Papers 051 - 066 Papers 051 - Technical Journal Technical 04

Technical Journal PAPERS 051 - 066 04 Printed on 9Lives Offset, part of the family Printed on 9Lives Offset, of ’born again’ papers, manufactured with National Association of Paper Certification (NAPM) using Merchants pulp, fibre 100% recycled approved the ultimate in offers 9lives Offset management. The environmental manufacturing mill and printer have also with the internationallybeen accredited ISO 14001. standard recognised © ATKINS Ltd except where stated otherwise. The Ltd except where © ATKINS logo, the “open A” device and the strapline ATKINS Ltd. trademarks of ATKINS “Plan Design Enable” are Welcome to the fourth edition of the Atkins Technical Journal. This edition of the Journal is dedicated to Cressida Spachis who sadly passed away earlier this year. Her paper, “Delivery of bus priority projects - a partnership approach”, was submitted to the Technical Journal to showcase the success of working in partnership on the Route 38 Project. Cressida was Project Manager for the Route 38 Corridor Management Pilot Study which has developed principles for intensified bus priority now being used across . The project received many accolades and Cressida was due to be part of the client/consultant team receiving the Improvements to Bus Services award at the London Transport Awards earlier this month. A further tribute to Cressida is made by Andy Southern, Managing Director for Transport Planning and Management, on page 69 immediately prior to the text of her paper. Once again there has been a great response to the call for papers across all businesses and it is particularly encouraging to have been able to publish papers from a wide range of staff grades covering Graduate Engineers to Technical Directors. The Journal continues to highlight the extensive range of technical disciplines Atkins can offer its clients and our Defence and Offshore Energy capabilities are prominent in this edition. With the recent appointment of three new Network Chairs we can look forward to even more effective showcasing of the diverse and complex problems that Atkins’ staff solve so well in their everyday activities. It is also particularly pleasing to see the excellent working relationships we have developed with our clients and partners being manifested in co-authored papers such as those written with the University of Bath, University of Southampton and English Heritage. This demonstrates that Atkins creates solutions that both we and our clients are proud of.

I hope you enjoy the selection of Technical papers included in this edition.

Chris Hendy Chair of H&T Technology Board

Highways & Transportation F O R E W O R D

5 Technical Journal 4 Papers 051 - 066

051 - Structures The launching of the River Esk Bridge 5 052 - Structures Recommendations for assessment Eurocodes for bridges 17 053 - Structures Limit equilibrium assessment of drystone retaining structures 29 054 - Intelligent Transport Solutions ITSO smart cards in Welsh public transport 39 055 - Intelligent Transport Solutions On balance - the see-saws of congestion charge business cases 45 056 - Intelligent Transport Solutions Improvements to ramp metering system in England: VISSIM modelling of improvements 51 057 - Highways ‘Scrambled’ pedestrian crossings at signal controlled junctions - A case study 57 058 - Highways Delivery of bus priority projects - A partnership approach 71 059 - Water & Environment Water quality modelling as a tool for assessing new water resource management options: The case of the River Stour, Kent 81 060 - Water & Environment High accuracy recording for heritage applications - Dover Castle graffiti 87 061 - Aerospace Conjugate heat transfer study of a spin pit rig: Application to the lifing of HP turbine disc firtrees 93 062 - Aerospace Non-deterministic thermo-fluid analysis of a compressor rotor-stator cavity 105 063 - Energy The optimum position for a tidal power barrage in the Severn estuary 115 064 - Energy Greater Plutonio project - Subsea flowline design and performance 123 065 - Defence Development of an armour mass estimation tool for land vehicles 147 CONTENT 066 - Defence Explosives safety - An oxymoron? 155 S The launching of River Esk Bridge 51

Rob Liddle Abstract Senior Group Engineer The River Esk Bridge carries the southbound carriageway of the M6 Highways & Transportation motorway over the River Esk estuary and forms part of the recently constructed M6 Carlisle to Guards Mill Motorway Extension which provides the missing link in the motorway network between England and Scotland. The bridge is a 180m long four span steel composite viaduct. Due to site constraints the most economical and practical method of construction was to launch the steelwork into position rather than to lift the steelwork in by crane. The bridge steelwork, weighing over 800 tonnes, was assembled on the northern approach to the bridge and launched southwards over the River Esk estuary This paper demonstrates that the design and construction of a launched bridge can be significantly more complicated than the design of a comparable bridge which is constructed using conventional methods, and that there are many more load cases and construction details to consider. This paper covers the approach for carrying out the analysis, design, detailing and construction of the structure and focuses specifically on the key issues relating to launching steelwork. It serves as a useful guide and quick point of reference for the design of launched steel bridges.

Introduction This paper covers the approach for carrying out the analysis, design and detailing of the River Esk Bridge on the M6 Carlisle to Guards Mill Motorway Extension. The Esk bridge is a four span 180m long steel composite viaduct over the River Esk and carries the southbound carriageway of the new section of M6 Motorway. The span arrangement is 31.4m/51.9m/51.9m/44.9m. The south abutment is fully integral, forming the point of fixity, and the superstructure is free to move longitudinally over Figure 1 - River Esk Launch the piers and north abutment. The bridge steelwork was assembled on the northern approach to the bridge and launched southwards over the River Esk estuary. The total weight of the steelwork was over 800 tonnes. STRUCTURE The reason for the launch was that access to the river for steelwork assembly and lifting was not possible due to the tidal nature of the river and the high risk of flooding. Also, lifting the beams in from the adjacent A74 ‘Metal Bridge’ was not possible due to traffic management issues. The A74 Metal Bridge was refurbished to carry the M6 northbound carriageway. Figure 2 - River Esk Launch S

5 51 The launching of River Esk Bridge

Analysis, design and buildability

Line beam analysis of launch phases Launch Phases The bridge steelwork was launched southwards in two phases: Phase 1 - The southern half of the steelwork was assembled on temporary land bases together with a launch nose and tail, and then launched southwards. The launch platform comprised the landbases, spaced at 28m/28m/28m, constructed in the partially completed approach embankment with the landbase levels set so that the steelwork could be set out and assembled to the Figure 3 - Views of the completed bridge adjacent to the existing A74 metal bridge vertical alignment required for the launch. The piers in the river were constructed in temporary cofferdams which were then converted into temporary islands to allow safe access and working at the piers without the risk of flooding. The cofferdams and islands were constructed higher than the maximum predicted flood level. Phase 2 - The tail was removed from the Phase 1 steelwork. The northern most steelwork spans were assembled and attached to the steelwork already launched. The tail was reattached to the rear of the Phase 2 steelwork. The whole bridge steelwork was then launched. On completion of the launch the temporary launch nose and tail steelwork were removed. The launch equipment comprised strand jacking tendons which were attached to the north abutment using Figure 4 - Views of the completed bridge adjacent to the existing A74 metal bridge a steel hauling frame which was embedded into the concrete stem of The M6 Carlisle to Guards Mill A temporary nose and tail assembly is the abutment to resist the hauling scheme provides the missing also required to be spliced to the main load. The tendons were attached link in the motorway network steelwork to accommodate the large to the launch tail via a large steel between England and Scotland. deflections that occur. Many bridge cross-beam which was attached to substructure and steelwork details The main difference between a static the end of all five beams to transfer were modified from a ‘standard bridge design and a launched bridge the hauling load into the beams. design is that each section of the design’ to accommodate the effects Choice of nose and tail length S steelwork elements is designed to of the bridge launch and use of work in different ways, depending on temporary launching equipment. The longer the nose, the less permanent steelwork would be

RE its position during the launch and also This paper focuses specifically on in the completed in-service condition. the launch-related aspects of the cantilevering at the leading span U The main girders and splices act in design and construction of the prior to ‘touch down’ at a support. T both hogging and sagging and the bridge, not on standard bridge A longer nose tends to reduce the leading front span behaves as a very design and construction topics. cantilever moment as the nose long cantilever during the launch. weight per unit length is normally UC lighter than the permanent works.

TR S 6 The launching of River Esk Bridge 51

Figure 5 - Cross section of the deck The use of a curved soffit is a different approach from that commonly used for post-tensioned incrementally launched concrete bridges in which the nosing usually has a hydraulic height adjustable tip to take up the deflection prior to touch down at a support. The tail needed to be profiled to take account of the back span ‘dropping’ off the land bases and bridge supports. Land bases are temporary supports in the steel launch platform area where the steel is assembled. A curved profile is required in the tail to gradually take up the cantilever deflection and to prevent the steelwork making a sudden Figure 6 - View of the launch showing signifi cant leading span cantilever drop as it moves off a support. defl ection and temporary cofferdam/island around an intermediate pier At each stage of the launch the The nose bottom flanges need The nose profile should be detailed ‘touch-down’ positions in the to be curved to compensate for to allow sufficient tolerance to ensure nose were calculated as they the cantilever deflection of the the nose travels past the support were required by the steelwork leading span and to prevent the bearings a sufficient amount before contractor – these being the location steelwork clashing with the piers and ‘touch-down’. 1m was specified and in the nose which would touch abutments as it passes over them. the vertical clearance at the tip was down first and also the angle of 165mm before ‘touch-down’ after the nose flange at touch down. considering deflections. The maximum cantilever deflection was 1112mm. STRUCTURE

Figure 7 - Phase 1 steelwork assembled ready for launch

Figure 8 - Phase 1 steelwork launched and phase 2 steelwork attached ready for launch S

7 51 The launching of River Esk Bridge

Figure 9 - Phase 2 launch complete

Launch bearing type The steelwork was launched on temporary launch bearings, set 351mm higher than the final position and then ‘jacked' down onto the permanent bearings. Launch bearings are usually a steel roller/skate or PTFE pads with lubricant applied to the underside of the bottom flange. Skates can sometimes damage the paint to the underside of the bottom flange and have a higher resistance/ coefficient of friction against the launch forces. PTFE bearings were chosen for the Esk launch. Deflection analysis during launch The software package STRAP was used to analyse the structure. Line beams were used to analyse the nose and tail deflections. At some stages during the launch uplift occurred at a support, indicated by a negative support reaction. The support was removed from the line beam model and the model solved again to obtain the correct results. It was important to accurately model the steelwork weight distribution, especially the plan bracing in the leading span, nose and tail, to obtain the correct results. Calculated deflections were Figure 10 - Hauling frame attached to north abutment required to enable accurate profiling on the nose and tail to accommodate the predicted deflections. Analysis of steelwork S Determination of launch phases The steelwork was analysed for the The most cost effective solution Due to the curve of the road cases where it was supported on RE was to launch from the north end alignment and compulsory purchase the land bases as well as on the

U only. On some very long bridges the order (CPO) boundary, an assembly permanent supports. The former is

T steelwork is launched from both ends area was prepared which was large especially important for modelling but this requires heavy plant, access, enough for only half the steelwork tail deflections and to check that assembly area and welfare facilities and therefore the steelwork had the steelwork does not bottom- UC in two locations rather than one. to be launched in two phases. out on the ground when leaving a Launching from one end only land base. To check for bottoming-

TR also minimises the number of out, the steelwork precamber

S temporary land bases required. also needed to be considered. 8 The launching of River Esk Bridge 51

Design details to consider Changes between hog and sag affecting effective length calculations At every location along the bridge each section of the steelwork acts in either sagging or hogging, depending on its position during the various stages of the launch as the steelwork is pushed over and between supports. Effective length calculations and section checks were required for every scenario as follows: • The number of bracings in a span could vary • The steelwork spans varying span lengths as it is moved over and between land bases, abutments and piers Figure 11 - View of leading span and nose during the launch • Sections changed from acting in hogging to sagging • Both the launch phase and in- service case were considered. For each steelwork section the most critical hog and sag case was chosen to calculate the beam effective length. Splices were also subject to both hog and sag moments and this was considered in the splice design. Intermediate bracings were treated as discrete torsional restraints to BS 5400 Part 31 clause 9.6.4.1.2 and BD13/062 designed to clause 9.12.2. Effects of differential precamber, setting out tolerances and fabrication tolerances At any point along the bridge the levels of the beam bottom flanges Figure 12 - Touch down at an intermediate pier position will vary between adjacent beams. The difference in level will be mainly due to differences in precamber but also a component from setting out and fabrication tolerances. This would not normally present a problem for a bridge constructed conventionally, however, for a launched bridge as the steelwork is pushed over the launch bearings at a support, the steelwork will be forced up or down STRUCTURE as the level of the bearings is fixed. Put another way, the difference in level between adjacent launch bearings at a support is fixed but the difference in level between adjacent flanges will vary throughout the span. This level difference or ‘lack of fit’ results in distortion of the bracing frames as the steelwork moves over the supports imparting large Figure 13 - View of tail, temporary land base, launch bearings and strandjack launch system forces into the bracing members. S

9 51 The launching of River Esk Bridge

If the deflections are too large to be accommodated, then beam lift-off at a support will occur causing additional large forces in the bracing members. Plane frame analysis models of the bracing were used to model these effects as summarised below: (i) Precamber values were determined from the main grillage of the completed structure (for all in-service dead and superimposed dead loads) (ii) The differences in precamber between beams across individual braced bays were calculated (iii) A plane frame model including all 4 bracing bays was set up to account for any transverse distribution effects (redistribution of vertical load) (iv) The calculated differential deflections were applied to the plane frame model to determine the forces induced in the bracing and also calculate the altered support reactions based on Figure 14 - Placing of jacks for removal of launch bearings prior to jack down transverse distribution effects (v) The system was rechecked for lift-off effects as follows: (a) The vertical force required to produce the prescribed deflection in the plane frame was calculated by removing the relevant support and applying a force to the model iteratively (b) If the force required to produce the deflection was larger than the actual support reaction that occurred then it was assumed that lift-off would occur (c) The support reaction at the lifted-off beam was transferred to the two adjacent beams. The additional dead weight in Figure 15 - Nose approaching launch bearings at a pier prior to touch-down the bracing due to the lifted- off beam was taken as being (e) The resulting forces from The total fabrication and setting equal to the support reaction the effects of differential out tolerance alone was +/- divided by the number of beam levels were added to 10mm without taking account S bracing bays in half the span other coexistent loads effects of differential precamber. It is either side of the support. for design of the bracing important to allow for the possibility

RE This is because for a lifted-off members and connections. of beam uplift since applying the beam the weight of the beam full deflections to the plane frames U Of course, a more sophisticated and bracings is transferred models can generate massive T space frame model of the whole to the two adjacent beams steelwork could have been used to forces in the bracing members. via several bracing bays simplify the number of operations, Initially the steelwork contractor

UC (d) The bracing plane frame but it was decided at the beginning suggested we provide both a top was resolved by removing of the scheme to only use line and bottom ‘push-pull’ member for

TR the support where the beam models of the girders to each bracing set between beams beam had lifted off

S model the launching effects. 2 and 3 as shown in Figure 16. 10 The launching of River Esk Bridge 51

TOP TIE MEMBER ORIGINALLY BOTTOM TIE LIFTED TO CLEAR PROPOSED GUIDES IN BAY 2-3

BEAM 1 BEAM 2 BEAM 3 BEAM 4 BEAM 5

PUSH – PULL BRACING

LATERAL GUIDES AT SUPPORTS TO PREVENT LATERAL MOVEMENT Figure 16 - Arrangement of bracing

The contractor had concerns that The bracing would be difficult to The vertical support reactions during the bridge was not adequately remove, being over the river, and the launch comprised the steel self braced between the group of with the EMJ formwork already weight, formwork on all but the three beams and the pair and that in place. Therefore, from a health leading span and a proportion of the two sections could deviate and safety and CDM perspective it the deck reinforcement and so the apart. Launching 5 beams is not was decided to leave the bracing reactions were a lot less than the a usual scenario as beams would in place. The bracing was also in-service support reactions. However, normally be launched in pairs. checked for the in-service case. the web thicknesses required were They proposed continuous bracing Web patch load checks on slightly thicker than those required at all bracing locations should be unstiffened sections of web for the in-service condition to accommodate the unstiffened used. This made the whole bracing In the in-service condition the bearing support scenarios. BS5400 system effectively fully rigid and steelwork at supports is stiffened Part 31 clause 9.9.6 and Annex A, so prescribing large deflections with web bearing stiffeners to allow ‘web patch loading’, were used to generated large forces resulting in the web to carry the vertical support perform this check. It is noted that the need for excessively large bracing reactions. However, for most of some further refinement of the members and bolted connections. the time during the launch, the design would have been possible By removing the top member in position of web bearing stiffeners using the more recently developed bay 2-3 the bracing system became or intermediate transverse web patch loading rules in EN 1993-1-53. more flexible as the group of 2 stiffeners would not coincide beams (beams 1 and 2) could rotate with the position of the supports. Splices designed to or flex relative to the group of Therefore the unstiffened parts of resist hauling load three beams (beams 3, 4 and 5). the web were checked for carrying Splices were checked for carrying the the vertical support reactions. horizontal hauling load together with in span bending and shear effects. Plan bracing requirements During launching the leading span acts as a long cantilever prior to touch-down on a support. In BS5400 Part 3 the determination of the effectiveness of lateral restraints is based on how well the compression flange of a beam is restrained against lateral and rotational movement. STRUCTURE Without plan bracing, for a maximum cantilever of 51m, the tip of the steelwork would be unrestrained and would sway significantly and so the effective length would be too large to enable a realistic plate design. Triangulated plan bracing was provided throughout the leading span to prevent this sway and fully restrain the steelwork.

Figure 17 - View showing launch track marks on bottom fl anges and plan bracing left in place S

11 51 The launching of River Esk Bridge

The analogy being a car being pushed around a corner without turning the wheel i.e. overcoming full friction of the tyres on the ground. This is a gross overestimate as this ‘steering’ effect does not occur. What actually happens is that the curved steelwork ‘drives’ itself around the curve naturally when pushed. The hauling load must overcome friction to move the steelwork. The largest horizontal force that can be imposed on the supports longitudinally in the direction of movement is the vertical reaction at that support multiplied by the coefficient of friction of the bearings. The horizontal load imposed on the lateral guides, taking into account the Figure 18 - Idealisation of symmetrical constant radius curve launch driving action, can be derived from considerations of force and moment equilibrium. A simple grillage or line The plan bracing provided effective The bracing, which would eventually beam model of the deck in plan can discrete lateral restraints where form the support bracing at the be used to consider equilibrium of the the effective length was limited to piers and abutments on completion system to determine the hauling load the distance between intermediate of the launch, was also checked and lateral guide forces. The bearing bracings and enabled an efficient as intermediate restraints for the friction at each support is applied use of steel plate. The plan bracing launch phase when positioned as a load parallel to the direction was provided between at least two away from a support. of sliding. The hauling load can be beams near the compression flange. Launch load effects due to plan found by putting a pin support at Plan bracing was also required curvature of the superstructure the hauling end of the deck, acting between all 5 beams in the end bay At first glance it might appear that parallel to the deck. The resulting of the leading end of the permanent the lateral launch forces on the lateral guide forces are determined steelwork i.e. just behind the nose PTFE lateral guide bearings would as the horizontal supported reactions splice. This plan bracing was required be equal to the force required to acting perpendicular to the deck. The for erection purposes and during push the steelwork sideways over assumption is made that the bearing launching to prevent adjacent beams the support bearings i.e. the total friction on the guides is negligible moving longitudinally relative to vertical load at the support multiplied and can be ignored. This would seem each other, known as ‘leading’. by the coefficient of friction. reasonable since the guide reactions are of relatively small magnitude compared to the vertical reactions. The hauling load and lateral guide forces need to be calculated for each key stage during the launch. This is because the geometry, spans and sum of support reactions, and hence the hauling load, will vary as the launch progresses. For a simple symmetrical structure on a constant radius horizontal curve, the lateral guide loads can be easily

S determined by hand calculation using simple rules of equilibrium, as shown in Figure 18. The simple example RE assumes the steelwork is pushed U and pulled at both ends whereas the

T more usual case, as with the River Esk, is to push the steelwork from one end only. Equal supports reactions UC are also assumed for simplicity.

TR Figure 19 - Lateral PTFE guide bearing anchored to pier and

S abutment bearing shelves with pig tail anchors 12 The launching of River Esk Bridge 51

Taking all moments about Launch load effects on When three or more beams are point A summed to zero: abutments and piers connected together, either by cross bracing or top and bottom ‘push-pull’ F = 3P The substructures were checked for the load effects from the bracing, then load sharing can occur. Summing all loads in plan launch hauling load, lateral guide This is termed ‘participating bracing', along the y axis to zero: forces and friction effects of the where, as the deck deflects under steelwork while being launched. load, some vertical load is transferred The steelwork was also being through ‘transverse distribution'. launched uphill by a gradient 0.3% The participating bracing was The above assumptions are correct and this gradient was considered. modelled to determine the additional for a deck on a constant radius The lateral and longitudinal forces load effects in the bracing members. horizontal curve. For the case where on the piers and abutments due to Lubricant coefficient the deck is on a varying radius moving the steelwork were considered of friction values horizontal curve, or a combination in the design of the abutments and A lubricant was brush applied to of constant curves and transition piers. The coefficient of friction the bottom flanges of the beams curves, then additional lateral guide in the longitudinal and transverse during the launch to minimise forces need to be considered. directions was taken as 11% (0.11) friction. The lubricant was actually These additional forces result from and 5% (0.05) respectively. a soap normally used for removing changing the plan curvature of the The north abutment was designed grease and oil, namely ‘Tufenega deck from the initial constructed to resist the full hauling load and Green Gel’ (like Swarfega). shape that is forced at any instant by the hauling frame was anchored During the approvals process of the guides. It is worth noting that if into the north abutment stem. these forces were too large then the the AIP the Highways Agency (HA) The hauling load effectively pulled SSR requested a justification for the assumption of ignoring friction on the abutment into the fill and so the guides would not be correct. assumed coefficient of friction values. passive resistance could be utilised. As mentioned in the AIP, the main The additional lateral guide forces Participating bracing girders and substructure were checked can be determined by modelling the for additional temporary loads due to effects of the forced deflection of The optimum number of main beams was investigated at the preliminary launching of the main girders, these the deck in a simple grillage model loads are summarised as follows: or by using simple beam formulae. design and value engineering stages of the design. Cost comparisons (a) Wind loading to BD37/01 An impact factor of 1.5 was were made between 4, 5 or 6 beams. (b) Loading due to differential applied to the calculated lateral With an even number of beams guide forces and these forces were vertical pre-camber as the bracing is usually arranged in specified in BS5400 considered in the bracing design. discrete unconnected pairs i.e. for 6 (c) Lateral sliding friction loads beams - three braced pairs. However, exerted by temporary for an odd number of beams the launching guides due to the bracing is arranged so that three curved horizontal alignment of beams are connected together. For the deck, assuming a friction the five beam case on the Esk there is coefficient of 0.05 based on a braced pair and a group of three. data used for similar past schemes as detailed below (d) Longitudinal sliding friction loads, assuming a friction coefficient of 0.11 based on data used for similar past schemes as detailed below (e) Longitudinal launching forces applied on the tail end of the steel deck. STRUCTURE The same launch method had been completed successfully by the steelwork contractor shortly before the launch of the River Esk on the new Swale Crossing forming part of the 30 year DBFO contract for the A249 Iwade Bypass between the Isle of Sheppey and mainland Kent. To ensure that the structure would not Figure 20 - Strand-jack launching system with stiff cross beam transferring be subjected to excessive loads during loads to tail beam ends - monitoring station under cover in background

launching, the steelwork contractor S had carried out extensive in-house 13 51 The launching of River Esk Bridge

Furthermore, to provide an additional margin of safety a higher load factor than that specified in BD37/01 was Tapered bearing plate conservatively adopted to check the Plate permanent works during launching to girder allow for dynamic and impact effects. A load factor of 1.5 was used for both dead load and superimposed dead loads, instead of 1.05 and 1.2, together with the friction coefficients stated above. Initial structural checks Adaptor plate Bearing 30mm thick showed that the load effects on the bolted to welded to substructure due to the in-service adaptor plate base plate condition would be far in excess of those generated during launching, Cast in base plate with resulting in a very high safety factor shear during launching ranging from 4.3 for overturning up to 29 for shear; the main girders safety factor was 1.75. Figure 21 - Adaptor plate arrangement The friction coefficients to be adopted testing in order to determine the most If unexpectedly high loads were during the launch were deemed safe suitable friction coefficients specifically indicated then the launch was to be to use and accepted by the HA SSR. used for the scheme. The friction halted and the cause investigated The main girders were checked coefficients stated above were based and resolved to continue the in biaxial bending for the effects on data taken from continuously operation. The monitoring system of major axis bending due monitoring the load effects on the provided a permanent digital to the launch and minor axis structure during the Swale Crossing record of the hauling loads. bending due to wind loads. launch. The same monitoring Prisms were attached to each system was adopted for the River pier, at the top, middle and Geometry arrangements, Esk scheme described as follows: bottom levels in order to check details and buildability Strand jack loads were monitored for displacement and rotation of The steelwork contractor would during the launch operation - the substructure. Measurements have preferred the bearing taper calibrated pressure transmitters were taken continuously during plates to be ‘through bolted’ to the were installed in the hydraulic lines the launching operation by bottom flange to avoid having to site to the strand jacks and connected surveyors using EDM equipment. weld and paint up over the piers. to a laptop computer to monitor However, due to space limitations strand jack launch loads. This for bolt locations the taper plates provided a digital and graphical were welded on site post launch. display of the load being applied Obviously the taper plates could by the two jacks every five seconds only be added once the steelwork that were continuously monitored was launched otherwise they would by an engineer who was in radio snag with the launch bearings. contact with the Launch Controller. A steel adaptor plate was used to attach both the temporary launch bearings and the permanent bearings. At each bearing location a base plate with shear studs was cast into the bearing plinth. An adaptor plate was welded to the base plate. The temporary launch bearings were bolted to the adaptor plate and S then removed after the launch. The permanent bearings were then

RE bolted to the same adaptor plate prior to the steelwork being lowered U down on jacks from the elevated T launch position onto the permanent bearings. This was a practical detail

UC to allow the use of two types of bearing in the same location. All site welding was painted to full TR Highways Agency specification.

S Figure 22 - Split bottom fl ange splice plates 14 The launching of River Esk Bridge 51

The temporary launch bearings This also applied to the launch nose At the piers sufficient room was comprised two 300 long x 200 wide and tail. In the final design of the available to jack down the steelwork PTFE bearings in a line (600 long). permanent steelwork all bottom from under the bottom flanges of The whole rocker bearing assemblies flanges were 900mm wide and the main beams as the piers are were typically 780mm x 370mm. all top flanges 600mm wide. The 1500mm wide. However, at the During the launch at least two flange sizes varied in the nose and north and south abutment the points of lateral restraint were tail for economy except in bay 2-3 bearing shelves were much narrower required at all times to prevent the where the same outstand as for and so no room was available to steelwork shifting out of line. the main steelwork was required. place the jacks under the main flanges. At the south abutment no During launching the steelwork was Web to flange welds were prepared in permanent bearing plinths were 351mm higher than its final in-service a T & I machine and were not ‘fitted’. to be provided and so jacking was position. At each support a pair of Therefore for web patch load checks made more difficult. To solve the lateral guides was provided between from the support reactions the weld problem, the bracing adopted at beams 2 and 3. The bracing in this was checked for resisting the direct these locations was mid–height bay was lifted higher than in adjacent patch load since the the patch load horizontal I sections, with stiffened bays in order to clear the guides. A would not be transferred from the jacking points provided either side of distance of 350mm from underside flange to the web in direct bearing each main beam to enable jacking. of bracing to underside of bottom but via the web to flange welds. flange was required to clear the A safe maximum launch wind speed guides. Bay 2-3, being a push-pull was specified on the drawings. Conclusions bracing bay, was easier to modify The wind speed corresponded to This paper demonstrates that than an adjacent cross-bracing bay. a 10 year return period with a 3 the design and construction of a Once the launch was complete the second gust speed of 18m/s. launched bridge can be significantly structural steelwork was jacked The permanent formwork and a more complicated than the design down into its final position. Initially proportion of deck reinforcement was of a comparable bridge which is the preferred sequence was for the added to the steelwork prior to the constructed using conventional steelwork to be jacked down fully, one launch on all but the leading span methods. There are many more support at a time. However, deflection to minimise construction operations load cases, load effects and details calculations were carried out to working at height and also mitigate to consider. However, in situations determine what bending moments access problems. A cost comparison where access to the bridge site or would be induced in the steelwork was made between EMJ and Omnia traffic management issues make by this imposed deflection (351mm). formwork. EMJ was chosen as it conventional bridge construction It was found that lift-off from the is half the weight of Omnia. difficult, bridge launching can offer an economical solution. This paper jacks would occur at some support At the integral south abutment the summarises some of the key issues locations before reaching the required wall reinforcement was required to relating to launching steelwork. level. The jacking sequence was run vertically up into the diaphragm revised to allow only jacking in stages and lap horizontally into the deck The first phase of the launch was at some supports to prevent lift-off. slab. In normal construction starter successfully completed on 16 May Bottom flange splice cover plates bars are usually left protruding where 2007 and the second phase on 11 comprised two separate plates the diaphragm is to be cast both June 2007. The scheme was opened 350mm apart rather than a single in the front and back face of the to traffic in December 2008 plate. This was to ensure the splice abutment stem/diaphragm. Because plates ran either side of the launch of the launch operation these bars Acknowledgements bearings. Bottom flange butt welds would clash with the steelwork and The author would like to acknowledge were ground flush where there nosing. A 300 wide x 1800 deep Fairfield Mabey Ltd, the Steelwork was to be contact with the launch box-out was provided on the back Contractor, for sharing their bearings and the guides in order to face so that the kicker bars could knowledge and expertise on launching prevent snagging and unnecessary be left protruding below the level steelwork. The author would also like high hauling loads. This was the of the steelwork. For the front face to acknowledge the Main Contractor central section (300mm wide) bars couplers were used instead of - Carillion, the Lead Consultant - of the underside on all girders, starter bars. The reduced thickness STRUCTURE Capita and also the Sub-Consultants and the edges of the two inner of abutment stem was checked (Bridge Designers) Grontmij. girders, which were in contact with for the temporary load cases. the lateral guides. All other butt At the north abutment the welds were to be dressed only. inspection gallery starter bars were Bottom flanges needed to have required to lap into the back of the same outstand dimension the gallery wall. These bars were in bay 2-3 so that the lateral kept as low as possible by slightly guides would always be in lowering the gallery floor. contact with the flange edge. S

15 51 The launching of River Esk Bridge

References 1. BS5400 : Part 3 (2000): Design of steel bridges. British Standards Institution, London. 2. BD13/06 (2006). Design of steel bridges. Use of 5400-3:2000. Highways Agency, UK 3. BS EN 1993-1-5 (2006) Design of Steel Structures. Part 1.5: Plated structural elements. British Standards Institution, London. S RE U T UC TR S 16 Recommendations for assessment 52 Eurocodes for bridges

Chris R Hendy Abstract Head of Bridge Design The new Structural Eurocodes offer increased economy in design over and Technology most existing codes of practice. This is achieved through codification of Highways & Transportation more advanced calculation methods, such as non-linear analysis, and also through codified rules which give increased economy in many situations compared to the old British Standard rules because they are based on recent testing and numerous non-linear finite element parametric studies. This presents great opportunities for improving sustainability in design. However, the assessment of existing structures is specifically outside the scope of the Eurocodes, so many of the benefits these new codes bring cannot be directly applied when reviewing old structures. Increased assessments of strength would Jessica Sandberg be highly beneficial in demonstrating that old structures with increased loading Group Engineer or small amounts of deterioration were still adequate to continue in operation Highways & Transportation without costly refurbishment, modification or reconstruction. The Eurocodes lend themselves to adaptation for assessment of structures because the rules are generally based around the application of first principles; this gives greater scope for determining the true ultimate strength of a structure and for the incorporation of structure specific information on loads and material properties in a rigorous manner. The assessment criteria can also be differentiated to take account of differences in consequences of failure of a bridge. This paper investigates the areas where the steel and concrete Eurocodes give increased resistances compared to existing codes and makes Navil K Shetty recommendations for sections which could be directly applied to the Director assessment of existing structures. Areas where the Eurocodes cannot be Rail applied directly to existing structures are also identified together with the reasons why, such as reliance on modern material and execution specifications. Recommendations are then made for how the Eurocode design rules could be modified for assessment situations, including the use of measured strengths and imperfections in calculation. Actions and the format for combining actions, are also investigated and recommendations made for modifications to these aspects for use in assessing bridges. Finally, an overall assessment of the scope of work required to produce an assessment suite of Eurocodes is made.

Introduction The new Structural Eurocodes offer assessments of strength would be in situ tests on a structure can be increased economy in design over highly beneficial in justifying that old rigorously incorporated into the limit most existing codes of practice. This structures, with increased loading or state verification method given in the is achieved through codification of small amounts of deterioration, were Eurocodes. In addition, assessment more advanced calculation methods, still adequate to continue in operation criteria can be differentiated by such as non-linear analysis, and also without costly refurbishment, taking account of the differences in through codified rules which give modification or reconstruction. consequences of failure of bridges. increased economy in many situations The Eurocodes lend themselves The approaches to be used for compared to the old British Standard to adaptation for assessment of developing the relevant rules for rules because they are based on much structures because the rules are assessment are discussed in this paper. more recent testing and non-linear generally based on the application of The main area of difficulty in applying finite element parametric studies. first principles; this gives greater scope the Eurocodes to the check of existing This presents great opportunities for for determining the true ultimate structures is that the assumptions improving sustainability in design. strength of a structure than do more of particular material properties However, the assessment of existing prescriptive or empirical rules. and workmanship inherent in the structures is specifically outside the Traffic data (collected using weigh-in- codes may not be borne out by the scope of the Eurocodes, so many of motion systems for long span bridges actual construction. Many of the the benefits these new codes bring or from known traffic counts for short enhanced resistances in Eurocode 3 cannot be directly applied when span bridges), measured dimensions for example stem from the ductility reviewing old structures. Increased and material properties derived from and strain hardening characteristics 17 52 Recommendations for assessment Eurocodes for bridges

HA standards are applied to them because resistances to Eurocodes tend to be higher on average. The avoidance of this problem leads to a potential third. Inevitably, engineers will seek to use large parts of the Eurocodes in their assessment as a departure from assessment standards, thus creating a pseudo assessment version of the Eurocodes, but without the control, expert input, consistency and consensus that would be present if an appropriate committee had assembled the assessment code. The remainder of this paper investigates areas where the steel and concrete Eurocodes give increased resistances compared to existing codes and makes recommendations for sections which could be directly applied to the assessment of existing Figure 1 - BSALL is used for the operation of the Forth Road Bridge structures and the modifications required to the sections which could not be directly applied. Recommendations are also made for the modifications needed to action calculation and the combinations of actions when assessing bridges. Key features of an assessment version of a Eurocode The process of assessment is of crucial importance for maintaining bridges in a safe and serviceable condition. The objective of assessment is to evaluate the safety of an Figure 2 - Imperfections in longitudinal stiffeners following an impact existing bridge quickly and with of modern steels that are produced characteristics. An additional factor a minimum of effort. However, to the specification EN 100251. is that the Eurocodes do not directly the assessment rules and criteria Older steels, or indeed wrought and deal with reduction in strength due to need to be established rigorously cast irons, may not comply with the deterioration arising from unexpected and judiciously. If assessments are requirements of this specification. behaviour or lack of maintenance. unduly conservative, structures will be unnecessarily strengthened, Similarly, all steelwork designed to In the absence of assessment versions or needless load restrictions will Eurocode 3 should be fabricated of the Eurocodes, a number of be imposed. Conversely, if the and erected in accordance with problems and inconveniences will rules are too lax, the safety of EN 1090-22. This document sets arise for UK engineers. First, there is bridges could be compromised. out, among many other things, the the obvious problem that engineers tolerances for steelwork such as will need to remain conversant with Eurocodes employ characteristic bow imperfections in columns. The two design standards: the Eurocodes actions and material properties

S strut curves in Eurocode 3 assume and the old Highways Agency (HA) combined with load and material that these bow imperfections will assessment standards which were partial factors to ensure an not be exceeded but this may not heavily based on BS 5400. Whilst appropriate level of safety and RE be the case for old bridges which this is not unmanageable, it is reliability. These factors guard U were not constructed to the same unfortunate because a prime driver against extreme variations in design

T tolerances; it is necessary to use for the Eurocodes was to avoid the parameters (e.g. material properties the actual tolerances in calculation. need to use more than one code of and applied loads) which could occur The Eurocodes may not even cover practice. Second, there is the more during service. In order to ensure UC design with certain materials. Plain significant problem that structures that the design rules are simple for round bars, for example, are not designed to the Eurocodes will mostly routine use, the values of the partial

TR covered by Eurocode 2 because of fail their assessment when the old factors have been chosen such

S their poor and unpredictable bond that they cater for a wide range of 18 Recommendations for assessment 52 Eurocodes for bridges

Figure 3 - Physical test and non-linear model of a plate girder structure/component types and failure with the design code The work needed to develop modes. It follows, therefore, that (i) Outmoded forms of construction assessment provisions for the above the rules tend to be conservative for not covered by the design code are discussed in this section. The the majority of bridges and the level remainder of the paper discusses the (j) Deterioration of structural of conservatism varies considerably evaluation of the ultimate strength members through condition from structure to structure. of bridges based on modifications factors or modification to Eurocodes 2, 3 and 4. The objective of the assessment code to material factors. would be to produce a more realistic A feature of previous UK assessment assessment of a structure than is Bridge specific assessment loading codes has been the introduction of possible at the design stage. This in modified, and usually significantly The procedures for the derivation of part is achieved through using current more complicated, calculation rules Bridge Specific Assessment Live Load state of the art analysis techniques whereby some of the simplifying (BSALL) for long span bridges with and design rules based on physical and conservative assumptions in spans greater than 50m are already testing where appropriate, and in the equivalent design formulae are well established in BD 503, and can part by taking advantage of the removed. The need to do this in be readily extended to Eurocode information available to an assessing a Eurocode assessment standard traffic loading. BSALLs are often vital engineer in respect of the loading, would often be negated because in justifying continued use of bridges material strength, geometrical the rules are usually already more without the need for strengthening properties and imperfections which general and computer modelling to meet full design standard loading necessarily have to be conservatively can often be used directly to see Figure 1. The derivation of BSALL estimated at the design stage. The obtain the required results. involves collecting traffic data at assessment code must therefore the bridge site through weigh-in- enable account to be taken of: Overview of motion systems and using this data to (a) Traffic loading appropriate to simulate extreme load distributions. vehicle flows over the bridge modifications to The live load partial factors given in (b) Bridge specific partial account for bridge the UK National Annex to Eurocode factors accounting for BS EN1990: Annex A2 can be used specific loading and unchanged with the BSALL. consequences of failure resistance information (c) Measured imperfections and For bridges of spans less than 50m, live load reduction factors to the load sizes different from those General assumed in the code models LM1 and LM2 given in the In assessing an existing bridge, UK National Annex to BS EN 1991- (d) Measured material strengths 24 can be derived for assessment

account may be taken of a number STRUCTURE (e) Construction not fabricated of factors that are specific to a bridge purposes in a similar manner to those or erected in accordance and the assessment criteria modified given in BD 215 depending on traffic with the relevant current to reflect these, for example: density and road surface roughness execution specifications for different levels of assessment • Bridge specific loading (f) Material not complying with the loading. The surface roughness factors relevant current product standards • Bridge specific resistance in BD21 may be readily applicable properties (material but the live load reduction factors (g) General configurations and shape and geometrical) will require some calibration. limitations not complying with the limits in the design code • Consequences of failure. (h) Minimum stiffnesses and

strengths not complying S

19 52 Recommendations for assessment Eurocodes for bridges

‘medium consequence’ structures (Consequence Class 2). Designing for a higher Consequence Class may be considered for bridges with span greater than 200m. However, small (i) Accumulated stresses determined (ii) Accumulated stresses used to culverts or bridges with span less on gross cross section determine an effective steel than 10m or medium span bridges cross section for re-calculation on a minor road with light traffic may of accumulated stresses be regarded as ‘low consequence’ Figure 4 - Illustrative procedure for determining effective cross section in Class 4 composite beams structures (Consequence Class The load models LM3 (abnormal Section D7.2 of BS EN 19906. Further 1). The annual target reliability (as loads) and LM4 (crowd loading) details of this methodology are given defined in BS EN 1990 Annex B) for given in the UK National Annex in reference 7. The calculation of a low consequence structure can to BS EN 1991-2 can be used for characteristic strength values takes be taken as 4.2 as opposed to 4.7 assessment of existing bridges without into account the number of test used for a medium consequence further modification, although the results, variability in test results and structure. As a result the partial derivation of bridge specific crowd prior knowledge about the statistical factors on live load can be reduced loading models is also possible. distribution of material properties. The by a factor of 0.9 for Consequence Where the structural dimensions and in situ characteristic strength values Class 1 as given in Table B3 of BS EN the thickness of road surfacing are can then be used together with the 1990, Annex B. In assessment, such measured for an existing bridge, the material partial factors given in the a reassessment of the consequences measured dimensions, together with UK National Annexes of material of failure may be appropriate. measured values of material density, Eurocodes EN 1992, EN 1993 and can be used to calculate more realistic EN 1994 for bridges. Where the Reliability based assessment values of self weight and surfacing ultimate resistance of a component (Level 5 Methods) or product is established through load for the bridge. For surfacing, The Eurocodes allow the direct testing, the design resistance can this will allow the tolerances on use of probability based structural be calculated using the procedures range of surfacing depth given in the reliability analysis methods for the given in Annex D of BS EN 1990. Eurocodes to be reduced, provided design of structures. It is important controls are in place to limit changes In some cases, construction tolerances that their use is permitted for the to surfacing depths in the future. (imperfections) or impact distortions assessment of existing bridges as Depending on the level of variability may also affect the structure’s they provide significant benefits in measured values and the controls strength and will therefore need for assessment. These methods are applied, the partial factors for self to be considered explicitly if they referred to as Level 5 methods in the weight, super-imposed dead load are greater than those assumed in UK8. In addition to accounting for and surfacing load given in the the Eurocode design rules Figure bridge specific loading, resistance UK National Annex to BS EN 1990 2 shows an extreme case of post- parameters and consequences Annex A2 could be reduced further. impact deformation. Greater of failure, the reliability based discussion on this is included below. assessment methods enable account Bridge specific resistance to be taken of a bridge’s previous parameters Consequences of failure exposure to known abnormally Where material properties, such as Consequence of failure reflects both heavy loads (e.g. Special Order yield strength of steel or cylinder the economic impact of the structural loads in excess of 180 tonne gross strength of concrete, have been failure (in terms of rebuild costs and weight), warning of failure related obtained through intrusive in situ disruption costs) and also the potential to structural form and material and testing on an existing bridge, the for loss of life. The UK National Annex quality and frequency of inspections. in situ characteristic strength can to BS EN 1990 Annex A2 recommends It is, however, important that be calculated using the Bayesian that all bridges up to 200m span the reliability-based assessment updating methodology given in should normally be regarded as procedures are fully standardised, including the probability distributions for the load and resistance random

S variables and the target values of reliability index. The necessary background research in this regard RE has already been undertaken during

U the calibration of load partial

T factors and load combination factors for use in the UK National Annex to BS EN 1990, Annex A2. UC TR Figure 5 - Typical effective cross section for design to EN 1993-1-5 S 20 Recommendations for assessment 52 Eurocodes for bridges

Overview of the economic benefits of Eurocodes 3 and 4 and the modifications required to produce a steel assessment code The following discussions are based around Eurocodes EN 1993-29 and EN 1994-210 and the other Eurocodes they cross reference and identify, in outline, the modifications required to produce an assessment standard. The various sections of the codes are reviewed one by one. General The scope of EN 1993-2 in its section 1 would need fairly extensive modification because of the number Figure 6 - Typical interaction diagram for shear and moment to EN 1993-1-5 of cross references to modern product and execution standards, but this This will require either modifications Structural analysis would mostly involve deletions. to the strength equations and Much of what is written on structural The terms, definitions and symbols procedures or the use of non-linear analysis would require very little defined in the section however analysis using known material amendment as the principles given would mostly be retained. Guidance properties for the structure. Other are to model the structure in a would have to be provided also on requirements for minimum toughness way that reflects its behaviour; inspection for assessment and the to guard against brittle fracture this applies equally to design and information on material properties, would need less modification, but the assessment. Elastic analysis itself, with imperfections and condition that Charpy energies for use in calculation subsequent member cross-section should be obtained and recorded. This would again need to be determined and codified buckling checks, needs could largely be taken from existing via the means above. The sections little adaptation except that facility documentation such as BD 5611. on brittle fracture referenced in would be needed to enable the EN 1993-1-1012 permit the use Basis of design buckling curves to be modified to of fracture mechanics methods allow use of measured imperfections. The material on the principles of to check adequacy, so actually limit state design and design assisted lend themselves to assessment. by testing would require very little The corresponding section in an amendment for use in assessment; assessment code could usefully give indeed design assisted by testing key material characteristics such would be a useful assessment tool. as yield strength, ultimate tensile Materials strength, ultimate strain, toughness The section on materials would and modulus of elasticity for historic require significant modification. steel grades. This could be readily This section of the Eurocode provided by extracting the required sets requirements for minimum properties from old superseded steel ductility and strength. The relevant specifications. An assessment National parameters would need to come from Annex would be a good place for as-built records and specifications this information to be recorded. or from intrusive investigation Durability and testing of the structure. STRUCTURE Most of the design requirements Non-compliance with the prescribed involving durability would simply be limits on these parameters, however, deleted in an assessment version would mean that some of the of the standard as inspections remaining Eurocode resistance would determine to what extent rules could not be used so the the design had succeeded in that implications of non-compliance regard. Deterioration would need on the remaining design rules to be accounted for via a condition would need to be addressed. factor in the resistance equations in the same way as is currently required

in UK assessment standard BD 21. S

21 52 Recommendations for assessment Eurocodes for bridges

Non-linear analysis, particularly of This aims to fix material factors stresses against yield when calculated slender structures, will often give the such that a base level of reliability is on the resulting reduced cross-section. most accurate and realistic picture obtained when using the assessment The procedure for composite beams is of true structural strength and is rules, based on the true resistances to first calculate accumulated stresses thus a very effective assessment tool. obtained in tests. The rules in EN 1993 on the gross cross section, following The model in Figure 3, for example, utilise relatively few values of γM for the construction sequence, and then predicted the physical test failure load simplicity. Background papers 13 are, to determine the effective areas of to within 4%. Non-linear analysis however, available in a number of the compression elements based on methods are, however, influenced cases to support the use of reduced this stress distribution. Finally, the by imperfections (residual stresses values where adequate reliability accumulated stresses are recalculated and tolerances on straightness and would still be obtained and advantage using the reduced effective steel cross flatness) and those given in Eurocode of this could be taken in assessment. section at all stages of construction. 3 may not be appropriate for an The Eurocode cross section resistances This is illustrated in Figure 4. existing structure. Requirements make some allowance for strain Class 4 beams with longitudinal for non-linear analysis would hardening in a number of areas stiffeners are treated in the same therefore require either magnitudes (for example in determining the way as beams without longitudinal of equivalent imperfections to be reduction of strength due to holes stiffeners in EN 1993, unlike in BD 56 provided for old structures or for and in the shear-moment interaction where a completely different approach imperfections to be measured in curves) so it would be necessary to to calculation was employed. In BD the existing structure and residual set out what to do if the ductility 56, individual panels and stiffeners stresses determined by calculation limits were not met. The simplest are checked for buckling once stresses based on, for example, weld details. situation in this case would be to were determined in them, generally Often, doing the latter will in any revert back to the more conservative using gross cross sections other case be beneficial compared with provisions of BD 56, adapted to than for flange plates where some the alternative of using the design make them fit the general Eurocode allowance is made for loss of strength code values. The rules would need to terminology in terms of notation. due to local plate buckling. There provide guidance on the application of A similar exercise has already been is therefore limited load shedding imperfection patterns to the structure. done in the drafting of reference between components and a single Guidance would also be needed 14 for salvaging non-contradictory overstressed component can govern on the material properties to use in material from BS 5400 Part 315 to the design of the whole cross section. analysis where the material behaviour be used in design with Eurocode 3. In EN 1993-1-5, effective widths does not conform to the assumptions Bending cross section resistance are again used to allow for buckling and limitations in the design code, of web and flange elements, as for For Class 1 to 3 cross sections, the but non-linear analysis is, in principle, unstiffened Class 4 cross sections, but calculation procedure and results for the best form of analysis provided the same approach is also used for bending resistance are very similar that the material characteristics are stiffeners see Figure 5. This effectively to that for previous practice. One known and can be modelled. allows load shedding between all small area of economy for Class 3 Plastic analysis would also be the various elements such that their cross section design is that where the more readily justifiable for use combined strengths are optimally bending resistance is based on first in assessment since in-service used. This represents a significant yield at an extreme fibre, EN 1993 performance can be visually change from previous UK practice defines an extreme fibre as the mid- assessed. The Eurocodes set out the and can give rise to a significantly thickness of the flange, rather than requirements for plastic analysis so increased prediction of load carrying the outer surface as was previous UK only a relaxation in the conditions resistance. These rules could be used practice. This gives a small increase in of use is likely to be needed in the with very little modification other economy over BD 56, particularly for assessment version, provided that the than to allow adjustment to the shallow beams with thick flanges. For requirements for ductility are met. reduction factors for plate buckling steel-concrete composite design, EN to allow for measured imperfections. Ultimate limit states 1994 employs a similar rectangular stress block for concrete for plastic Shear buckling resistance and Material factors and cross design as is used in BD 6116, web transverse stiffeners section resistances although the resisting compressive The rules for shear buckling in EN

S An important feature of the design stress is slightly higher. This leads 1993-1-517 and BD 56 are based on code is the application of the partial to slightly increased resistances. quite different theories but produce safety factor for material strength, The treatment of Class 4 cross similar results for shear resistance. RE γM. Eurocode partial factors sections and beams with longitudinal Little change would be required for an U could generally be used with the stiffeners (which are treated as Class assessment version of the Eurocodes

T characteristic material properties 4 cross sections in EN 1993) differs when the required material properties determined from as built records or significantly, however, from that in were met in terms of ductility; the from tests. Additionally, EN 1990 sets BD 56 and brings significant benefit. shear resistance of webs is not very UC out means of determining material Class 4 beams without stiffeners are sensitive to imperfections, so no resistances and factors in its section treated by making reductions to the adjustment would be necessary.

TR on design assisted by testing. compression areas and then checking S 22 Recommendations for assessment 52 Eurocodes for bridges

The implications of the different • Even if the cross section is in One important caveat is that theories are, however, more Class 3 or 4 (so that the bending longitudinal stiffeners must not be significant in the design of transverse resistance is limited to first yield), susceptible to torsional buckling if stiffeners as Höglund’s theory (used the interaction is performed using this method is to be used, otherwise in the Eurocode) places less demand the plastic bending resistance. adequate ductility does not exist. If on their strength. This is reflected The interaction is truncated by stiffeners are susceptible to this mode in EN 1993-1-5, which allows lighter the requirement to limit the of buckling, a method analogous transverse shear stiffeners to be moment to the elastic moment. to that in BD 56 would again be designed than would be permitted This has the effect of permitting required. EN 1993-1-5 Section to BS 5400 Part 3. The rules in EN almost full web shear resistance 10 provides such a method. The 1993-1-5 have still been shown to with full bending resistance, as rules in EN 1993-1-5 for checking be conservative and could be refined shown in Figure 6, which reflects susceptibility to torsional buckling are further in the assessment Eurocode, the findings of recent non-linear themselves a little conservative as they as suggested in Reference 18. parametric finite element studies. ignore restraint from the parent plate Shear - moment interaction The rules do rely on ductile to which they are attached. Rules could easily be provided to include EN 1993 produces a more behaviour and an amount of strain this restraint and also measured economic check of shear and hardening being available. If imperfections and make a less moment interaction than does BD relevant limits for these were not conservative estimate of torsional 56 and consequently has been met, the simplest resolution in this buckling strength. Modified versions used in assessment to justify not case would be to revert back to the of the rules in BA5619 could be used. strengthening existing bridges. It is more conservative provisions of BD more economic for three reasons: 56, adapted to make them fit the Lateral torsional buckling general Eurocode terminology. and distortional buckling • Shear does not interact with lateral torsional buckling Longitudinally stiffened cross sections Whilst BD56 gave extensive empirical resistance. It only interacts are treated in essentially the same guidance on lateral torsional with cross section resistance way in EN 1993-1-5 as for unstiffened buckling, EN 1993 takes a more cross sections, so the same economic theoretical approach. EN 1993, • The interaction diagram is benefits can be obtained. To BD as a general approach, gives only a continuous curve, rather 56, the check of the cross section an expression for slenderness, than a series of straight would have to be performed on a lines as was the case in BD panel by panel basis in such a way 56, as shown in Figure 6 that any shear stress at all has the effect of reducing bending strength.

where Mcr is the elastic critical buckling moment. No guidance

is given on the calculation of Mcr, which tends to lead the designer towards performing a computer elastic critical buckling analysis for its determination. The advantage of this in an assessment context, is that it usually produces a less conservative prediction of slenderness and strength reduction than does a simplified hand calculation see Figure 7. The reduction factor slenderness curves themselves would need to be adapted to allow use of measured imperfections in the assessment code, but this would be straightforward to do and could be adapted from provisions in BD 56. For quicker estimates of strength, STRUCTURE EN 1994 gives simpler methods of calculation of slenderness without the need for computer analysis. S

Figure 7 - Elastic critical buckling analysis of historic bridge edge girder 23 52 Recommendations for assessment Eurocodes for bridges

Flexural buckling of struts Fatigue added to the determination of Flexural buckling slenderness is The approach to calculation of characteristic material properties, similarly written in terms of a fatigue life is unlikely to be different as discussed previously, where no as-built records are available. The critical buckling force Ncr so that, between design and assessment, but it may be appropriate to use section on creep and shrinkage a different vehicle spectrum for would also require some modification assessment than for design based as usually the detailed concrete on measured traffic conditions. EN composition information required Once again, this allows the assessor 1993-1-9 already allows reduced for these calculations will not be to use a computer elastic critical factors of safety to be used where available for older structures. In buckling analysis which usually damage tolerant conditions exist (i.e. such cases it would be possible to produces a less conservative the structure’s details are regularly give recommended values based on prediction of slenderness and strength inspected for fatigue cracks and there common concrete mixes that were reduction than does a simplified is some redundancy available in the used at the time of construction. hand calculation of slenderness, event of a component failure) and For reinforcement, requirements for particularly in cases where otherwise this approach is particularly suitable compliance with current execution simplified effective lengths would for assessment where monitoring standards will need to be removed need to be used. A tapered arch of the structure is possible. It would and the properties assumed for with hangers and transverse also be possible to incorporate validity of the design rules, such bracing, see Figure 8, provides a modifications to the S-N curves where as ductility, clearly stated. Where good example of a case where an fracture toughness properties were reinforcing steels do not comply with effective length would be difficult measured directly in the structure. EN 1992, alternative rules such as to determine without a computer those found in BD 4422 may need analysis. The reduction factor- Overview of the economic to be used, although generally it slenderness curves themselves would benefits of Eurocode 2 should be possible to modify the EN again need to be adapted to allow 1992 rules to suit. This might include use of measured imperfections in the and the modifications removing consideration of strain assessment code, but this would be required to produce a hardening in the flexural resistance straightforward to do and could be concrete assessment code check where the existing steel is adapted from provisions in BD 56. brittle. Guidance will also need to be Serviceability The following discussions are included for the treatment of plain based around Eurocodes EN 1992- round bars which are not covered by Most of the design requirements 220 (and its cross-references to EN 1992 because of their very poor involving serviceability would be EN 1992-1-121) and identify, in bond characteristics. The properties retained in an assessment version of outline, the modifications required of commonly used historic reinforcing the standard, but inspections would to produce an assessment standard. steels could usefully be given in this determine to what extent the design The various sections of the code section. An assessment National had succeeded in that regard. are reviewed one by one. Annex would be a good place for Fasteners and welds General and basis of design this information to be recorded. Most of the provisions for bolts The modifications required for the and welds would be able to remain Durability and cover general and the basis of design in an assessment standard but section would be similar to those As for structural steel, most of greater coverage of the resistance discussed for steel above. the design requirements involving of other connectors, such as rivets, durability would simply be deleted in Materials would need to be added from an assessment version of the standard older codes such as BD 56. The section on materials would as inspections would determine to require modification for concrete what extent the design had succeeded to remove the need for compliance in that regard. The implications of with current execution standards low concrete cover would, however, and reference would need to be need to be considered at the ultimate limit state where this adversely affected anchorage and lap lengths; S the EN 1992 rules assume a minimum cover equal to the bar diameter in this RE regard. Deterioration would need

U to be accounted for via a condition

T factor in the resistance equations in the same way as is currently required in UK assessment standard BD 21. UC TR

S Figure 8 - Typical tapered arch 24 Recommendations for assessment 52 Eurocodes for bridges

Shear The variable angle shear truss model provided in EN 1992, see Figure 10, provides significant benefit for reinforced concrete sections, and using this model for assessment will provide an important tool. The resistance model is essentially a plastic solution that allows the resistance to be tailored to the reinforcement provided. Where concrete shear stresses are not high, greater link resistances can be developed by using a flat truss angle. Greater Figure 9 - Non-linear analysis of slender piers (defl ected shapes shown for various load cases) shear stresses can also be allowed compared to BD 44, which will often Structural analysis give improved assessed load rating As for structural steel, the structural Plastic analysis would also be more for beams with slender webs. analysis section would require readily justifiable for use in assessment Torsion little modification as many of the since in-service performance can As for shear, the rules for torsion assumptions would remain valid for be visually assessed. The Eurocodes provided in Eurocode 2 would require assessment. For non-linear analysis set out the requirements for plastic very little modifications and could however, measured imperfections analysis so only a relaxation in the be directly adopted for assessment. would need to be used if the actual conditions of use is likely to be needed Once again, the variable angle truss construction tolerances used were in the assessment version, provided model gives the designer an important not known. Non-linear analysis is a that requirements for rotation tool to tailor the resistance model particularly appropriate assessment capacity are met. To that extent, it to the reinforcement provided both tool because it produces the most would be necessary to give guidance longitudinally and transversely. realistic prediction of structural on rotation capacity for reinforcing Strut and Tie behaviour and resistance. However, steels not covered by EN 1992. the practical application of the The strut and tie rules could also be technique is more limited for concrete Ultimate limit states directly adopted for assessment, see structures than for steel ones because Figure 11. Since strut and tie analysis Bending of the added complexity. For is a special case of plastic analysis, members where the predicted failure The rules for flexural resistance once again the rules allow models is a flexural one, such as for beams would require little modification to be developed which suit the and slender piers, see Figure 9, the but reinforcing steels would need actual arrangement of reinforcement analysis is relatively straightforward. to be classified for ductility; there provided. Some additions to the For more complicated cases where is a small amount of benefit that Eurocode for assessment would, concrete failure under a multi- could be gained from these rules however, be useful. The stress limit axial stress field is concerned, the because of the consideration of in struts where there is transverse analysis can be much more time reinforcement strength beyond yield, tension is currently rather conservative consuming, iterative and difficult though this would depend on the in cases where the tensile strains to agree with checkers because of ductility of the reinforcement and are small and the stress limit could the number of concrete models the strain hardening characteristics. be refined to include the effect of that exist. Guidance on the latter Moment redistribution would also tension angle in relation to the strut could usefully be taken from fib provide benefit, as allowed by EN axis and the tensile strain. Further, bulletin 4523 among other sources. 1992, though there was previously more guidance should be included allowance in BD 44 for this also. on by how much the idealised model can depart from the elastic flow of forces because concrete has limited ductility and excessive STRUCTURE departures from such conditions will reduce the allowable strut and node compression limits. Guidance on quantifying this reduction could also be usefully added. S

Figure 10 - Variable angle shear truss model from Eurocode 2 25 52 Recommendations for assessment Eurocodes for bridges

Figure 11 - Simple application of strut and tie rules for bridge pier Membrane rules and Serviceability Detailing Rules sandwich models Most of the design requirements Section 8 of EN 1992, dealing The membrane rules in EN 1992-2 involving serviceability could be with detailing, would have to be Clause 6.109 would be useful for retained in an assessment version modified substantially to deal with assessment where shell element are of the standard, but inspections are non-compliant details. Detailing required in analysis due, perhaps, to generally more useful in determining rules tend to either be provided the complexity of structure geometry serviceability performance. to ensure adequate workmanship and the lack of applicability of other This again would be consistent or to guarantee compliance with codified assessment rules. These rules with previous UK practice. assumptions in the design rules. allow stresses to be used directly Fatigue Detailing rules relating to from an FE analysis model for the Checks of fatigue in concrete and workmanship could simply be deleted assessment calculations. However, the steel were not covered to any as poor workmanship could be significant limitation of the approach significant degree by previous UK accounted for via a condition factor. is that design is done element by assessment standards, but fatigue This would include, for example, element and hence no redistribution can be important in some structures minimum bar spacing requirements; between elements is possible. This where the details are unusual. Details non-compliance will potentially lead makes the use of the membrane that may need assessment include to poorly compacted concrete but rules conservative compared to other bars connected with threaded this should be possible to identify member design rules (such as those couplers and bent bars used at from examining the surface or taking for shear and torsion) and these locations where the bars are highly a core sample from the concrete. member rules should therefore always stressed. Both these situations can Detailing rules which affect the be used where they are applicable. lead to fatigue problems24,25. EN applicability of design rules will need Within an element, however, there 1992 provides a realistic assessment to be maintained and guidance is some scope to tailor the analysis of fatigue in both these situations provided on how to handle non- to suit the reinforcement directions. and would require minimal compliant details. For example, modification. The coupler failure minimum cover rules are required in Figure 12 was predicted by the to ensure that the rules for bond rules in Eurocode 2 for example. strength can be used. Where the minimum cover requirements are not met, additional rules will need to be provided to determine the effects on the bond strength. Rules on minimum bend radii for

S bars are also given in the design code to prevent crushing of the concrete inside the bends and to limit RE bending cracks in the reinforcement

U to acceptable levels. Where these

T minimum radii are not complied with, explicit checks on concrete stress can still be made inside the UC bends but additional information may need to be provided for fatigue

TR Figure 12 - Broken reinforcement coupler due to fatigue strength calculation of such bars. S 26 Recommendations for assessment 52 Eurocodes for bridges

Conclusion The new Structural Eurocodes lend themselves to adaptation for assessment of structures because the rules are generally based around the application of first principles. This not only means that the rules can be adapted relatively easily but they also provide scope for determining an accurate prediction of the true ultimate strength of a structure based on structure specific information on loads and material properties in a rigorous manner. This paper has identified areas where the Eurocodes could be directly applied to the assessment of existing structures and those which would need modification. It has also identified areas where the Eurocodes give increased resistances compared to existing codes. The amount of work to produce an assessment version of the Eurocodes would not seem to be prohibitive, even given a modest budget, if the scope was limited to adapting the existing rules to assessment without refining them further for economy. This should be the priority for the work because the Eurocodes themselves provide a better estimate of strength than many of the UK assessment standards and the lack of a Eurocode assessment standard is likely to lead to inappropriate application of the design Eurocodes to assessment. The secondary task of updating the rules and the partial factors would be considerably more time consuming and would mostly be equally applicable to design. It is therefore unlikely that this refinement would be undertaken solely for an assessment code. STRUCTURE S

27 52 Recommendations for assessment Eurocodes for bridges

References 1. BS EN 10025 (2004): Hot rolled products of structural steels. British Standards Institution, London. 2. BS EN 1090-2 (2008): Execution of steel structures and aluminium structures - Part 2: Technical requirements for steel structures. British Standards Institution, London. 3. BD 50/92: Technical Requirements for the Assessment and Strengthening Programme for Highway Structures Stage 3 - Long Span Bridges, Design Manual for Roads and Bridges, Highways Agency (1992) 4. BS EN 1991-2 (2003): Eurocode 1: Actions on structures - Part 2: Traffic loads on bridges. British Standards Institution, London. 5. BD 21/01: The Assessment of Highway Bridges and Structures, Highways Agency (2001) 6. BS EN 1990 (2002): Eurocode - Basis of structural design. British Standards Institution, London. 7. Shetty, N. and Chubb, M. (2001), Probabilistic Methods for Improved Bridge Assessment, International Symposium on Bridge Management, Singapore. 8. Shetty, N. Chubb, M.S. and Manzocchi, G.M.E. (1998), Advanced methods for the assessment of bridges, Int. Symposium on Management of Bridges, Institution of Civil Engineers, London. 9. BS EN 1993-2 (2006): Design of Steel Structures. Part 2: Steel bridges. British Standards Institution, London. 10. BS EN 1994-2 (2005): Design of Composite Steel and Concrete Structures. Part 2: General rules and rules for bridges. British Standards Institution, London. 11. BD 56/96: The assessment of steel highway bridges and structures, Highways Agency (1996) 12. BS EN 1993-1-10 (2005): Design of Steel Structures. Part 1.10: Material toughness and through-thickness properties. British Standards Institution, London. 13. B. Johansson, R. Maquoi, G. Sedlacek, C. Müller, D. Beg (2007), Commentary and worked examples to EN 1993-1-5 "Plated structural elements", JRC Scientific and Technical Reports, Italy 14. PD 6695-2:2008 Recommendations for the design of structures to BS EN 1993-2:2006, BSI (2008) 15. BS 5400:Part 3 (2000): Design of steel bridges. British Standards Institution, London. 16. BD 61/96: The Assessment of Composite Highway Bridges and Structures, Highways Agency (1996) 17. BS EN 1993-1-5 (2006): Design of Steel Structures. Part 1.5: Plated structural elements. British Standards Institution, London. 18. Presta, Hendy et al Numerical validation of simplified theories for design rules of transversely stiffened plate girders, The Structural Engineer, Volume 86, Number 21 pp 37 - 46 (4/11/2008) 19. BA 56/96: The assessment of steel highway bridges and structures, Highways Agency (1996) 20. BS EN 1992-2 (2005): Design of Concrete Structures. Part 2: Concrete Bridges. British Standards Institution, London. 21. BS EN 1992-1-1 (2004): Design of Concrete Structures. Part 1.1: General rules and rules for buildings. British Standards Institution, London. 22. BD 44/95: The assessment of concrete highway bridges and structures, Highways Agency (1995) 23. fib bulletin 45 (2009), Practitioners' guide to finite element modelling of reinforced concrete structures, fib, Lausanne, Switzerland 24. Hendy, C.R and Smith, D.A, Design of the Dubai Metro Light Rail Viaducts - Substructure, ICE Bridge Engineering, Volume 162, June 2009 25. Sandberg, J.M and Hendy, C.R, Replacement of the stays of a major cable stayed bridge, to be published ICE Bridge Engineering S RE U T UC TR S 28 Limit equilibrium assessment of 53 drystone retaining structures

Chris Mundell Abstract Graduate Engineer A limit equilibrium analysis program has been developed as part of an Highways & Transportation investigation into the stability of drystone retaining structures. Initial verification of the programme’s function was in relation to field trials conducted in 1834 by Lieut-General Burgoyne, which have been the main reference to date for checking numerical modelling of drystone retaining walls. Parametric studies and investigations of bulging mechanisms are reported Paul McCombie and analysed. Program predictions have been compared with the initial Senior Lecturer results from new small scale and full scale drystone retaining wall tests. University of Bath

Claire Bailey Project Engineer Clarke Bond

Andrew Heath Senior Lecturer University of Bath

Peter Walker Senior Lecturer University of Bath

Introduction Drystone technology is an ancient Most construction in the UK dates Such figures highlight the need form of construction, suitable for to the 19th and 20th centuries. for the means to assess these applications ranging from simple Though poorly constructed walls structures adequately, as current field walls to large earth retaining presumably collapsed shortly after design standards often deem them structures several metres high. their construction, the majority of insufficiently safe4. There are several Typically utilising undressed stone walls have remained perfectly stable difficulties when attempting to assess and constructed without mortar, over decades of usually steadily drystone structures; often, very structural integrity is maintained increasing loading and weathering little is known about any particular through self weight, inter-block of the constituent stone. wall, as their construction usually friction and overlapping of stones. However, many otherwise stable walls pre-dates the strict design guides The technique relies upon the have deformed or bulged and with which are adhered to today, leaving skill of the mason in selecting a little guidance currently available to uncertainties regarding wall thickness, suitable stone for each location in assist engineers in assessment of these age, construction quality, foundation turn, placing each appropriately. structures2, the authorities responsible capacity and the mechanical There are estimated to be some for any potentially unstable walls properties of the material retained. 9000km of drystone retaining are often forced to replace or There are also regional differences structures lining the road and rebuild them - in many cases as many styles of wall construction rail networks1 of the UK, whilst unnecessarily and at great cost. The exist, either necessitated by the globally the total length is many total replacement cost for the walls material properties of the stone or times this figure, with walls found lining the UK’s highways is estimated for purely aesthetic purposes. throughout Europe, parts of to be in excess of £10 billion3. Asia, Africa and the Americas. 29 53 Limit equilibrium assessment of drystone retaining structures

There is in any case an important Objectives and scope Element Code) may provide precise philosophical difference between details regarding wall stability and assessing an existing structure and As part of an ongoing investigation, the potential failure mechanisms designing a new one, in that many several unmortared retaining given sufficient data and careful of the uncertainties that are to be structures of both large and small modelling, but the analysis can take covered by factors of safety in design scale have been built and tested several hours, making parametric have been resolved by the fact that to failure. These experiments are studies of any particular structure the wall has been standing and carefully monitored and both the a lengthy and expensive process. stress changes and deformations at serviceable. While this fact gives no Neither option is acceptable for critical locations are recorded, and assurance that the structure has ever routine use. A computer program has then used to determine the underlying experienced a full design applied therefore been developed to explore mechanisms behind the failures. loading, there remains the important efficient approaches to analysis and In parallel with these studies, the fact that the assessor is most design which might be carried out authors are developing alternative concerned by possible changes from by hand calculation or by a range means of analysing drystone the status quo, in which the factor of simple computing approaches. structures which are then verified of safety must at least exceed one The program is based on a rapid with the gathered physical test data. under permanent loading conditions. 2-dimensional limit equilibrium (LE) Inappropriate interventions such as Current analysis techniques for appraisal for structures of any size, pointing become of greatest concern, drystone walls are either simplistic with the ability to account for any because while they may increase the by considering the static equilibrium deformations or bulges that might compressive strength and stability at of the wall as a monolithic structure, occur. In addition this program is the face of the structure, they can or too complicated, using time being utilised to further understand lead to catastrophic changes in the consuming numerical packages to the mechanisms behind the pore water pressure regime. It is model each element within the wall deformations within drystone walls, therefore very desirable to be able to and backfill. Numerical packages as well as the critical factors that assess the possible impact of changes such as UDEC (Universal Distinct affect this particular construction. in geometry and loading on the structural stability of an existing wall, especially given that old structures often appear to have departed from their originally constructed geometry. It is also important to understand the extent to which structural stability is dependent upon precision in geometry and quality of construction. Standards for modern drystone retaining wall construction are very high, with good practice resulting in very strong structures with a high degree of integrity. However, such construction is time consuming and expensive. A proper understanding of drystone retaining wall stability could lead to narrower structures requiring less volume of carefully placed stone and significantly less construction time. Similarly, an understanding of their tolerance of deformation and of the actual sensitivity to variability within the construction could lead to faster construction. These factors would make it easier to repair

S drystone structures rather than replace them, and easier to replace

RE with a new drystone structure which will be sustainable, re-using materials U where possible, or using locally T sourced materials, and resulting in structures that are in keeping with Figure 1 - Comparison between single and double faced walls. their surroundings. It would also UC make this highly sustainable form of construction a more attractive TR proposition for new constructions. S 30 Limit equilibrium assessment of 53 drystone retaining structures

Drystone construction were built to various standards by the Based on these field tests, several project masons within timber boxes. numerical studies have recently Although many differences exist As the density of the wall material been conducted. UDEC has been between the various drystone is known, together with the overall used by various authors1,5,7 - 10 to construction styles, several common volume of the sample, a stone/void test both the validity of various features are usually exhibited. Typical ratio can be easily determined. A very modes of analysis and to study drystone walls are built in horizontal carefully constructed tightly packed further the various parameters at layers or ‘courses’, with each course double faced wall with almost ideal work within drystone structures. ideally consisting of stones of a Cotswold limestone has around Although highly informative, these 'uniform' thickness, retaining a 20% voidage, while over 40% is investigations are both complex and straight and level appearance. The possible within poorly built walls. The time consuming, often requiring cross section of the wall usually consequences of a high void ratio several hours to run a single cycle consists of a well made, tightly- are more extensive than just reduced of analysis. Work is currently packed outer face, with a core of weight of the structure: a loosely being carried out in conjunction smaller random material packed packed wall gives the blocks within it with this project to develop three behind. Some drystone retaining a much greater opportunity to rotate dimensional models of the full scale walls follow this core material directly and slide, facilitating bulging and tests described in this paper. with the retained backfill material, other deformation or even collapse. whilst others have a second inner face, usually less well finished than Previous work the outer face. ‘Tie-stones’ span from the outer to the inner face, To date, despite its widespread binding the wall together. Where use, only a limited number of there is no inner face, tie-stones are investigations into drystone behaviour often used to anchor the outer face have been conducted. Until recently, further back into the packing fill. the only physical test data for full Coping stones can act in a similar scale drystone walls dated back over manner, spanning the entire width of 170 years to work conducted by the wall at the crest, see Figure 1. Lieut-General Burgoyne in 18346. Each block within the wall should Burgoyne built four full scale granite ideally be in contact with several other walls, up to 6.1m tall, 6.1m long and stones, and pressure upon any part of varying thickness, in an attempt of a freshly placed stone should not to quantify the effect that the wall cause any rocking or lifting at the profile has upon stability. These walls opposite corner. In practice it is usually were then gradually backfilled until necessary to wedge in small pieces either full retention was achieved or of rock known as pins to prevent collapse occurred. Movements and rocking. The unavoidable presence general observations were recorded of these pins presents a weakness for upon the placement of each layer of all drystone structures, especially as fill but only reported posthumously weathering of these smaller elements in 1853 from Burgoyne’s notes. has a substantial effect much more quickly than for the larger stones. Pins are often used to allow a more even appearance to the face and assist in drainage by tilting stones so that their outer surface is in the W plane of the face. Thus the face of a Earth structure can often give a misleading pressure, Pa impression of a very tight well-ordered Pv construction, whilst behind the face there are substantial voids held open STRUCTURE by a large number of small pins.

Depending on the quality of Ph workmanship and the material used, the density of the walls can

Resultant differ vastly. Void percentages within W + Pv drystone walls have been generally considered to range between 10% 5 and 20% , however, measurements Ph carried out within this project on a range of test sections showed these values to be too low. Sections of wall S Figure 2 - Determination of resultant forces 31 53 Limit equilibrium assessment of drystone retaining structures

Program operation By analysing the stabilising forces within the wall and using Coulomb’s earth pressure coefficients to determine the horizontal and vertical stresses acting at each level up the back of the wall11, the magnitude and direction of the overall thrusts are determined, see Figure 2. The initial wall geometry is entered along with the material properties of both the wall and the backfill (mass, friction angles, etc) and the eccentricity (ε) is calculated at a number of levels to generate a thrust line as shown in Figure 3. In addition to forces arising from the self-weight of the backfill, patch surcharging may also be applied. Figure 3 - Program operation screen Additional pressure is then applied to the backfill, spreading out by a ratio of the masonry were zero, a thrust bulges. Once the parameters are set of 1h:2v. This is clearly a simplification, line in front of the middle third and the wall profile is generated, the as used for example in BS800612, would simply result in the stone user may move the individual blocks compared with the more rigorous at the back of the structure being to any position, either by typing in a approach of Bolton13 as suggested in progressively unloaded, which need new co-ordinate, or by clicking the BS80024, but for the present purposes not have any immediate serious cursor at a new position for the front this approximation allows the consequences. As the thrust line face. Recalculation and redisplay combination of rapid calculation and moves further forwards, the area of the new thrust line is virtually reasonable accuracy required here. carrying the vertical load reduces, so instantaneous, see Figure 4. In such Further justification of this approach increasing the stress. The compressive a manner, idealised wall structures was given by Corte14. It is currently strength of most masonry, including can be deformed to represent assumed that the surcharge will drystone, is usually relatively high commonly observed drystone bulging have no effect upon the calculated compared with the stresses acting. patterns, to assess their effect thrust line until the expanding area Therefore, compression failure of upon overall stability. Conversely, over which it is distributed crosses the main stones is very unlikely, but existing walls which have bulged the boundary of the wall. Although a concentrated thrust may cause may be quickly recreated using the the analysis is two-dimensional, localised crushing of weakened pins program to ascertain their stability. three-dimensional load dissipation or a flexural fracture of some stones can be assessed as spreading in both leading to further deformation. In horizontal dimensions. The most addition, foundation settlement might problematic loading is wheel loading give rise to significant deformations. from a heavy vehicle, so it is important Given sufficiently strong masonry and to model the three-dimensional foundation, failure would only occur distribution, even if crudely. A more once this thrust line breaches the wall sophisticated stress distribution face (ε > 1/2). However, individual calculation is simple to implement, block rotation will occur before this,5 but given the uncertainties in wall as the block at the face must carry and backfill stiffness and anisotropy, all of the lateral thrust as well as a this may overcomplicate the vertical load which is moving closer analysis without adding value. and closer to its leading edge. This Standard masonry construction could result in an immediate rotational S recommends that for stable failure of the entire structure, and construction the eccentricity of the indeed such a mechanism accounts for the heights reached by the

RE thrust line must remain within the Burgoyne walls which failed5. It middle third of the structure (ε = 1/6 U of the base width away from the may also be noted that a crushing T neutral axis). If the masonry were failure at a point of contact or of a to behave in a linear elastic manner pin may lead to collapse before the and deformations were very small, line of thrust reaches the front face. UC this would result in no tension being A significant advantage of the taken at the back of the structure. program is the ability to allow the user TR However, even if the tensile strength to deform the wall model and induce

S Figure 4 - Example of user defi ned bulging 32 Limit equilibrium assessment of 53 drystone retaining structures

Program validity check Parametric analysis Initially, the program was validated Due to the nature of the program, a the backfill. As this vertical force against Burgoyne’s four test walls. parametric analysis of any structure acts against the overturning forces The geometries of each wall is a rapid process. This has a two- and stabilises the wall, this is a value were recreated, and the material fold application; firstly it allows users which would ideally be as high as the properties entered from Burgoyne’s to quickly grasp which parameters interface allows, although in reality tests6. Backfill heights were then have the greatest impact upon it is not always guaranteed that the systematically increased by 300mm wall stability and secondly it allows full friction angle will be achieved. (simulating Burgoyne’s test procedure) engineers in the field a greater One of the most variable and until the thrust line reached the wall flexibility when assessing existing difficult to ascertain parameters is face, indicating failure via toppling. walls. Once the cross section of the density of the walls themselves. The final simulated heights were a wall has been recreated within Non-destructive testing methods, very similar to those recorded by the program, each variable may be such as ground penetrating radar, Burgoyne, and indeed also similar altered to examine the safety factors or horizontal coring can be used to to previous attempts using other at the worst possible conditions. give some indication of wall depth, more sophisticated and complex From the program it is apparent profile and even voidage. Whilst 8 numerical packages , see Table 1. that for any given geometry of wall, the density of the rock will not vary Both the first and second of several parameters are dominant for greatly, its age and the construction Burgoyne’s test walls were backfilled stability. For example, the assumption style, and the skill of the mason to their full height without excessive of a 1h:2v load spread from will all affect a wall’s overall density movement, and by using the LE surcharges means that the loads must and hence the total volume of voids program it can be demonstrated be close to a wall to have an effect, within. Whilst this voidage has that the thrust line lies within the but it is also found that loads must little impact on wall stability when boundaries of the wall. It should be relatively large, corresponding to changed by a few percent, the value be noted that for both these walls wheel loads from the heaviest trucks. may vary by much more than this, the eccentricity is outside of the This corresponds with anecdotal as mentioned above. Low density middle third at the base, indicating evidence, confirmed by numerical reduces the wall’s stability in terms of uplift at the heel. The third and modelling studies9, that there is a both sliding and overturning. Perhaps fourth walls both fell before full relationship between increasingly most critically, a reduction in density height of retention was achieved. heavy traffic and failures of walls allows easier movement and rotation For both these wall geometries, which had been safe for many years. of the individual blocks, determining the LE program predicted failure The friction angle of the backfill the flexibility of the wall and the at a height similar to that found material is critical to wall stability. amount of bulging that may occur. by Burgoyne, although it has been This angle determines the coefficient demonstrated that consideration of active pressure (k ), which in turn Bulging investigation of individual block rotation gave a determines the magnitude of the Bulging is common in drystone walls, a tighter correlation with actual horizontal forces upon the retaining failure heights5. To allow this to be usually occurring at roughly a third to wall. A stiff, tightly-packed backfill half the height of the wall, creating seen in the program, the direction material might have a high density, of the resultant force at each level a distinctive ‘belly bulge’ shape, but its consequently high angle of see Figure 5. Upon investigation is also shown at the point at which friction is likely to result in a lower it acts. A resultant which points of the effects of bulging and wall horizontal pressure than a much deformation, it was discovered in front of the toe of an individual looser, yet less dense material. block may result in rotation of that that far from causing instability, There is also the friction between the block. With regard to the results a moderate bulge may indeed wall and backfill to consider. Due to shown in table 1, this would indicate increase the safety of a wall against the nature of drystone construction, a failure at 5.2m for Wall C, bringing certain failure modes when subject the wall faces are generally rough, it in line with the observed results. to surcharge loading conditions. which allows the inner wall face adjacent to the retained material to attract some of the vertical load from STRUCTURE S

Table 1 - Comparison of Burgoyne tests 33 53 Limit equilibrium assessment of drystone retaining structures

Bulging begins when the loads Bulging and movement can also Below the bulge, the wall is leaning behind the wall cause blocks or occur much higher up a wall – usually forwards, causing the active pressures entire sections of wall to move, caused by localised surcharging, or within the backfill to have a much and the resulting movement disturbances to the wall itself, such greater effect upon this portion of causes both the forces acting on as growth of vegetation, although the wall. The magnitude of the force the wall and the equilibrium of its this is generally detrimental to will be greater, but the downwards own mass to change, such that a wall stability and can easily lead component will be most increased, new equilibrium position is found. to partial or full wall collapse. so increasing the stability of this Were this not the case, the wall Once a bulge is formed, the pressures portion of the wall, provided that would continue to move resulting acting upon the wall must change the face has not moved so far in collapse. This rearrangement in response to the new geometry. A forwards that individual blocks are usually occurs lower down the section of a typically bulged wall is no longer supported. Overall, these wall, and can be due to slips in the shown in Figure 6, highlighting the changes tend to be in favour of retained earth, increased pore water common features. Above the bulge, increasing wall stability, and walls pressure or an increase in loading the wall is leaning back somewhat, have commonly remained perfectly conditions, or the equilibration of having a two-fold effect. Firstly, safe for years whilst displaying this negative pore pressures within the it stabilises the wall by moving its type of bulge without any detrimental backfill. Bulging probably occurs centre of gravity away from the effects. However, new works, such much more commonly than is toe of the wall, which is usually as excavations for services at the appreciated, but is usually on a the overturning point. Secondly, it toe of deformed walls, or changes scale too small to be noticed. reduces the magnitude of the forces in loading, are common factors applied to the wall by the backfill. attributed to triggering collapse. Due to the flexible nature of these walls, significant movements may take place before a failure occurs, giving visible warning signs. Final collapse can occur either by toppling or bursting, but is usually a combination of both. S RE U T UC TR Figure 5 - Deformed wall Figure 6 - Section of typically bulged wall S 34 Limit equilibrium assessment of 53 drystone retaining structures

Small scale testing A series of small scale tests has To overcome three-dimensional the collapse heights of these small been conducted to determine if effects long blocks were used, each scale tests. It was assumed that the observed drystone behaviour can spanning almost the whole width interface friction between the wall be recreated in smaller, simpler of the steel box but with gaps at blocks and the backfill was ⅔ the full experiments. To house the tests, a either end to allow small rotations value of the backfill’s internal friction steel box was constructed, with the of the wall elements. Both timber angle. Evidence gathered by the small capacity to hold scale walls 500mm and concrete block walls were tested scale tests supports this assumption, in height and 500mm in width, see independently; the timber blocks although in practice it is difficult Figure 7. The box was lined with a were quickly discarded as their to ascertain precisely how much of double layer of plastic sheeting, to densities proved too low for realistic the backfill’s full friction angle has help reduce friction at the edges modelling of drystone behaviour been mobilised against the wall. This and hence minimise end effects16. (5.5kN/m3 as opposed to 24kN/m3 obviously has a large impact upon As the goal of these experiments for the concrete blocks) although the wall stability, although it is expected was to reproduce full scale drystone data proves useful for comparison that ground settlement over time behaviour, small pellets (2-3mm with the LE program results. and the rough nature of drystone diameter) of lead shot were used For each test, the scale walls were structures results in the full friction as backfill to induce sufficient fully constructed without any angle being mobilised for in-situ walls. pressures to cause deformations and retained backfill, and then slowly failures. The lead shot used has an backfilled. Results from the small Large scale testing 3 uncompacted unit weight of 50kN/m scale tests are shown in Table 2 In addition to small scale testing, full O and an internal friction angle of 31 , together with the backfill heights scale walls have been built and tested allowing the generation of sufficient predicted by the LE program. to failure to validate modern theories lateral pressures to overcome the From table 2 it is clear that the and analysis tools. To test the drystone stabilising forces within the test walls. program is accurately predicting walls, a bespoke test facility has been constructed, allowing recreation of localised or general foundation settlement, backfill settlement and localised surcharging, see Figure 8. Each wall was constructed of Cotswold limestone by skilled masons. At 2.5m high and over 12m long, the test walls are large enough to be representative of many of the walls found throughout the country, and are built using traditional methods, including regular placement of through-stones and a line of coping stones at the peak of each wall. The first wall varied in thickness from 600mm at its base to 400mm at its peak, and was constructed to a high standard. The second wall was of Figure 7 - Test Setup an intentionally poorer quality, and 100mm more slender throughout. A large range of instrumentation is used to monitor each wall, including extensive surveying, multiple transducers, load and pressure cells,

high resolution imagery and video STRUCTURE footage. With this vast range of data, the critical events that lead to the failure of each wall can be better understood and incorporated into both our general understanding and the theories and programs which are used to evaluate a wall’s stability. For each test the LE program was used to check initial stability. Ideally, each wall should be able to retain

Figure 8 - Bespoker test facility S the full height of fill intended

35 53 Limit equilibrium assessment of drystone retaining structures

relatively lightly loaded end sections help support the central section. This load shedding subsequently allowed the first wall to deform to a greater extent than would have been possible had the wall been acting purely in plain strain, with the wall profile showing the coping overhanging the toe by some 500mm prior to failure. Following this experience, the second wall was intentionally of a poorer quality than the first, showing a number of ‘running joints’; a term used to describe when the joint between two adjacent blocks is similar in position to a similar joint above or below. These prevented the transfer of load along the length of the wall, and opened up as loading proceeded to allow the central section to move more easily relative to the adjacent sections. The result of this was that Table 2 - Small scale testing results the wall behaved in a manner more akin to that represented in the LE with a sufficient margin of safety Throughout each experiment program, and probably more akin during installation of remaining the geometry of the wall face is to sections of wall which actually instrumentation and loading plates, constantly monitored, allowing the fail in practice. Most wall failures whilst being close enough to its wall profile to be recreated within are localised at weak points, rather ultimate conditions that the proposed the LE program, so that stability than being general failures along the movements and loadings can take can be assessed as the loads are full length of a wall. Thus three- the structure to failure. Both walls changed and the walls deform. The dimensional effects, in particular complied with these criteria; however images of the thrust lines generated the transfer of load along the line the second wall was significantly immediately prior to failure for the of a wall, can certainly help support less stable, with the eccentricity first two tests are shown in Figure 11. a weak section, but failure is much on completion lying in front of more likely to occur where loss of From Figure 11 it is evident that the the wall’s middle third (fig 11c). such support results in a behaviour structures are both on the verge of which is nearer two-dimensional or The initial phase of each test to collapse, although three-dimensional plain strain and two-dimensional date has involved the raising of effects may have given added stability, analysis is therefore recommended for the platform up to ensure that especially in the case of the first test most situations. The final observations the maximum possible friction wall. Both walls developed bulges were recorded just minutes before is generated at the wall/backfill only through the central region of the failure, when the wall was remaining interface. In real walls, full friction is wall adjacent to where the surcharge stable in the unloaded state but likely to be achieved due to settlement loading was applied. The high friction giving indications of imminent of the backfill following construction generated between the courses collapse (increased movement for of the wall. In order to take full allows a tensile strength to develop only minimal surcharge loading), control over this important parameter along the length of the wall, so the the inverse movement is applied, and and this was successfully reproduced the wall is moved upwards in relation to the backfill via the jacked platform. Load cells supporting the wall show a steady increase as the platform is lifted, until the full friction is mobilised S when the loads level out, see Figure 9. Following the initial raise of the RE platform, Wall 1 was subjected

U to a combination of forwards

T rotation and surcharging, whilst Wall 2 was simply surcharged until failure occurred. Both walls failed

UC by toppling, though each displayed a great deal of movement prior to

TR collapse, including block rotation and sliding, see Figure 10a, b and c. S Figure 9 - Load cell readings during platform raise 36 Limit equilibrium assessment of 53 drystone retaining structures

Figure 11 - Limit equilibrium analysis of physical tests: (a) wall 1; (b) wall 2

by the LE program. The surcharge loading was displacement rather than load controlled, allowing a progressive deformation past peak load with full control, allowing safe collection of data, until eventual collapse associated with excessive distortion of the geometry of the structure. Both structures remained absolutely stable, with no ongoing deformation, when the surcharge loads were removed, even though they were very severely distorted. Conclusions The limit equilibrium analysis program described in this paper has enormous potential compared to numerical analysis packages; its simplicity allows any engineer with a basic knowledge of a wall’s geometry and material properties to obtain a reliable understanding of the factors influencing its stability, without the need for the detailed knowledge, advanced design parameters, time and expertise that are needed for reliable numerical analyses. The program's flexibility in use allows walls of any geometry with variable backfills to be analysed, and the application of surcharging can be applied to

represent circumstances such as new STRUCTURE constructions in the proximity of the wall or increased vehicle loading. Acknowledgements The work carried out in this paper Figure 10 - Physical test results: (a) wall 1 prior to failure; (b) wall 2 was conducted at the University prior to failure; (c) graph of wall displacements of Bath in conjunction with Southampton University, funded by the Engineering and Physical Sciences Research Council (EPSRC). S

37 53 Limit equilibrium assessment of drystone retaining structures

References 1. POWRIE, W., HARKNESS, R. M., ZHANG, X., and BUSH, D. I. Deformation and failure modes of drystone retaining walls. Geotechnique, 2002, 52, No. 6, 435-446. 2. O'REILLY, M. P. and PERRY, J. Drystone retaining walls and their modifications - condition appraisal and remedial treatment. 2009, RP723, 3. O'REILLY, M. P., BRADY, K. C., and BUSH, D. I. Research on masonry-faced retaining walls. 2nd European Road Research Conference, 1999, 4. BRITISH STANDARDS INSTITUTION(BSI) Code of practice for earth retaining structures. 1994, BS 8002, 5. CLAXTON, M., HART, R. A., MCCOMBIE, P., and WALKER, P. J. Rigid block distinct-element modelling of drystone retaining walls in plane strain. ASCE, 2005, 131, No. 3, 381-389. 6. BURGOYNE, J. Revetments or retaining walls. Corps of royal engineers, 1853, 3, 154-159. 7. DICKENS, J. G. and WALKER, P. J. Use of distinct element model to simulate behaviour of drystone walls. Structural engineering review, 1996, 8, No. 2/3, 187-199. 8. HARKNESS, R. M., POWRIE, W., ZHANG, X., BRADY, K. C., and O'REILLY, M. P. Numerical modelling of full-scale tests on drystone masonry retaining walls. Geotechnique, 2000, 50, No. 2, 165-179. 9. BRADY, K. C. and KAVANAGH, J. Analysis of the stability of masonry- faced earth retaining walls. 2002, TRL Report TRL550, 10. WALKER, P. J., MCCOMBIE, P., and CLAXTON, M. Plane strain numerical model for drystone retaining walls. Geotechnical Engineering, 2006, 159, No. GEI, 1-7. 11. COOPER, M. R. Deflections and failure modes in drystone retaining walls. Ground Engineering, 1986, 19, No. 8, 28-33. 12. BRITISH STANDARDS INSTITUTION (BSI) Code of practice for strengthened/ reinforced soils and other fills. 1995, BS 8006, 13. BOLTON, M. D. Geotechnical stress analysis for bridge abutment design. 1991, Transport and road research laboratory, report 270, 14. CORTE, J. F. Reinforced earth retaining walls under strip load: Discussion. Canadian Geotechnical Journal, 1981, 18, 324-326. 15. WONG, H. N. and HO, K. K. S. The 23 July 1994 landslide at Kwun Lung Lau, Hong Kong. Canadian Geotechnical Journal, 1997, 34, 825-840. 16. BAILEY, C. Model tests of dry stone retaining walls. MEng Dissertation, University of Bath, 2008, S RE U T UC TR S 38 ITSO smart cards in Welsh public transport 54

Vivien Collins Abstract Integrated Transport This paper will discuss the progress of the Welsh Assembly Government Division, Welsh Assembly Government project to introduce smart card technology to public transport in Wales and addresses the achievements of the project to date, how the lessons learnt during its initial phases can benefit other smart card schemes looking to implement within the Integrated Transport Smartcard Organisation (ITSO) environment and how the further exploration of smart card technology will move forward in Wales. Joanna Scott Senior Consultant Intelligent Transport Systems

Sarah Warlow Project Support Intelligent Transport Systems

Wales’ ITSO environment The benefits of ITSO

for smart cards standards in Wales The ITSO operating licence INTELLIGENT TRANSPORT SYSTEM Wales is one of the four nations that The scheme makes use of integrated After becoming a member of ITSO, make up the United Kingdom and has systems that follow a common open signature of the ITSO Operating a population of just over 2.9 million. specification developed by ITSO2, an Licence is the first step in setting up a The Welsh Assembly Government organisation formed by leaders in live transport smart card environment, is the devolved government for the UK transport industry to establish and forms the contractual basis of Wales and is responsible for health, standards in smart card technology a scheme’s relationship with other education, economic development, with the intention to make national scheme owners and ITSO themselves. culture, the environment and multimodal interoperability a reality. In It represents the agreement between transport. Residents aged over 60 contrast to other “closed” or “black all parties to protect the integrity of and disabled passengers of all ages box” smart card schemes that act the environment, and includes clauses have been entitled to free travel on within self-prescribed boundaries, the that cover use of the ITSO systems local bus services across Wales since interoperable ITSO specification is built themselves, service level agreements 2002, and the concessionary fares on the vision that multiple schemes and fees. Although called a Licence, scheme remains part of the Top 10 (both commercial and public sector) it is a binding contractual document Assembly Government commitments will be able to share and interlink that must be treated as such, and to the electorate, where over £50m of acceptance of their ticketing products adequate time should be given to public money is provided per annum and services so that passengers may allow a thorough understanding in revenue support to bus operators experience a seamless journey on any of the implications of signature by each year. In 2008, the Welsh mode of public transport they choose senior members of the operator’s Assembly Government began to issue using electronic ticketing. In addition organisation and, more than likely, standardised contactless smart cards to this, the chip embedded within lawyers. Because the document is to over half a million concessionary the smart card allows opportunities owned by the ITSO membership, pass holders, to further enhance the for extension beyond ITSO and changes to clauses are not easily concessionary fares scheme, provide the transport environment, with negotiable and would have to go quicker boarding times, accurate enough memory to hold additional through a full consultation period data capture and more efficient applications that may be used for with all ITSO members if it were and auditable reimbursement. services such as prepayment, citizen to change - meaning potential entitlement privileges, loyalty bonuses, delays to the start of any scheme. In and beyond. Services of this kind are Wales the Operating Licence took already being used effectively by ITSO months to agree and signature was schemes in Scotland and England. ultimately raised to Cabinet level. S

39 INTELLIGENT TRANSPORT SYSTEMS 54 and POSTmessagingtobegin. card-encoding, ISAMcommissioning, System (CMS),enablingpersonalised and linkedtoaCard Management the systemcanbecommissioned this thattheHOPScanthenbebuilt, on time,resource andcost.Itisfrom in thefuture withminimumimpact remains flexibleenoughtochange to ensuringthatanITSOscheme relationships heldtherein, iscritical given toanOIDstructure, and the goals. However, theconsideration types offlexibilityanddifferent has itsownrequirements, different other OIDsbelow. Everyscheme OID identification,withaseriesof retains thetoplevel(orparent) the Welsh AssemblyGovernment an “umbrella” OIDstructure, where changes inthatstructure. Wales has configurations, andfuture proof for encode theITSOartefactsand provide theabilitytouniquely allocate ticketingproduct ownership, of thestructure ofyourscheme, to set ofidentifiersthatallowdefinition structure towork-thisisessentially a their OperatorIdentification(OID) must firstunderstandhowtheywish addressed before anyother. Schemes mission-critical systemthatmustbe suppliers andadvisorsasitisthe environment andrequires experienced aspects ofenteringtheITSO of themosttechnicallychallenging processed. SettingupaHOPSisone which allsecure ITSOtransactionsare databasethroughpassenger journey simply as“HOPS”),andactsa Processing System-usuallyknown Management System/HostOperator is knownastheAMS/HOPS(Asset The centralhuboftheITSOsystem Setting uptheHOPS without signature ofthisdocument. as thesystemscannotbebuilt lengthy process, andisveryimportant may wellhelptoexpediteapotentially ITSO attheearliestopportunity. This to discussandunderstandthemwith any potentialissuesatthestartand the project inceptionstage,identify have earlysightofthisdocumentat Our recommendation wouldbeto with theHOPSandCMS. maintaining real-time communication and switchedoneachday, whilst thousands ofcards tobeencoded running perso-POSTs toallow formed usingmultipleconcurrently remote centralbureau wasalso and rollout inWales, atemporary system. Duringthebulkcard re-issue model thepassengerdataheldin and LocalAuthoritiestoanalyse the Welsh AssemblyGovernment reporting capabilitieswhichallow The systemalsohascomprehensive with appropriate systemprivileges. functions are onlyaccessedbythose of thedataandtoensure thatcertain partitions toprotect boththe integrity Wales withhierarchy basedsecurity accessed byall22LocalAuthoritiesin the centralisedCMS.TheCMScanbe and secure web-basedinterface with deleted bytheissuerthrough adirect be added,confiscated,hot-listedor products andservicesonthe card to encoded ondemand,andfor smart cards tobeproduced and Local Authoritiesenablenewlive equipment (Perso-POSTs) inWelsh The personalisedsmartcard issuance important forreimbursement. from thenext-especially transactions canbedifferentiated its life,andsothateachoperator’s tracked throughout thecourse of may beindividuallymanagedand public transportoperatorsothatit identification, itisassignedtoa as containingtheuniqueWales When thechipisencoded,aswell identified foroperatorreimbursement. also allowstheITSOtransactiontobe to thesmartcard onbus.TheISAM ITSO ticketingproducts andservices ability tosecurely addadditional may beacceptedandincludesthe which ticketingproducts andservices interoperability, whichdetermines Module (ISAM),aliteral“key”to commissioned ITSOSecure Application basis. InsideeveryPOSTisaspecially depot systemtotheHOPSonadaily from theticketmachinetobus which allowtheflowofdatatoand Machines (ETMs)on-board buses these are theElectronic Ticket Point ofServiceTerminals (POSTs); front office busdepotsystemsand The HOPScommunicateswiththe and theCMS POSTs, ISAMs,PersoPOSTs, ITSO smartcards inWelsh publictransport 40 geographical boundariesorroutes. principles, forexampleadjustmentsin against changeinreimbursement to adjusttheparametersprotect system alsoretains enough flexibility thathastakenplace.The journey) (orpart occurs forthejourney service provider andreimbursement grouped according tolocationand of reimbursement thetransaction is boundaries orregions, atthe point multiple operatorsacross multiple travel). Therefore, ifapassengeruses administration andpaymentforthe Local Authorityresponsible forthe (the boarding pointsignifies the the passenger’s pointofboarding basedon concessionary journey operators are reimbursed foreach CurrentlyGovernment. inWales, factor setbytheWelsh Assembly value usingareimbursement then translatedintoamonetary or geographicalboundaries,and analysed usingtimeparameters to beidentified,grouped and to operatorsallowtransactions this pointthattheISAMsallocated for concessionarytravel.Itisat the amountsowedtobusoperators Authorities canaccuratelycalculate transactionssothatLocal journey extracted datafrom theHOPSon partitioned, receives detailed system, alsoweb-basedandsecurely The centralisedreimbursement Reimbursement system which isespecially importantwhen POSTS across theITSOenvironment, types ofmessagingtoISAMsin provides theabilitytosendcertain transaction isrecorded. TheISMSalso via theISMSasandwhenthat smart transaction isforwarded toitsHOPS different systemssuppliers,theITSO others' ticketingproducts buthave to acceptand/orrecognise each supplier. So,iftwoschemesagree that share ourback-office systems with HOPSownedbyschemes allows Wales tocommunicatedirectly and a“HOPStoHOPS”capability communication withotherschemes; System (ISMS)facilitatesmulti-HOPS The ITSOSecurityManagement ITSOsystems. is linkedtoexternal transactions, theWelsh architecture architecture andprovide interoperable Finally, tocomplete theITSO and ITSOcompliance HOPS, ISMS, external ITSO smart cards in Welsh public transport 54

dealing with “hot-lists” that protect The avoidance of a “big-bang” Mandatory field requirements all schemes against fraudulent card rollout has been beneficial to and data handling business rules or product use. Hotlisting is described Wales, especially in the confidence within the system also ensure that in further detail later in this paper. that now exists with regard to the data recorded accurately and are To maintain system security and quality and security of our national consistently across the country, integrity, each aspect that sits within passenger database. The ability ensuring ongoing quality. the Wales architecture that requires to show flexibility in addressing These extra steps in addressing reciprocal messaging from the HOPS 22 different sets of needs in each data quality have given the Welsh must be ITSO compliant, and tested location has also been essential to Assembly Government a great and certified to this standard. This retain enthusiasm, momentum and deal of confidence in the class of is extremely important, not only buy-in from all the Welsh authorities. the data held in the system - very for the immediate needs of Wales’ Prior to migration, the quality of important given that each Local scheme, but also for the integrity of the data held by Local Authorities Authority is grant-aided based on any operator that sits within the ITSO varied widely and, as part of their the number of live concessionary environment. True interoperability responsibility as data owners, travel card users they have on means that all the secure systems are extensive data cleansing was carried their system each year. Again, the interacting constantly, so the ITSO out by each Authority prior to accuracy of this process and ability community as a whole must remain migration to the new database. to audit this information centrally if confident that any new scheme For the majority, this phase usually required enabled the government owner or “ITSO Licensed Operator” worked through two cycles over two to ensure that public money is protects the security and integrity of years to ensure maximum benefit and spent appropriately (and saved the entire estate and not merely its minimum risk. Usually data were sent where possible), and apportioned own portion of it. On signature of to a contracted cleansing service and fairly. It has also meant that the the ITSO Operating Licence, this also the remaining records were manually, requirement for large investment becomes a contractual obligation, and sometimes painstakingly, searched in the additional integration of although this certification comes at a and updated in-house based on the large-scale data cleansing tools has premium in terms of contractual and up-to-date information available. The been avoided as the scale of any financial commitment, the ultimate relationships between local registrars issues remains relatively small and benefits of cross-scheme capability for during this process also both helped data quality can continue to be passengers merits the investment. and hindered progress depending managed in-house. For any scheme on the strength of information looking to migrate to live smart card INTELLIGENT TRANSPORT SYSTEM Project Implementation that could be obtained at each use, ITSO or not, this aspect can Phase 1: Commissioning Local Authority. Recommendations be endorsed as a major benefit. systems and migration have therefore been taken forward As well as renewed confidence in to further improve internal the quality of the national database, The first major phase of the project relationships and an understanding has been system commissioning records are also partitioned within of the real and sometimes assumed the CMS allowing Local Authorities and the bulk reissue of bus passes boundaries around sharing data. to eligible concessionary travellers. access to the data of residents within Prior to this project, concessionary In addition to these steps, during the their area, with access to others’ travel pass data were held on migration process itself authorities partitions being granted on a per- local databases by each of Wales’ have been able to use a “holding request basis, which allows updating 22 Local Authorities. For a cost- area” within the CMS for any record of records to take place locally. effective implementation of an which may have had a question Allowing this responsibility to remain all-Wales scheme the migration to mark on its validity. This has allowed local, with residents contacting their the centralised CMS database was Authorities to continue checking their own Authority directly in all related necessary to benefit from economies data immediately after their migration, matters, retains a sense that a local of scale; the systems were procured ‘flushing out’ erroneous data after service is being provided, even within in 2006, commissioned in 2008, the initial reissue, although this has the larger national scheme. This with card rollout finishing in 2009. been minimal. This has ensured that is arguably particularly significant The project worked through many a proportion of cards that could given the vulnerable demographic challenges, including technical have been reissued in error have of the concessionary fare scheme. problems mainly experienced been managed over a designated The economies of scale brought by during the integration and testing period of time without any undue the migration to one centralised phases in the first two months after negative impact on the card holder. national system have therefore system commissioning. Building in This has saved time, money, and any been achieved without sacrificing contingency weeks for testing and unnecessary heartache for families the benefits of local services. for each migration phase also proved with recently deceased relatives whose invaluable and although it has taken records may not have been identified. 18 months in total this has been an acceptable timescale for the phased authority-by-authority approach. S

41 INTELLIGENT TRANSPORT SYSTEMS 54 all ITSOimplementationsinthe UK. This hasbeencommonandimpacted undiscovered problems are resolved. the project planasnewpreviously that delayshaveoftenimpacted schemes inthisarea hasmeant public. Therelative immaturityof for theoperatorandgeneral sufficient leveltoprovide confidence need tobetrainedandeducateda to becreated orupdated,anddrivers multiple depotsystemsneedeither requires sufficient time tobetested; machine (e.g.GPSsystems)also that communicatewiththeticket of themanyothersoftware products alongside ITSOcapability. Integration capability forticketingisretained detailed leveltoensure that allformer Software needstobetested toavery met before progress canbemade. set ofdependenciesthatneedtobe occurred mainlyduetothe complex prepared forthechange.Delayshave that alloperatorsare adequately of wholesalemigrationandtoensure involved, bothtominimisetherisks to thenumberofstakeholders is plannedasaphasedactivitydue beyond originalexpectations.Rollout been subjecttosignificantdelay began inNovember2008andhas Rollout ofticketmachinesinWales use byitssmalleroperators. four Welsh transportconsortia for a leadauthorityonbehalfofthe within eachtransportregion by have beeninstalledandare managed directly totheHOPS,central systems that allowdatatobetransferred installed theirownbackoffice systems number oflargeroperatorshave precluded involvement.Whileasmall equipment mightotherwisehave operators, where thecostofsmart locally basedsmalltomediumsized particularly significanttoWales’ ninety transport businesses.Thiswas systems tosupportsmallerpublic and provision ofregionalised depot for ITSOcompliantticketmachines operators forpurchase agreements award ofsliding-scalegrantstobus promoted through theproject bythe inward economicinvestmentisalso aim ofencouraginglocalbusinessand The Welsh AssemblyGovernment’s Machine (ETM)rollout phase 2:Electronic Ticket Project implementation are usedeverywhere by default. in thefieldandsmarttransactions happen whenallticketmachines are all transactionsare smart.Thiswill be trueuptoacut-off pointwhen in theshortterm,butthiswill only card doesnotworkwillbeabated dealing withscenarioswhere the pass willmeanthatthedifficulty in use ofthenewsmart-card asaflash entitlement onboard abus.Parallel the pointwhere theywishtousetheir to facilitateresolution fortheuserat identified clearlyandquicklyenough scenarios cannotbeinvestigatedand the root causeineachofthese or isfraudulentlyused.Unfortunately, scenarios where acard malfunctions and howdriversshoulddealwith on-bus doesnotworkasexpected, scenarios where thetechnology account policydecisionsthatmanage The project hasalsohadto takeinto data thatthesystemhasgathered. basedontheauditable journeys reimbursement forconcessionary have submittedtheirclaimsfor through theHOPSandbus operators the transactionsthathavepassed have beenabletoimmediatelysee the summer. In southeastWales we quicker rollout overthecourse of confidence forasmootherand on theirequipment,givinggreater achieved livesmarttransactions month period,eachETMsupplier Perso-POSTs. thisinitialsix Within live andfittedtoETMPOSTs and 25% oftheWelsh ISAMestate was end ofApril2009,approximately machines inNovember2008.Atthe with ITSOcompliantelectronic ticket The firstoperatorsinWales wentlive been affected andplanschanged. especially where timescaleshave operator hasbeencriticaltosuccess, andeachbus Consortia, Government Authorities, RegionalTransport partnership workingbetweenLocal installation. Goodrelationships and with minimumornoimpactduring can maintaindaytobusiness a verylargegeographicalarea) ensure thatallbuses(oftenacross this hasmeantcomplexplanningto has proven thebestoption,although Scheduling around weekendservices resources hasalsobeendifficult. installation timescaleswithlimited buses across manyoperatorsinshort delivering hundreds ofmachinesto Managing thelogisticsinvolvedin ITSO smartcards inWelsh publictransport 42 implementations intoaccount. principle thattakesfuture commercial product orashelloriginatesfrom a decision tohotlistbasedeitherona ITSO partitionofasmartcard). The products (tickettypes)andshells(the are basedonprioritisingtypesof simple businessrulesaround hotlisting Wales, theinitialapplicationsof guidelines onimplementation.In discussed butthere are nodefinitive hotlisting mightworkhavebeen formed, manytheoriesonhow Since theITSOspecificationwas that thecard isnotvalidfor travel. message indicatingtothebusdriver further useofthecard will result ina is senttoeachticketmachineand is uploadedeachday. The‘hotlist’ the ‘hotlist’whentransactiondata on-board abus,itwillbecome partof HOPS. Ifacard issubsequently used become partofa‘warm-list’inthe are flaggedwithintheCMS and ISAM. Cards reported lostorstolen ticket machineviatheHOPSand on-bus viaITSOmessagingtothe identifying fraudulentorinvalidcards implementation. Thisisaformof the shortterm,hotlistingrequires with changesinthetechnology. In challenges inkeepinguptodate structure, aswellongoing a simpleconcessionaryticketing opportunities topursuebeyond in placeWales hasaselectionof Once theITSOinfrastructure is Welsh ITSOenvironment phase 3:extendingthe Project implementation become thedefaultposition. are installedandsmarttransactions across Wales untilallticketmachines large, wewillnothaveaclearpicture date indicatesthattheproblem isnot machines. Althoughexperienceto cards anddefiningerrors withticket functioning cards vs.fraudulent these scenarios,quantifyingnon- involves identifyingreal examplesof work currently beingimplemented for planningpurposes.Partofthe data, bothforreimbursement and technology tocollectvaluable the purposeofusingsmart not anoptionasitwillundermine Using aflashpassindefinitelyis ITSO smart cards in Welsh public transport 54

Although at the moment an invalid The flexibility of the system will also concessionary card’s shell would be have been demonstrated by the hot listed if it was lost or stolen, in introduction of a commercial smart the future it may be that a product card scheme by Cardiff Bus, one of sits within an ITSO shell containing Wales’ major bus operators. The multiple products. If one product expansion into a commercial scheme became invalid - for example, a will go further towards highlighting cancelled prepaid product - no other the key benefits of the replacement of product would be affected and there paper ticketing by smart technology, would be little point in incurring the such as the invaluable ability to additional cost and time involved in capture data on passenger travel cancelling and reissuing a full card habits to better inform marketing if only one aspect of it is affected. practices and the potentially vast Products could also then be prioritised reduction in fraud. Wales remains on the hotlist based on their value: committed to ITSO, interoperability, a season ticket has inherently more and the implementation of further value than a prepaid day ticket. ticketing types within the smart We plan to use one hotlist that card environment which will is broadcast to our OID group on extend across other modes of a weekly basis with incremental public transport in the future. updates of new or prioritised shells or products, avoiding the additional messaging traffic that would incur from a point to point approach (HOPS to individual POST). There are also certain constraints within this simple implementation: the capacity of memory on ticket machines dictates the maximum size of the list; the size of warm-list within the HOPS needs to be managed and pruned as the INTELLIGENT TRANSPORT SYSTEM scheme grows; software changes will require implementation in the back office as well as changes to the ITSO specification itself; and interacting with other ITSO scheme hotlists in the future will become more complex as more operators become live and implement their own hotlisting policies. In Wales we are planning to implement our initial hotlisting structure by the end of 2009 and are currently likely to be the first ITSO scheme to do this. As the Welsh scheme progresses towards a fully-smart environment, further benefits of the technology will also have been highlighted. Before the end of the year we will have also migrated the scheme to a new card media type where all new passes will be issued using a different specification of card. The ITSO platform has a selection of certified card media available for use, and Wales will shortly be using two card types in the scheme, retiring one card type to allow for better technology to take its place through another. S

43 INTELLIGENT TRANSPORT SYSTEMS 54 ITSO smartcards inWelsh publictransport 44 On balance - the see-saws of congestion 55 charge business cases

Paul Grayston Abstract Technical Director The design of urban congestion charge schemes within business case Intelligent Transport proposals to the UK Department of Transport’s Transport Innovation Systems Fund has presented a variety of competing and sometime conflicting considerations for designers. This paper looks at how balances need to be struck between geographic coverage, revenue, effect on traffic, technology choice, and the treatment of discounts and exemptions.

Andrew Ferry Senior Consultant Intelligent Transport Systems

Adrian Bourne Graduate Engineer Intelligent Transport Systems

Introduction

Overview Many transport practioners see the a referendum when the general priming money for studies to develop use of a congestion charge as a public is unable to comprehend what proposals. The original enthusiasm potential part of their armoury to the whole package of measures, has dampened somewhat. In some change travel patterns and reduce never mind the congestion charge areas politics put paid to the idea, congestion. The success of schemes in scheme! A full, well thought out but in others the clear benefits and/or London and Stockholm has provided business case may help with this in positive business cases through traffic a degree of interest, even support other cities considering road pricing modelling have not been proven. from previously wary politicians. In such as Cambridge and Bristol. the case of London the charge was When announced in 2006, the The basis of this paper part of Mayor Livingstone’s mandate, UK Department of Transport (DfT) In general sponsors of congestion and in Stockholm a referendum 1 Transport Innovation Fund (TIF) charging schemes will want to keep was only held once citizens had offered English Local Authorities the their schemes simple to understand, actually experienced road pricing. opportunity to bid for major funding and aim for the low operational However, as was seen in the “no” for packages of transport investment costs and efficiency seen within the vote Manchester in 2008 (for a that would enable step changes in tolling industry. From Atkins work for scheme not due to go live until at provision (new bus, tram and rail some of the TIF authorities wanting least 2013) political support was not services, better interchange facilities, to develop draft business cases, we enough. This was despite indications improvements in information and have learned that this can be a very that a positive business case for the smartcard ticketing etc). In return difficult balancing act given that there charge could be made. Perhaps the schemes would implement some form are actually few operating schemes biggest lesson from Manchester is of congestion charging. The process in place and comparisons are difficult not for politicians to be pushed into attracted many bids for pump- to make. This paper gives some

45 55 On balance - the see-saws of congestion charge business cases

insight into those “see-saws” and The level of charge (for a local But, reduce the geographic coverage how to account for them within the charge) is driven by a need to of an area scheme and there comes business case for charging schemes. generate a certain level of revenue. a point where it might as well be a The graphical plots used in this As well as covering the operating cordon because there are few if any paper have been generated using costs, this could be to provide funds places where an internal detection cost and volume models developed for paying off loans, paying for point would be worthwhile. by Atkins for road pricing scheme additional works, or even general And then, as scheme concepts go design option analysis and sensitivity contribution to authority funds. To out for consultation, local objections testing. They are used to give a avoid overloading the see-saw with will raise their head. Often they will typical indication of results and do unachievable objectives it is essential have an emotional aspect to them (for not represent any specific scheme. to derive an early indication of: instance giving exemptions to lower (a) the average level of charge paid workers such as nurses) that The see-saws users will be required to pay will be persuasive to local politicians (b) the transaction cost target. to champion. Cost implications can See-saw 1 - What to charge for be severe, even to the extent that See-saw 2 - The scheme versus the political requirement exemptions might result in a net size and boundary cost to the scheme because of the The political instruction to investigate human administration involved. the options for introducing a local Schemes will either be charging road pricing scheme is almost certain for crossing one or more cordons See-saw 3 - The technology choice to contain some caveats. They could (they could be concentric as the For the moment, despite perhaps be about: the level of charge, the Manchester proposals or adjacent justified claims by suppliers, Time revenue expectations, the discounted zones), or for travel within an area. Distance Place is not a viable option or exempted users, geographical In considering the size and shape for an urban type of scheme to coverage and so forth. More often of a scheme boundary, the extent be implemented. Under the UK than not (and quite understandably) of the traffic modelling will set the TIF programme the technology to the scheme designer this initial basic coverage. However, reference choice see-saw for the local ‘back of an envelope’ concept to detailed maps and on the ground authority scheme promoters is will be fraught with complication. surveying are needed to produce a weighted down at one end with Furthermore, at an early stage little set of “gateways” into the scheme. a cautious, risk averse attitude. thought is usually given to how They may be junctions, bridges,

S the charge will be presented to or other physical constructions. It Realistically this means any business end users. Keeping it simple is a is essential that traffic funnelled case must propose the use of byword that is often adhered to, but towards the gateways is the traffic technology already proven on a the consultation processes for the that will be liable for the charge wide scale. This leaves Automatic Number Plate Recognition (ANPR) TEM Manchester and Cambridge schemes and can be charged. Similarly suggest that a large proportion or a combination of ANPR S communities should not be isolated of people didn’t even understand by the boundary of the scheme. and tag and beacon solutions Y that the proposals were for peak as the two main options. The issue of splitting conurbations by S

hour charges only, not all day. a scheme boundary is a vexed one, Tags have attractions in that The only way to demonstrate robustly bound to cause local resentment. they have a high accuracy, as a to decision makers what is likely to detection technology the monetary RT This may result in considerations work and what is not is to carry out for discount schemes which will transaction begins immediately, and O comprehensive traffic modelling. But, complicate the operation of the interoperability can provide users and

P because of its theoretical nature, it scheme. The two ring Manchester operators with incentives for use.

S can sometimes cloud incorporation concept was an interesting But on the other end of the see-saw of the obvious such as natural variant. The fact remains though the permanent scheme in Stockholm boundaries and gates to funnel that in traffic modelling terms has elected to use ANPR as its main traffic in and out of a scheme area, the most attractive options, both solution, with technology improving or that a more simple tolled crossing financially and in terms of affecting the accuracy of the system to a level solution might be more appropriate. traffic are often the simplest. deemed acceptable by the authority. Traffic modelling provides the key If the scheme is a cordon then the ANPR must be considered for the input to any congestion charge cordon should not be too long, purposes of identifying non-payers. scheme design. Identifying the otherwise it might result in too little This creates a further balancing point

ENT TRAN number (and type) of users of effect or be too difficult to divert away relating to the expected number of the scheme per day and relating from. But a long boundary for an area Penalty Charge Notices (PCN) to be G it to a particular level of charge scheme may mean that a relatively issued. In the London scheme PCN is intense and complex work. large number of internal charge income is a considerable contributor points will be required, increasing the to the overall revenue. If there is capital cost for setting up the scheme. a high take up of tags within a Also detection/charge points with an scheme there is potentially less PCN area scheme will need to cover both revenue because of the high level

INTELLI directions of traffic in most locations. of identification of scheme users. 46 On balance - the see-saws of congestion 55 charge business cases

Figure 1 - Example of balancing revenue sensitivity to tag take up

In the draft proposals for the See-saw 4 - The discount The group of users qualifying for Cambridge scheme (October 2007)2, and exemption see-saw a discount can be given stricter it was considered that a tag solution interaction requirements. For example, offered the most efficient scheme Public opinion expressed during in London paying for a weeks worth option. To encourage tag take up, scheme consultation will often result of residents’ discount is one method discounted charges would be offered. in the need to provide discounts, to reduce scheme operation costs. even exemptions to certain types of Different scales of charges and Using expected transaction volumes user. These discounts may be quite degree of take up of the tag option the cost and volume models substantial. The to be discontinued were analysed using the scheme created can be used to carry out Western Extension to the London cost model. Even small differences various analyses on the sensitivity scheme gives residents a 90% in the charges caused quite large of take up of particular solutions

discount for the whole scheme area. INTELLIGENT TRANSPORT SYSTEM (20%) differences in the overall on the overall revenue. Figure 2 Has this level of discount actually annual revenue for the scheme. illustrates an example where a contributed to a rise in traffic The possible scheme illustrated in the residents' discount scheme using within the zone? We might never draft proposals for Cambridge used a tag might be in place as well as know, but the level of (discounted) the output of traffic modelling that a general tag scheme, an ‘Express’ residents’ charge needs to be suggested an average charge of £4 registered user declaration channel sufficiently high to persuade at least in the modelled year would provide where automation of the process some proportion of the residents the benefits in traffic reduction is maximised (e.g. payment by text not to use their cars and continue required. An objective was for more message), and a general declaration to contribute to the congestion. efficient, more accurate operation channel (e.g. payment at a shop through high take up of tags. An In Stockholm the Lidingö exemption terminal) where some form of human analysis of the optimum scale of is managed using tags. Tags are interaction may need to take place. useful for this application, as they charges to give a flat net daily revenue It must be noted that this graph is provide a certainty of registration profile was undertaken, see Figure based on a mix of solutions being with the scheme. However, there 1. This was to ensure the scheme used (so when there are 80% users is a cost for their provision and income would not be dependent on of a solution the other 20% is split operation – in some cases this may a particular split between tag users, between the other three), and that not be recovered if those having a ANPR declaration and PCN revenue, confidence levels are lower below tag do not actually interact with the essentially making sure discounts 10% and above 85% because scheme (for instance they may have a for tag use did not end up as a cost where a different scheme design tag but not interact with the scheme liability. It is important to note that philosophy is likely to be more during its hours of operation). tags increase compliance, therefore appropriate. As would be expected, reduced enforcement revenue from Once discounts or exemptions a greater proportion of declarers PCN’s drops as tag use increases. are introduced into a scheme, the increases the transaction cost, and a Using the cost model also showed dynamic of the cost modelling and greater proportion of tag detection that if there was low take up of the traffic modelling is severely reduces the transaction cost. But, tags (perhaps because the discount tested. Every interaction with the when restrictions are put on users incentive was too low) then there scheme has a cost, even the detection to create the ‘Express declaration’ was a point where, instead of the of a vehicle to be exempted. The option (typically some form of discounted rate, a standard rate and discount has to be commensurate payment guarantee) the slope of an ANPR declaration scheme (i.e. with the overall transaction cost. the line is actually downward.

like the London scheme) became S the most attractive to operate.

47 55 On balance - the see-saws of congestion charge business cases

See-saw 5 - Transaction cost and the number of daily users It would be reasonable to assume that transaction costs decrease with higher numbers of users. A selection of scheme scenarios has been examined using the cost and volume models to see what the effect of variance in conditions produces on the transaction costs. In Figure 4 the intersection of the x and y axis represents a typical scheme example default condition with a transaction cost calculated as Figure 2 – Typical plot of transaction cost against take up of solutions £0.91 with these characteristics: • it uses a set average number of daily users derived from the transport model • 50% of users are expected to be using tag detection. A range of schemes has been tested (including using figures relating to the London scheme) and all have produced plots with similar characteristics. The three plots are independent of each other, but plotting them together shows which of three factors (number of users, proportion of users Figure 3 – Typical plot of revenue against solution assumptions

S using tags and cost of workforce to administer the system) is most sensitive to variation. The workforce effort plot is linear, and reflects the sensitivity of staff salaries on TEM transaction cost. The number of users S in detection plot is reasonably linear, Y some slight wobbles occur reflecting

S the step increases in IT provision as

the number of users increases. The number of users plot, however, is RT a curve that shows low numbers

O of users can result in significantly higher transaction costs resulting in P low revenues and threatening the S Figure 4 - Typical plot of transaction cost against scenario assumptions required profitability of the scheme. For this particular scheme the plot The effect of this scheme option on Tailoring the scheme by running also demonstrates (see the dotted the annual overall revenue raised is different tariff and solution mixes line central box) that variations of shown in Figure 3. Applying similar through the models can help to ± 10% on the design case only techniques as used in the example balance the discount and exemption result in transaction cost sensitivity in Figure 1 could minimise revenue see-saw. But it must be noted that of less than ± 5%. At business case variation for general users, but, such work has a theoretical basis and derivation stage this is an important as the plot shows (as would be is designed to support options with indication of how risk is being ENT TRAN expected) increasing the numbers a business case for road pricing. If in limited in the scheme design. using the residents’ discount solution future the scheme is implemented it G causes a distinct reduction which will be the operational model of the is a function of the slope of the system supplier that drives the plot, residents' discount plot in Figure 2. and the contractual arrangement should apportion the risk accordingly. INTELLI 48 On balance - the see-saws of congestion 55 charge business cases

Conclusion Designing an urban congestion charge scheme is a matter of balance between practicalities and politics. What seems straightforward at first impression often turns out not to be. In developing a business case for a scheme a wide range of tests and scenarios needs to be analysed in order to progress to procurement of a system that will meet the economic and operation targets required of it. INTELLIGENT TRANSPORT SYSTEM S

49 55 On balance - the see-saws of congestion charge business cases

References 1. UK DfT Transport Innovation Fund Guidance (2006/7) http://www.dft.gov.uk/pgr/regional/tif/ 2. Cambridgeshire CC Outline Proposal for Funding (October 2007) http://www.cambridgeshire.gov.uk/transport/strategies/tacklingcongestion/ourproposals/bid.htm S TEM S Y S

RT O P S ENT TRAN G INTELLI 50 Improvements to ramp metering system in 56 England: VISSIM modelling of improvements

Jill Hayden Abstract Managing Consultant The Highways Agency has installed ramp metering at over 80 sites in England, 1 Intelligent Transport with evaluation results showing an average 13% reduction in journey times . Systems Despite the success of the existing system, some potential improvements have been identified. Due to the complexity of the algorithms, the improvements were modelled using the micro-simulation package VISSIM2. This paper describes the method and results of the modelling. The modelling exercise has shown that all the algorithms have worked as Roger expected. It has also shown that all the algorithms designed could provide additional performance benefits over the existing system. Higginson Senior Systems Engineer Intelligent Transport Systems

Matthew Hall Managing Consultant Intelligent Transport Systems

Background INTELLIGENT TRANSPORT SYSTEM Sukhvinder Ramp metering was first trialled in These improvements are described in England in 1986, with a small number detail in the paper “Improvements to Ubhi of sites around Birmingham and then Ramp Metering System in England: Project Sponsor Southampton. From the outcome Detailed Description of Algorithm Highways Agency of these trials the Highways Agency Development”3. As these are new (HA) commissioned Atkins to design ideas and some of the solutions a system for wider roll-out, and IPL are quite complex, it was decided & PEEK to implement it. This system that modelling should be used to was first installed in 2005 and there check their feasibility and that they are now over 80 sites in England. The would work as expected, before existing system was not modelled producing the final specifications. prior to implementation and no sites Although there are a range of were modelled before being installed, micro-simulation modelling packages for two main reasons. Firstly the basic available such as PARAMICS, AIMSUN concepts of the algorithms (ALINEA and DRACULA, the VISSIM software on the main carriageway, queue from Visual Solutions was chosen for management and queue override on this assessment due to its strength in the slip road) were well understood junction modelling, both signalled and and had been successfully used un-signalled. VISSIM is a microscopic elsewhere, giving confidence that traffic flow simulation model based the system would work. Secondly, on car following and lane change the timescales did not allow it. logic. Furthermore, it includes the Since the system was first Vehicle Actuated Programming (VAP) installed, several improvements integrated package that enables have been identified with the complex algorithms to be modelled. following potential benefits: • A reduction in the amount of calibration required • Improvements to the operational benefits

• Widening of the applicability S of technique 51 INTELLIGENT TRANSPORT SYSTEMS 56 main carriagewayandsliproad. of 4.2%wasachievedforboth timereductionthat anoveralljourney in themodel.Themodelalsoshowed a similarorder; thisgaveconfidence showed a7.4%decrease whichisof main carriageway. Themodelresults reduction timeonthe injourney southbound, showeda9.1% system implementedatM1J33 The evaluationoftherampmetering the purposesofproject. that themodelwassuitablefor model. Theseresults gaveconfidence MIDAS dataovertwodays,andthe ramp meteringconditions,forreal upstream occupanciesforexisting Figure 1showsthecomparison of modelling results. Forexample, between actualon-road and showed excellentcomparisons metering conditions.Thevalidation for bothpre andexistingramp The existingmodelwasre-calibrated, of existingrampmetering Calibration andvalidation of eachimprovement. be drawnaboutthesuccess This enabledconclusionsto • • • • measures were usedsuchas: without eachimprovement, various ramp meteringsystem,withand compare theperformance of the VAP codeandthencalibrated.To were eachimplementedinthe be compared. Theimprovements which theimprovements could ramp meteringmodel,against project. Thisprovided thebase calibrated forthepurposeofthis The modelwasimproved andre- • • Ageometricallycorrect road • southbound, whichcomprised: metering systematM1Junction33 VISSIM modelofthecurrent ramp The project madeuseofanexisting Methodology the real systemontheroad accurately locatedcompared to layout ofthejunction Speed profile RM controller outputs Travel/delay times Network performance represent thesystem VAP programming to algorithms, codedusing The rampmeteringsystem MIDAS loopsandtraffic signals nln:Dtie ecito red timeisreduced, andviceversa. of AlgorithmDevelopment”. England: DetailedDescription release thetraffic level.Iftheactualflow signaltimingsforeach to RampMeteringSystemin in thepaper“Improvements detailed descriptionsare wastoreduce available thesignificanttime andeffort required tocalibrate are onlybrieflyexplainedhere, Themainpurposeofthisalgorithm algorithms. Althoughthealgorithms results ofmodellingthefourkey The followingsectionsdescribethe improvements Modelling of Figure 1-Comparisonofupstream occupanciesformodel andreal data Figure 2-ComparisonofspeedswithexistingRMandauto-sig Table timesforauto-signaltimings 1-Comparisonoftotalvehiclejourney Auto-Sig Base RM Scenario

Motorway Speed (mph) Pre-RM Occ [%] 10 15 20 25 30 35 40 10 20 30 40 50 60 70 80 0 5 15:45 15:45 0

900 England: VISSIMmodellingofimprovements 1200 16:00 16:15 16:30 16:45 17:00 17:15 17:30 17:45 18:00 18:15 18:30 18:45 19:00 19:00 18:45 18:30 18:15 18:00 17:45 17:30 17:15 17:00 16:45 16:30 16:15 16:00

1500 Improvements torampmeteringsystemin 1800 2100 Veh Hrs 1008.6 934.2 2400 900.8 2700 52

3000 Mainline 3300 3600 3900 Upstream Occupancyby Upstream Comparison Day -7.4%

4200 %Diff -10% Speed Profile - Average- Profile Speed Seeds Random 10 of 4500 4800 - 5100 5400 5700 Simulation Second 6000 Veh Hrs 112.6 105.6

6300 84.5

6600 Time

6900 Slip Road timings (Auto-Sig) Auto-calibration ofsignal improved, asshowninTable 1. through thejunctionissignificantly times performance intermsofjourney The modellingshowedthatoverall than therequired flow, thenthe achieved byarelease level islower 7200 7500 7800 33.3% 24.9%

8100 %Diff 8400 - 8700 9000 9300 9600

9900 Veh Hrs 1046.8 1093.1 1006.3 10200 10500 Base RM Base 10800

11100 Total 11400 11700 Model 31/01/08 25/01/08 HAQ024 HAQ024 -4.2% -7.9%

12000 %Diff - Improvements to ramp metering system in 56 England: VISSIM modelling of improvements

It can be seen that both the main Auto-calibration of carriageway and slip road journey times are reduced, compared to the Queue Management existing ramp metering scenario. Parameters (Auto-QM) This is because having accurate This algorithm dynamically updates The figure shows that the Auto-QM release flows allows more efficient the values of two key parameters scenario is slower to break down operation of both the ALINEA and in the QM algorithm; the desired than base ramp metering, makes a queue management algorithms. The combined queue occupancy o and quicker minor recovery but is then modelled overall journey times are descq the queue management gain factor slower to recover in the latter periods. improved by 7.9% from the pre- K . This is designed to remove This algorithm has an opposite effect ramp metering scenario, compared poqm the significant effort required to to the Auto-sig algorithm; the main with a modelled improvement of calibrate the proportional occupancy carriageway journey time is slightly 4.2% with existing ramp metering. queue management algorithm. worse than with base ramp metering This is an excellent result because as but the slip road is significantly better well as fulfilling the main objective Figure 4 shows the comparison of the (because the queue is managed more of successfully reducing calibration speed throughout the peak, for the efficiently). This results in a slight time, the algorithm should also existing ramp metering (in red), and benefit overall, as shown in Table 2. significantly reduce journey times. the auto-QM scenario (in green).

Figure 2 shows the comparison of Red Time by Release Level for 10 Iterations the speed throughout the peak, 20.0 for the existing ramp metering (in 19.0 RL1 18.0 red), and the auto-signal timings 17.0 RL2 16.0 scenario (in green). It can be seen RL3 15.0 that speeds are higher with auto- 14.0 RL4 signal timings for the whole of the 13.0 12.0 RL5 peak, which explains the significant 11.0 improvement in journey times. 10.0 RL6 Time in Seconds 9.0 RL7 The modelling showed that even 8.0 7.0 RL8 when the red times for all release 6.0 levels are initially set to the same 5.0 RL9

4.0 INTELLIGENT TRANSPORT SYSTEM value (15s), the algorithm calibrates 3.0 RL10 2.0 the signal timings fairly successfully 1 2 3 4 5 6 7 8 91 0 by the end of one day of operation Iteration (iteration 1 in Figure 3). The system Figure 3 - Variation in red times per release level for fi rst 10 days operation improves further to converge to sensible signal timings within a week Speed Profile - Average of 10 Random Seeds of being installed (iteration 7). 80

70

60

50

40

30 Motorway Speed (mph)

20

10 Base RM HAQ020

0 900 1200 1500 1800 2100 2400 2700 3000 3300 3600 3900 4200 4500 4800 5100 5400 5700 6000 6300 6600 6900 7200 7500 7800 8100 8400 8700 9000 9300 9600 9900 10200 10500 10800 11100 11400 11700 12000 Simulation Second

Figure 4 - Comparison of speeds with existing RM and with Auto-QM

Table 2 - Comparison of total vehicle journey times for Auto-QM

Scenario Mainline Slip Road Total Veh Hrs %Diff Veh Hrs %Diff Veh Hrs %Diff Pre-RM 1008.6 - 84.5 - 1093.1 - Base RM 934.2 -7.4% 112.6 33.3% 1046.8 -4.2% Auto-Sig 938.4 -7.0% 103.6 22.6% 1041.9 -4.7% S

53 INTELLIGENT TRANSPORT SYSTEMS 56 Control wouldalsobebeneficial. and onthoseoccasionsMerge cause asaturatedmergearea problem, incidentsmayoccasionally junctions withnoregular merge saturated bythesliproad flow. At where themergearea isregularly at arelatively smallnumber ofsites These benefitswillonlybegained overall benefitsinthemodelling. This algorithmshowedthelargest as farandisquickertorecover. Control modeldoesnotbreak down The figure showsthatthe Merge the MergeControl scenario (ingreen). existing rampmetering(inred), and speeds throughout thepeak, forthe Figure 5showsthecomparison of significant benefit(seeTable 3). ramp metering,withanoverallvery road are lowerthanforexisting main carriagewayandontheslip timesonboththe The journey limits theflowfrom thesignals. used bythetraffic signals,i.e.it algorithm limitstherelease levels this occurs,themergecontrol merge area iscongested.When this model)todeterminewhenthe downstream ofthestoplinefor the nosearea (approximately 200m downstream ofthestop-linewithin sites. Anewsetofloopsisplaced override algorithmsatcertain by queuemanagementand caused bythehighflowsrequested of themergearea whichcanbe attempts toprevent saturation The MergeControl algorithm Merge Control Table timesforMerge Control 3-Comparisonoftotalvehiclejourney to provide performancebenefits. the DCequation).Thiswasexpected output oftheALINEAequationinto by cascadingthem(puttingthe Demand Capacity(DC)algorithms, best characteristicsoftheALINEAand The ACDCalgorithmcombinesthe Demand-Capacity (ACDC) AINEA Cascadedwith Figure 6-ComparisonofspeedswithexistingRMandACDC Figure 5-Comparisonofspeeds withexistingRMandMergeControl Auto-Sig Base RM Scenario Motorway Speed (mph) Motorway Speed (mph) Pre-RM 10 20 30 40 50 60 70 80 10 20 30 40 50 60 70 80 0 0

900 900 England: VISSIMmodellingofimprovements 1200 1200

1500 1500 Improvements torampmeteringsystemin 1800 1800 Veh Hrs

2100 2100 1008.6 934.2 888.1 2400 2400 2700 2700 54

3000 3000 Mainline 3300 3300 3600 3600 3900 3900 -11.9% -7.4% 4200 4200 %Diff Speed Profile - Average- Profile Speed Seeds Random 10 of Speed Profile - Average- Profile Speed Seeds Random 10 of 4500 4500 - 4800 4800 5100 5100 5400 5400 5700 5700 Simulation Second Simulation Second 6000 6000 Veh Hrs 112.6 102.5

6300 6300 84.5 6600 6600

6900 6900 Slip Road overall slightbenefit(seeTable 4). lower thanforexisting,withan on thesliproad are marginally carriageway are slightlylowerand timesonthemain The journey break downfurtherthanthebase. quicker torecover, althoughitdoes slower tobreak downandslightly This showsthattheACDCmodelis and theACDCscenario(ingreen). the existingrampmetering(inred), speeds throughout thepeak,for Figure 6showsthecomparisonof 7200 7200 7500 7500 7800 7800 33.3% 21.3%

8100 8100 %Diff 8400 8400 8700 8700 - 9000 9000

9900 Veh Hrs 1046.8 1093.1

9300 990.6 102009300 9600 9600

10500 RM Base Base RM Base 9900 11100 10200 Total 11400 10500

1080011700 10800 HAQ027 HAQ027 HAQ035 HAQ035 -4.2% -9.4% 11100 %Diff 11400 - 11700

12000 12000 Improvements to ramp metering system in 56 England: VISSIM modelling of improvements

Table 4 - Comparison of total vehicle journey times for ACDC

Scenario Mainline Slip Road Total Veh Hrs %Diff Veh Hrs %Diff Veh Hrs %Diff Pre-RM 1008.6 - 84.5 - 1093.1 - Base RM 934.2 -7.4% 112.6 33.3% 1046.8 -4.2% Auto-Sig 922.0 -8.6% 112.2 32.7% 1034.2 -5.4%

Table 5 - Predicted journey time savings with four improvements

Algorithm Improvement JT Improvement in Number of sites model - above 4.2% with existing RM Auto-Sig 3.7% 80 Auto-QM 0.5% 80 Merge Control 5.2% 12 ACDC 1.2% 80 Conclusions The VISSIM modelling has shown that all algorithms developed work as expected in the model, which provides confidence that they could be implemented successfully. It has also been shown that all four algorithms provided additional performance benefits over and above INTELLIGENT TRANSPORT SYSTEM the existing system. Table 5 shows the predicted journey time savings for each algorithm. While the results of the modelling do not guarantee the same results on the road, they do give increased confidence that greater journey time benefits can be achieved. Acknowledgements This paper draws on the work undertaken for Highways Agency Network Services Directorate and is published with the permission of the Highways Agency. The views contained in this paper are those of the authors and not necessarily those of the Highways Agency. S

55 56 Improvements to ramp metering system in England: VISSIM modelling of improvements

References 1. “Ramp Metering Operational Assessment”, Highways Agency Report, April 2008, http://www.highways.gov.uk/knowledge/17375.aspx 2. VISSIM Visual Solutions software, www.vissim.com 3. Higginson R, Hayden Dr J, Charton T, Ubhi S, “Improvements to Ramp Metering System in England: Detailed Description Of Algorithm Development”, in Proceedings of ITS World Congress 2009 (Stockholm, 21-25 September 2009). S TEM S Y S

RT O P S ENT TRAN G INTELLI 56 ‘Scrambled’ pedestrian crossings at 57 signal controlled junctions - A case study

Chris Abstract Greenwood ‘Scramble crossings’ for pedestrians at signal junctions are widely used in Japan and have been reintroduced in Canada and the United States as a Principal Consultant way of prioritising pedestrian movement by stopping all traffic movement Highways & Transportation and allowing pedestrians to cross in every direction at the same time. Some examples exist in the UK but their use is not widespread. One potential reason for this is there is limited guidance on when ‘scramble crossings’ should be considered as potential schemes for promoting pedestrian priority. This paper focuses on the proposed scramble or diagonal crossing scheme at Oxford Circus in central London. This has been designed by Atkins on behalf of The Crown Estate, Transport for London (TfL), Westminster City Council (WCC) and the New West End Company (NWEC) and was implemented during 2009. The purpose of the paper is to identify some of the existing examples in the UK and overseas and review the existing guidance on their application of scrambled crossings. The paper then describes the design process and key features of the Oxford Circus scheme and based on this experience concludes by suggesting potential future applications of scramble crossings in the UK.

Introduction to scramble crossings For the purpose of this paper the term The author considers that the Through review of these papers ‘diagonal crossing’ will be used as popularisation of the motor car and the author’s experience well as ‘scramble crossing’ to describe in the post-war period resulted of developing the proposed signalised crossings which have an ‘all in the withdrawal of some of the scheme for Oxford Circus the key red’ stage and where pedestrians are scramble crossings particularly advantages of introducing scramble encouraged to cross in all directions. in North America. However, the crossings include the following: Scramble crossings for pedestrians acknowledgement of the need to • Promotion of pedestrian priority involve stopping all traffic movements promote more sustainable modes and the relief of pedestrian at signalised junctions and allowing of travel together with increased congestion on more traditional pedestrians to cross in every direction urbanisation has lead to scramble orthogonal crossings and at the same time. Their origins are crossings being considered once footways particularly where unclear but they are also known again as a way of dealing with the pedestrian volumes are very high. conflicts between pedestrians and as a ‘Barnes Dance’ after Henry • Reduction of walk distances vehicles in the centre of our cities. Barnes, a traffic engineer in the and times particularly where United States, who popularised New scramble crossings have recently pedestrians would otherwise the concept after overseeing their been introduced in Oakland (2002 use two orthogonal crossings to introduction in Denver and New York. at 8th Street / Webster Street) and reach their intended destination The general consensus is that the first Calgary (2003 in Quartier International and can now complete their scramble crossings were introduced district) and Toronto (2008 at journey through the junction in Vancouver and Kansas City in the Yonge Street / Dundas Square). by making a single diagonal 1940’s and spread to other cities in crossing movement. Potential advantages HIGHWAY North America including Los Angeles • Potential improvements in safety and disadvantages of in the 1950’s which at one point had by reducing conflicts between 25 examples. Over 300 scramble scramble crossings pedestrians and vehicles. Bechtel crossings now exist in Japan including A number of papers have been et al 2003 demonstrated in their the ‘ crossing’ situated close to published on the theoretical and analysis of the Oakland scramble ’s Shibuya railway station which actual benefits and disbenefits of crossing that safety benefits is said to be used by over 250,000 introducing scramble crossings were realised by the introduction pedestrians every day. This is probably including papers by; Ahuja et of a scrambled crossing. The the most widely known example of al 20081, Ishaque and Noland North American examples of a scramble crossing in the world. 20062 and Bechtel et al 20033. scramble crossings have particular S

57 57 ‘Scrambled’ pedestrian crossings at signal controlled junctions – a case study

benefits of reducing the conflict crossings at signal junctions are little guidance is given to allow the between pedestrians crossing usually of the orthogonal / straight designer to determine whether a and vehicles turning left and across or staggered type: these diagonal crossing layout should be right ‘on red’ which is legally provide crossings of the traffic used as opposed to a more traditional permitted in the United States. approaches to the junction. arrangement with orthogonal But this is not permitted when In the UK the Department for crossings. This along with the a scramble crossing operates. Transport (DfT) sets out the perceived risk of not following Ahuja et al 2008 concluded that design standards and guidelines guidance may be a key factor as to scramble crossings are beneficial for pedestrian facilities at signal why diagonal crossings are still the only when both pedestrian and controlled junctions. Design Manual exception to the norm in the UK. vehicular volumes are relatively high for Roads and Bridges (DMRB) Current UK examples of scramble and the junction is characterised by TD50/044 states that at traffic signal or diagonal crossing layouts significant delay for both modes of junctions ‘where a pedestrian need at signal junctions include: travel. In free flowing conditions is established then appropriate • Aberdeen – Scrambled for vehicles and pedestrians, they signal controlled facilities should be crossings at several locations concluded a conventional crossing provided’. Traffic Advisory Leaflet across the town centre. arrangement would be a better 5/055 describes the main options for • Balham High Road / Balham solution as scrambled crossings providing pedestrian facilities at traffic Station Road junction, Balham, generally lead to an increase in delays signal junctions. This states that: South London – urban signal at the junction. This may in turn "the designer has to consider the junction located directly outside lead to an increase in pedestrians pedestrian flow patterns, degree of Balham London Underground not complying with the signals and saturation and the topographical Station with two diagonal therefore an increase in conflict layout to decide on which option crossings as shown in Figure between pedestrians and vehicles. is best suited to a particular site". 1. This is the first example of a The potential drawbacks of The options identified include diagonal crossing arrangement introducing scramble crossings for underpasses and overbridges; introduced in London. pedestrians including the following: no pedestrian phase or stage; • A240 / A2022 Drift Bridge, • Increased delays to vehicles full pedestrian stage; parallel Epsom Downs, Surrey – suburban particularly where an ‘all red’ pedestrian facility; staggered signal junction providing access signal stage has to be introduced. pedestrian facility or displaced to a local centre with single • The effectiveness of scramble pedestrian facility. With respect to diagonal crossing from north- crossings relies on pedestrian providing a full pedestrian stage, east to south-west corner only. the advice from the DfT in terms of compliance with the signals and • Balliol Road / Pembroke Road, diagonal crossings is as follows: several examples are considered Bootle, Merseyside – urban to suffer from pedestrians “Diagonal crossings (crossing the signal junction with single failing to ‘clear’ the junction centre of the junction, say, from north diagonal crossing connecting at the end of the crossing east to southwest) are largely untried two separate sites of Hugh periods leading to potential but a small number do exist. There Baird Higher Education College additional delays to vehicles. are important design aspects to be (introduced 2004-2005). incorporated. Diagonal crossings are • Increase in the perceived risk to • Oxford Road / Dover Street, not considered appropriate for many pedestrians by the removal of Manchester – urban signalised disabled users, particularly those who barriers and guardrails to allow staggered crossroads are visually impaired. Also road safety crossing in all directions. The in University area. key consideration is the potential education generally teaches children The original justification for providing risk of pedestrians waiting to not to cross diagonally at junctions. diagonal and scramble crossings cross on diagonal movements Conventionally orthogonal crossing at these locations is not known. at the corners of junctions places should therefore always be However, justification appears to be being struck by vehicles turning provided with flush dropped kerbs, either a result of the existing layout left around the corners. tactile paving and audible / tactile signals. Flush dropped kerbs, tactile (such as a staggered crossroads) which Status and examples paving and audible and tactile signals lends itself to a diagonal crossing should NOT be provided on the and / or because they are in local in the UK diagonal crossing part. If a lowered and town centres, where pedestrian S connectivity is important and The basic prerequisite for introducing kerb is provided there should be a pedestrian flows are relatively high. scramble crossings exists at a large minimum upstand (after possible number of signal junctions in the resurfacing) of at least 25mm. Careful There are also a number of existing UK through the provision of an thought also needs to be given to the junctions where traditional orthogonal ‘all red’ traffic stage during which use of markings or coloured surfacing crossings are provided at signalised at the junctions so that partially crossroads and where the absence

HWAY all traffic movements are stopped and pedestrians use all marked sighted pedestrians are not misled". of barriers to movement such as G crossings at the same time. However, Although the advice is therefore guardrail permit ‘informal’ diagonal in the UK controlled pedestrian prescriptive in terms of the design, crossing. This includes the A24 HI 58 ‘Scrambled’ pedestrian crossings at 57 signal controlled junctions – a case study

one continuous movement – this is facilitated by the provision of an ‘all red’ traffic stage during which every pedestrian crossing operates. Each of the four crossings are set back between 7 and 15 metres from the junction which means they do not serve the key pedestrian desire lines along the footways on either side of the two streets. Stairways to Oxford Circus London Underground station are located at each of the four corners of the Circus and stone balustrades mark the kerbs around each of the corners preventing crossing by pedestrians anywhere other than at the formal crossings. The layout of the London Underground stairways and the balustrades means that pedestrians cannot walk between Figure 1 – Diagonal crossing in Balham, South London them creating areas of ‘dead space’. Guardrails are used in the vicinity of the orthogonal crossings on the central islands and further down each approach to the junction. Both Regent Street approaches have two lanes of traffic, as does the approach from Oxford Street west which has one lane dedicated to right turning buses. There is a single lane on the Oxford Street east approach. A number of turning movements are banned including the right turns from both Regent Street approaches and Oxford Street east as well as the left turn from Regent Street north. The Oxford Street approaches are generally perceived to be for the use of buses and taxis only during 0700-1900 hours but Figure 2 - Informal diagonal crossing in Epsom town centre in actual fact service vehicles are High Street / Waterloo Road / Ashley presents unique problems for its permitted to make certain turns to Road junction in Epsom town infrastructure and public realm and from Oxford Street. Separate centre close to Atkins’ offices and including congested streets, vehicular traffic stages are provided for Regent shown in Figure 2. This is situated and pedestrian conflicts and a Street and Oxford Street with an in an outer London suburban town tired looking street environment. extended phase for the right turning centre but also on the A24 London buses from Oxford Street west. to Worthing Trunk Road. Such Existing junction layout (2008) Current use and examples represent potential sites Oxford Circus is at the heart of where formal diagonal or scramble operating conditions the West End and one of the most crossings could be introduced. renowned junctions in the world, Oxford Circus is dominated by pedestrians and is extremely busy

marking the convergence of London’s HIGHWAY The Oxford Circus two most famous retail streets, at most times of day, particularly case study Oxford Street and Regent Street. at weekends. Typically, pedestrians account for 64 percent of all people The junction is a signalised crossroads London’s West End has been one passing through the junction. Bus with traditional orthogonal crossings of the UK’s dominant retail and passengers account for the next for pedestrians on all four arms, entertainment centres since the highest proportion, at 32 percent, each of which include central early 19th century. It continues to with around 1 percent of people islands as shown in Figures 3 and be a vibrant place with a distinctive passing though in private cars, taxis, 4. The crossings are ‘straight across’ retail offer and range of attractions on motorcycles and by cycle. with pedestrians crossing either

and is renowned on an international S level. However, its popularity Regent Street or Oxford Street in

59 57 ‘Scrambled’ pedestrian crossings at signal controlled junctions – a case study S HWAY

G Figure 3 – Plan of existing Oxford Circus layout (before scheme) HI 60 ‘Scrambled’ pedestrian crossings at 57 signal controlled junctions – a case study

The four pedestrian crossings are also extremely busy and do not always clear of pedestrians at the end of the crossing stage. Many of the existing pedestrians use two crossings as part of their journey through the junctions and would therefore be well served by a diagonal crossing. A number of vehicle turning movements are banned which results in 64 percent of pedestrians crossing during the ‘red man’ phase when either Regent Street or Oxford Street traffic movements are on ‘green’. Conflicts between pedestrians are also an issue, for example between those accessing the London Underground and others passing through the junction using the footways and crossings. Faced with these conditions a small number of pedestrians opt to walk in the road on the ‘traffic side’ of the balustrades

Figure 4 – Oxford Circus layout (before scheme) – view towards north east corner (Nike Town) and guardrail as shown in Figure 7. Both streets are major traffic thoroughfares - typically, hourly traffic flows peak at approximately 2,000 vehicles per hour during the week with around 500 to 700 vehicles per hour on Regent Street in each direction and 200 to 400 vehicles per hour on Oxford Street in each direction. The junction is generally considered to be operating at or close to capacity for a large part of the day with queues evident on all approaches but particularly Oxford Street west. The junction is a key ‘node’ in the network of traffic signals in this part of the West End. Buses account for a significant proportion of the existing traffic at Oxford Circus: a total of 24 bus routes pass through the Circus with over 400 buses per hour on a typical weekday. Figure 5 – Surface mode share at Oxford Circus: Weekday PM Peak Hour Several of the routes terminate in the (Source – 2006 traffi c and pedestrian counts) area and pass through the junction more than once in accessing and Oxford Circus has the highest stationary pedestrians means high egressing from their stands. Oxford pedestrian volumes recorded levels of pedestrian congestion are Circus is highly accessible by bus and anywhere in London. At the busiest a normal part of the Oxford Circus is also a key interchange between times, over 40,000 pedestrians per experience for large parts of the day. the various bus services and between hour pass through the junction which 6 HIGHWAY Fruin Levels of Service (LoS) D to F buses and London Underground. Bus includes those accessing the London have been calculated on all of the stops are located within 200 metres of Underground station. The existing footways in the vicinity of Oxford the junction on all four approaches. crossings are each used by between Circus including those between the 5,500 and 11,000 pedestrians per London Underground station accesses Scheme development hour with the crossing of Regent and building lines as shown in Figure Street south generally the busiest. 6. These LoS represent extremely Westminster City Council, The A combination of the high volumes congested conditions for pedestrians Crown Estate, Transport for London of pedestrians, together with the which dissuade walking at and and the New West End Company, constrained nature of the space and through the Circus and ultimately together with other key stakeholders, S the prevalence of street clutter and dissuade visitors to the West End. developed and adopted the ‘Oxford,

61 57 ‘Scrambled’ pedestrian crossings at signal controlled junctions – a case study

Figure 6 – Fruin level of service analysis of existing layout Regent and Bond Street (ORB) Action Plan, a vision for the future development of the West End to safeguard its position as one of Europe’s finest shopping districts. The action plan puts forward a number of proposals to improve the quality of the public realm and to enhance the visitor experience. As part of the ORB Action Plan, Atkins (working for The Crown Estate) proposed a comprehensive redesign of Oxford Circus to provide more space for pedestrians through reduced street clutter, selective footway widening and diagonal (scramble) crossings. The overarching objective was to transform Oxford Circus into a rejuvenated space befitting its status as gateway to the West End and address its problems in terms of discouraging visitors, shoppers and tourists. Figure 7 – Pedestrians walking in the carriageway A plan and image of the proposed

S scheme are presented in Figures 8 and 9 - the key elements are as follows: • Two new diagonal crossings for pedestrians. • Straight across crossings realigned to better serve the HWAY pedestrian desire lines with

G narrowed central islands. HI 62 57 HIGHWAYS 63 ‘Scrambled’ pedestrian crossings at crossings pedestrian ‘Scrambled’ case study – a junctions controlled signal Figure 8 – Plan of Oxford Circus scheme Circus 8 – Plan of Oxford Figure 57 ‘Scrambled’ pedestrian crossings at signal controlled junctions – a case study

therefore sought between providing access to all crossings for all users and the desire to not encourage use of the diagonals by the partially sighted who could become confused by the arrangement which is not yet standard practise in the UK. A pedestrian ‘countdown’ system was also considered for implementation in parallel with the scheme that would visibly and audibly count down the time in seconds until the end of the pedestrian crossing stage. As this technology is also untried in the UK and requires approval from the DfT it was decided to omit countdown from the scheme for the time being although it will be considered for implementation at a later date. Guidance and approval was sought from Transport for London’s

Figure 9 – image of Oxford Circus scheme – view towards north east corner (Nike Town) Directorate of Traffic Operations (DTO) who manage the signal • Footway widening on Regent The new diagonal crossings are junctions within the Greater London Street both north and south approximately 4 metres wide and Authority area on the appropriate of the junction which together between 25 and 27 metres in length. intergreen and pedestrian crossing with the removal of street clutter This compares with the realigned timings. These have been derived to create 63 percent more straight across crossings which are 4 with reference and in accordance usable space for pedestrians. to 6 metres wide and between 11 and with the appropriate guidance • Removal of balustrades and 15 metres in length. In common with (TA 16/817 and TTS68). the realigned straight across crossings, guardrail to facilitate diagonal It was decided to reduce the and scramble crossing. the diagonal crossings have dedicated signal heads and studs marking the pedestrian invitation from the • Introduction of or improvement width of the crossing. It should standard 9 seconds to 8 seconds. to bus lanes on the Regent be noted, however, that as with This decision was taken in order to Street approaches. other ‘scramble’ crossings already in minimise the impact of the scheme • Closure of Princes Street existence there is an expectation that on traffic conditions and therefore and Little Argyll Street, two pedestrians will also cross outside bus journey times. A comparison side streets south of the of the stud markings making full of the existing and proposed signal Oxford Circus junction. use of the space in the centre of the settings including traffic stage lengths, inter greens and pedestrian The new diagonal crossings are the junction between opposing stop stage is presented in Table 1 (for centrepiece of the scheme and were lines and their parallel stud marks. the two signal cycle times currently identified as potential solutions to Extensive public and stakeholder used in the weekday Inter Peak Hour the existing pedestrian congestion on consultation was carried out on the (1300-1400 hours) and the cycle footways and crossings, particularly in design philosophy to be applied to time used in the weekday PM Peak the light of the fact that pedestrians the diagonal crossings. The public Hour (1700-1800 hours). The table outnumber those travelling in vehicles consultation gave feedback which indicates that the overall length of the twenty to one and that a large was very positive with 95 percent of pedestrian crossing stage (inclusive number of the existing pedestrians respondents expressing their support. of subsequent intergreen) has been use more than one crossing to Consultation with accessibility groups increased from 26 to 32 seconds. complete their journey through the such as Guide Dogs for the Blind This increase is associated with the junction. The perception also was helped to influence the design and increased length of the longest single that as an ‘all red’ traffic stage already as a result of this consultation it was crossing to approximately 26 metres existed, that diagonal crossings could S decided that different coloured skid (the diagonal from south east to be implemented without a significant resistant surfacing would be used to north west) with the scheme in place impact on traffic conditions since mark out the diagonal and straight compared to the existing longest the broad order distribution of signal across crossings. It was also decided crossing (Regent Street South) which ‘green’ time between vehicles and that whilst tactile paving and dropped is approximately 17 metres in length. pedestrians and the overall cycle time kerbs would be used for the straight The new diagonal crossing therefore could be maintained. Minimising HWAY across crossings, none would be used requires a pedestrian crossing stage the traffic impact and the effect for the diagonal crossings which which is 6 seconds longer than G on buses in particular was also a would feature a 50mm upstand the existing stage inclusive of the key objective of the scheme. at the kerb. A compromise was

HI subsequent intergreen period. 64 ‘Scrambled’ pedestrian crossings at 57 signal controlled junctions – a case study

Table 1 - Proposed signal settings junction operating conditions and the impact on particular vehicle classes. This encompassed the prediction of before and after journey times for buses and other vehicles, predicted queues and predicted Degrees of Saturation for traffic. The VISSIM model included the full length of Oxford Street from Marble Arch in the west to St Giles Circus in the east and Regent Street from Langham Place in the north to Great Marlborough Street / Maddox Street in the south. Both models were audited and approved by the relevant departments within TfL. Detailed analysis of the pedestrian movement issues predicts that the scheme will have the following effects: • Average walk times through the junction will be reduced by 52 Scheme appraisal seconds by virtue of reduced pedestrian congestion and This increase is offset by reduced inter A significant investment was the provision of the diagonal greens between the traffic-traffic made in undertaking detailed crossings which will allow many phases and the traffic-pedestrian appraisals of the proposed scheme pedestrians to undertake a single phase achieved by moving the for a number of reasons: crossing movement rather than orthogonal crossings and their • The innovative nature of the use two of the existing straight associated vehicular stop lines closer scheme particularly the diagonal across crossings in succession. into the centre of the junction. crossing and scramble zone. Overall, a saving of 8 seconds has • Fruin LoS on the footways in • The high profile nature of been achieved through these changes the immediate vicinity of Oxford Oxford Circus and its role as which (taking into account the change Circus are predicted to improve the key junction in the West in pedestrian ‘invitation’ and crossing from Level of Service between End and the surrounding time) provides a net gain of 2 seconds D and F to Levels of Service network of traffic signals. per cycle for vehicular green periods. between A and C as shown • The large number of stakeholders in Figure 10. This represents a It should also be noted that the involved in any such scheme significant improvement from the proposed closure of the Princes in central London. exiting congested environment to Street side road is predicted to busy but generally free flowing remove approximately 100 vehicles Although the development of the conditions for pedestrians. per hour from passing through scheme was predominantly funded the Oxford Circus junction as this by The Crown Estate, the costs of • Typically the new diagonal street is eastbound (towards Regent its implementation will be shared crossings will attract use by up Street only) and traffic can only between The Crown Estate, Transport to 6,500 pedestrians per hour currently turn northbound towards for London and Westminster City who will be abstracted from the Oxford Circus. This is a reduction Council. For this reason a standard four straight across crossings in total junction throughput of TfL Business Case was required to improving conditions there. approximately 5 percent. evaluate the financial benefits of • The configuration of the space the scheme. This, in itself, required Whilst the saving in inter green on the footways and around the detailed appraisals of the effects times and the removal of the London Underground stairways of the scheme on the various users Princes Street traffic from Oxford will provide separate ‘wait’ zones that currently pass through the Circus has undoubtedly helped to for pedestrians congregate to

Circus, particularly, pedestrians, HIGHWAY achieve a scheme that minimises use all of the crossings and buses and other vehicles. the effect on operating conditions, ‘through’ zones for walking it should be noted that these Atkins Intelligent Space undertook a pedestrians. Movement either measures are actually offsetting detailed manual analysis of pedestrian side of the London Underground the effect of the footway widening movements including Fruin Level of stairways will also be possible. (and corresponding narrowing of Service (LoS) as well as developing This is illustrated by the images Regent Street northbound and Legion pedestrian micro-simulation from the Legion pedestrian southbound from two lanes to models of the Circus. Atkins Transport model presented in Figure 11. one lane) as well as the effect of Planning developed a parallel VISSIM vehicle micro-simulation model to the longer diagonal crossings. S evaluate the impacts on overall

65 57 ‘Scrambled’ pedestrian crossings at signal controlled junctions – a case study

Figure 10 – Fruin level of service analysis for Oxford Circus scheme Whilst the diagonal crossings are only one element of the scheme, they are key to its success: without them the realigned crossings are predicted to be overcapacity and function in much the same way as the existing straight across crossings. It is notable that the Legion model can only simulate the current conditions if pedestrians are permitted to cross during the red man phase (as actually happens) whilst the scheme can operate effectively assuming 100 percent compliance by pedestrians. A VISSIM micro-simulation model, an image from which is shown in Figure 12, has been used to demonstrate that the scheme could deliver the significant benefits for pedestrians and the public realm without a significant impact on junction Figure 11 – Legion space utilisation plans capacity and knock-on effects on journey times for traffic and buses. widening is considered to have the In this context and based on the greatest impact as it reduces the signal settings presented in Table Case studies of other schemes have Regent Street southbound movement 1, the VISSIM model has predicted indicated that the introduction

S from two lanes of traffic to one. the following impacts on traffic of diagonal crossings can have a The reduction in traffic-traffic and and buses of the scheme: significant impact on traffic. This traffic-pedestrian inter green times is not predicted to be the case at • With the exception of the Princes and the removal of the Princes Street Oxford Circus. Although the signal Street side road traffic which will side road traffic has also helped to setting standards have been relaxed be removed from Oxford Circus, mitigate this potential effect. New by reducing the ‘invitation to cross’ all other traffic currently passing

HWAY or improved bus lanes have been to 8 seconds this has in effect through the junction will still be provided on both sides of Regent able to do so on a typical day.

G been one of the measures used to Street to provide buses with priority minimise the impact of the overall on these approaches to the Circus.

HI scheme. The proposed footway 66 ‘Scrambled’ pedestrian crossings at 57 signal controlled junctions – a case study

• Very little change in bus journey times is likely to occur – average journey time improvements of 7 to 15 seconds are predicted for buses during the weekday Inter Peak Hour and weekday PM Peak Hour. Journey time improvements are predicted on at least five of the eight possible bus turning movements during these periods. • Very little change in the journey times for other vehicles is likely to occur – journey times are predicted to change by between +3 and -8 seconds during the two peak hours. Journey time improvements are predicted for around half the turning Figure 12 – VISSIM Microsimulation model of the scheme movements during these periods. benefits significantly outweigh the • Locations where a diagonal • The length of traffic queues is benefits and disbenefits for other pedestrian desire line exists but predicted to be similar to existing. modes given the huge number of is not served by the existing Although small increases in queue pedestrians passing through the crossings, for example between lengths are predicted for some junction. Despite this the business two trip generators or attractors approaches this will be offset by case results do suggest that diagonal located on diagonally opposite the relocation of the stop lines at crossing solutions could represent corners of a signalised cross Oxford Circus which will provide viable schemes elsewhere even roads. In these circumstances, the greater ‘stacking space’ between recognising that most locations scramble crossing as the potential the Circus and adjacent junctions. in the UK will have significantly to reduce walk times and distances The probability of ‘blocking fewer numbers of pedestrians. as well as reduce the pressure on back’ of queues occurring is existing footways and crossings. therefore similar to the existing. Overall conclusions & • Other locations where a Notwithstanding that the impact signalised crossroads exists on traffic and buses is essentially future applications of with an ‘all red’ traffic stage ‘neutral’, increasing the cycle time scramble crossings where further improvements for from the current 112 seconds to The scheme at Oxford Circus includes pedestrians would be welcome 116 or 120 seconds is now being an innovative scramble crossing perhaps in relation to a new considered by TfL to provide the arrangement which is untried as a development. In this situation, highway network with greater solution in such a high profile location benefits can be provided for resilience to un-planned events or in the UK where a constrained pedestrians whilst potentially accidents at Oxford Circus or nearby. environment is coupled with very high avoiding significant impacts on The overall financial benefits and pedestrian and vehicular demand. vehicle flows and delays. It is disbenefits generated by the scheme Limited guidance exists in the UK accepted, however, that this is as estimated by the Legion and VISSIM on scramble crossing solutions and likely to be highly site specific. models and subsequently included this tends to focus on the design The Oxford Circus scheme was in the TfL Business Case results in aspects of scramble crossings instead implemented at the end of 2009. It £6m of benefits per annum which of identifying the circumstances in is anticipated a monitoring exercise can be disaggregated as follows: which they should be considered will be undertaken in accordance • £5.1m - benefits for as a way of promoting pedestrian with the approvals for the scheme pedestrians per annum priority. Based on the experience of obtained from Transport for • £800,000 - benefits for developing the scheme for Oxford London. This exercise will examine bus users per annum Circus and reviewing previous and record the actual impact of HIGHWAY • £10,000 - disbenefits for case studies, suitable locations the scheme on junction operating car users per annum in the UK for scramble crossings conditions, particularly impacts on could potentially include: buses and other vehicles as well as • £10,000 - benefits for the changes in pedestrian behaviour. taxi users per annum • Town or local centres where there are crowded footways The results of the monitoring The overall result is a Benefit – Cost or crossings; where pedestrian exercise will help to further clarify Ratio (BCR) for the scheme of flows are significant and the potential of introducing 12.9:1. This is a very healthy BCR pedestrian connectivity needs scramble crossings elsewhere. as typically TfL is aiming to achieve to be improved but where an a BCR of at least 1.5:1 for schemes existing junction for vehicles S that it is funding. The pedestrian also needs to be maintained. 67 HIGHWAYS 57 8. 7. 6. 5. 4. 3. 2. 1. References TTS6, DesignStandards forSignalSchemesinLondon,Transport forLondon,August2002 Bridges (DMRB),Vol. 8,Section1.TheStationeryOffice (TSO)September1981 TA 16/81,GeneralPrinciplesofControl byTraffic Signals,DesignManualforRoadsand Fruin, J.J,PedestrianPlanningandDesign.Alabama:ElevatorWorld Inc,1987 Crossings, DepartmentforTransport (DfT)(March 2005) Traffic AdvisoryLeaflet(TAL) 5/05,PedestrianFacilitiesatSignalControlled Manual forRoadsandBridges(DMRB),Vol. 6,Section2.TheStationeryOffice (TSO).November2004 TD 50/04,TheGeometricLayoutofSignal-Controlled JunctionsandSignalisedRoundabouts,Design Scramble: AnEvaluation,UCBerkeleyTraffic SafetyCentre, December2003 Bechtel, A.K,K.E.MacLeodandD.R.Ragland,OakwoodChinatownPedestrian Paper submittedforpresentation attheTransportation Research Board AnnualMeeting,Washington D.C2006 Ishaque, M.MandR.BmNoland,Micro-simulationSignalizedPedestrianCrossings, ComparisonsofAlternative submitted forpresentation attheEuropean Transport Conference, theNetherlands2008 Guidelines forInstallationofScramble(allred phase)PedestrianSignalsatIntersectionsPaper Ahuja. S,SBose,TvanVuren andDRagland,Towards theProvision ofUltimatePedestrianPriority: signal controlled junctions–acasestudy 68 ‘Scrambled’ pedestriancrossings at HIGHWAYS 69 “Tragically the author of this paper, Cressida Spachis, passed Spachis, passed Cressida author of this paper, the “Tragically of 33 following a pulmonary at the age away on 6th March embolism. Cressida, is published not just in memory of The paper her contribution to Atkins and the industry. but as a tribute to Business Manager for Leader and Deputy was a Group Cressida within Transport (N&T) business unit & Traffic the London Networks 1999. and had been with Atkins since Planning & Management, first class honours graduate in Civil Engineering She joined us as a worked on a variety during her career Imperial College, and from the UK as well as clients across for public and private of projects a number of secondments. included Over the years, her clients authorities. for London, BAA, LOCOG and a range of local Transport Management Manager for the Route 38 Corridor was Project Cressida for intensified bus priority Pilot Study which has developed principles many accolades received London. The project now being used across team was due to be part of the client/consultant and Cressida to Bus Services award the Improvements receiving earlier this month. Awards at the London Transport well- was enormously talented, bright and Cressida leader but also as a not only as a group respected, displaying a passion for her work and team player, with her to all who worked a vitality that endeared missed. Her contribution to She will be greatly her. for many years to come.” Atkins will be remembered Dr Andy Southern Managing Director Planning & Management Transport Highways and Transportation Delivery of bus priority projects projects bus priority of Delivery approach - A partnership HIGHWAYS 70 Delivery of bus priority projects 58 - A partnership approach

Cressida Abstract Spachis Population and employment growth in London is projected to continue as part of a national and Europe wide trend seeing an increased proportion Managing Consultant of the population living in cities. In London, the residential, commercial Highways & Transportation and non-commercial densities are already high and the historic road layouts can not accommodate increased trips by private vehicle and this situation presented transport planners with a challenge.

Introduction Population and employment growth based on a robust benefit/ project changed in 2005 to that of a in London is projected to continue as cost framework which reflect “Whole Route Corridor Management” part of a national and Europe wide passenger requirements; and Pilot Project (now referred to as trend seeing an increased proportion • Other than the further extension the Route 38 Pilot Project). of the population living in cities. In of Congestion Charging (or This paper highlights the importance London, the residential, commercial road pricing) and other real of a strong partnership approach and non-commercial densities are traffic restraints, the clearest in developing and delivering the already high and the historic road potential cost savings are in Route 38 Pilot Project, a project layouts can not accommodate Intensified Bus Priority (provided that has undergone changes in increased trips by private vehicle and that enforcement is effective management, approvals and network this situation presented transport and sustained) and delivery of conditions throughout its life. planners with a challenge. the Cashless Bus strategy. Route 38 operational conditions In 2003, Transport for London (TfL) (Source “London Buses, MAYOR OF London Buses carried out a Strategic LONDON, The case for investing in London’s Route 38 operates in London, Review showing that the bus service buses, Presenting the Results of the London between Victoria (Westminster) and in London provides an indispensable Buses Strategic Review, September 2003”) Clapton Pond (Hackney) through and increasingly valuable service to In line with this Strategic Review, TfL the Boroughs of Camden and the whole of London, particularly in Surface Transport Bus Priority Team Islington. This area of the capital social inclusion terms. Key conclusions (BPT) promoted two Intensified Bus is very densely populated and the included, but were not limited to: Priority projects on the bus Routes services are in heavy demand for • The (at the time) current subsidy 149 and 38, aiming to introduce commuting and social use including of London Buses delivers a ‘step change’ in bus priority. Bus tourism. Route 38 is a 12.5km real ‘value for money’; Route 149 was mostly operated long, high-frequency corridor with HIGHWAY on the Transport for London • The bus is the only means of 25-30 Route 38 buses per hour, Road Network (TLRN), whilst bus increasing capacity rapidly as well as other routes, and serves Route 38 was mostly operated on to meet the challenges of up to 7000 passengers per hour. Borough roads. The Route 149 delivering the Mayor’s London project was not proceeded with. Route 38 was converted from Plan and forecasted population Routemasters to articulated buses in and employment growth; In 2003, the Route 38 “Intensified Bus October 2005, and it was announced Priority” Pilot Project was set up and • The service planning guidelines that it will be converted back to jointly promoted by TfL BPT and the used to determine the double deck buses at the end of 2009. London Bus Priority Network (LBPN) London bus network are S Central Sector. The focus of the

71 58 Delivery of bus priority projects - A partnership approach

The reliability and journey time of The project involved approximately The project structure is shown Route 38 and other bus services on a £20m investment programme in the Figure 1 below. the corridor was poor and buses incorporating measures for bus When initially set up, the SRO were subject to excessive delay and priority, streetscape, pedestrians, for the project was the Head of disruption at many locations. Much of cyclists, bus stop accessibility, TfL BPT. However, in order to this was due to the lack of highway enforcement, safety, signals (SCOOT, foster a partnership approach, capacity, conflicting demands of AVL), Selective Vehicle Detection the SRO role was transferred to other road users, inappropriate bus (SVD) and iBus, converting the the LBPN Central Sector, and stop layouts, limited waiting and Routemaster to articulated bus was undertaken by the Head of loading restrictions, junction and operation, and innovative ideas Transportation for City of London. side road disruptions, inadequate for traffic management. In addition to the formal Partners enforcement, and the fact that bus The importance of the project has of the project, relationships were priority measures were present in been demonstrated by the fact developed with other partners, some sections of the corridor only, that TfL has continually used it as referred to as Stakeholders. These and often limited to peak hours. a test bed for the Third Generation included TfL departments such as of Bus Priority (3GBP). The link has the Public Carriage Office, Cycling The Route 38 Pilot Project been achieved through TfL’s 3GBP Centre of Excellence and London Project Manager attending the Underground (LU), as well as Key objectives fortnightly formal progress meetings Borough departments other than the of the Route 38 project, and when The Route 38 Pilot Project was set delivery Partners, and Developers. relevant also the fortnightly technical up to provide a ‘step change’ in The existing road layout has emerged progress meetings, and feeding bus priority through Intensified through a process of refinement back to the 3GBP the experience Bus Priority schemes, building on and competition between the gained and lessons learnt. measures previously implemented modes and the various interests of under the London Bus Initiative (LBI) The Project Partners frontages and the Local Authority. and the LBPN. Following the change In order to make changes it is of focus to that of Whole Route The Route 38 Pilot Project is being essential that all interests are properly Corridor Management, the project promoted jointly by TfL BPT and the reflected in any planned changes. aimed to improve bus reliability and City of London, the Lead Borough Key delivery Partners included TfL’s journey times, whilst adopting a of LBPN Central Sector, and in Network Assurance Team (NAT) and holistic approach as far as possible partnership with the four London Department of Traffic Operations through balancing the competing Boroughs the route goes through, (DTO) as well London Buses Area user demands and providing namely Westminster, Camden, Traffic Controllers and the Borough’s benefits to other users as well as Islington and Hackney, who are Street Managers. In addition, close buses. Some examples include to: the highway authorities for much working was important at a local • Enhance the environment of the network concerned. level with town centre managers, along the route corridor; other consultants and, vitally, • Improve safety and facilities Borough Councillors and the public. for pedestrians and cyclists; • Create a more pleasant and safe The Mayor's Transport Strategy bus stop environment to make it easier for all passengers to Project Board get on and off the bus; and Senior Responsible • Provide additional benefits, TfL Bus Priority Team LBPN (Project TfL London Buses Offi cer (SRO) & Project where possible, for all road users, (Policy and Funding) Assurance) (End User) Manager City of London communities and businesses.

Key characteristics Atkins - Technical, Programme Borough and TfL and Managerial Support Political Approval The Route 38 Pilot Project started in May 2003 and is due for completion Project Delivery in March 2010. During its lifetime, TfL Surface City of LB Camden LB Islington LB Hackney

S the project has encountered a Transport Westminster number of changes. Partnership working with the Client, the various Partners and Stakeholders proved to Technical Approvals - TMA, TSSR, etc be key to the change management and successful project delivery. Delivery on Street HWAY

G Represented on Management Team

HI Figure 1 - Project structure 72 Delivery of bus priority projects - A partnership approach 58

• Local Council Elections in 2006 leading to a change of political lead in certain Borough Partners, hence a change of Council priorities; • The introduction of the Western Extension Zone to the London Congestion Charging in 2007, leading to changes in the traffic patterns of Central London; and • The development of a new TfL Business Case Assistant in 2007 requiring new business cases to be developed for all sections along the corridor. The sections below describe the project stages and how they have been impacted upon by these changes, and highlight Figure 2 - Theobald's road footway and bus stop improvement how partnership working was key to the change management The regular day-to-day management Change control and management and to the project delivery. team of the SRO, the LBPN Senior During its lifetime, the project Route Co-ordinator, the 3GBP Project stages Project Manager and Atkins was encountered a number of changes small. However, it was vital that in management, approvals and Although a higher profile project the management team was fully network conditions, which influenced than some, the Route 38 project still committed and involved with all the project progress, programme, covers the lifecycle of a standard project Partners and Stakeholders, and delivery. Some of the changes traffic engineering scheme, divided whether end users or delivery. encountered and managed include, into the following stages: but are not limited to, the following: By following a partnership (1) Scoping; approach working closely with • The introduction of new approvals processes, including (2) Feasibility (including Stakeholders and Partners, the public consultation); project was able to tailor schemes, the introduction of the Traffic (3) Preliminary Design (also approaches and methodologies Management Act 2004 including consultation); to the preferences, requirements (TMA), and subsequently the and needs of the ultimate end introduction of the TfL Network (4) Detailed Design; Assurance Team (NAT); and the users, the general public, whether (5) Implementation; and travelling or just using the street. new TfL Department of Traffic Operations (DTO) processes (6) Monitoring. Where a Borough had, for example, and signals guidelines; The route was initially divided into a safety priority to introduce an All 13 sections for ease of scheme Red Pedestrian Phase at a junction, • The project focus changed in 2005 from an Intensified Bus development, consultation, the scheme was adjusted to include it approvals and implementation. whilst at the same time maintaining Priority to a Whole Route Corridor Management. This meant that bus benefits. Or, where a Borough Scoping required provision of loading bays the project was no longer purely for businesses, the scheme was a bus priority project, but needed The scope of the project was set developed to remove the bays from to adopt a holistic approach as by TfL at the initiation, in 2003, the carriageway in order to improve far as possible, hence balancing and was to investigate the use bus and traffic flow, but bays were the competing user demands of intensive measures to give provided as inset or shared surface including those of pedestrians, priority to buses. The scope was with widened footways, hence not cyclists, and general traffic; changed to a demonstration only accommodating businesses, but • The conversion of Route 38 buses of corridor management and HIGHWAY also providing additional pedestrian from Routemaster to articulated a pilot for 3GBP in 2005. facilities. Another instance would be in 2005, requiring the redesign of where a Borough had inspirations to all schemes and bus stops along create an attractive town centre area, the route to cater for them; a scheme was developed diverting • The project Governance was general traffic away from the town transferred in June 2006 from centre, but retaining buses through TfL to the City of London, it, hence reducing bus journey times highlighting the importance whilst improving the public realm of the Borough partnership; S and pedestrian environment. 73 58 Delivery of bus priority projects - A partnership approach

Figure 3 - existing streetscape Feasibility Consultation on the preferred options Based on the scope, options were • Technology measures; such developed by Atkins for the route as GPS tracking of buses, Once Borough Partner officers were and set out into scheme-based dual SVD detection, bus door satisfied with the schemes jointly logical entities. Subsequently, activated signal detection, and developed, approval was sought to the chosen schemes were signal/ stop integration; carry out public consultation from grouped to form sections for • Mechanisms; such as corridor road the Lead Member, Cabinet Members the purposes of consultation, user charging, multiple zone based at Committee, or using delegated design and implementation. road user charging, statutory authority. The approval to consult The options included ‘blue authorities/builders renting was sought by the Borough officers, sky’ measures, developed as carriageway, roads policy for bus and following the processes of the a demonstration of all ideas corridors, signal policy and design individual Council. Public consultation with perceived benefit and guidance review, identification was undertaken on a section basis to application in London. Some and classification of highest allow all consultees to appreciate the of those considered were: priority bus routes, and review extent and effects of the options. • Infrastructure measures; such of Mayor’s Transport Policy; and A common project approach and as guided bus ways, bus ways, • Additional ideas and concepts; format for consultation were high occupancy vehicle (HOV) such as route based loading followed for all sections, whilst lanes, heavy goods vehicle strategy, permitted bus lane also tailoring them to the specific (HGV) lanes, median bus lanes, variants, and removal of requirements of the Partner. bus segregated strips, inflatable bus / pedestrian conflict. Common consultation approach humps, converting disused railway The choice of which options to All public consultations consisted into bus ways or general traffic develop was made in 2004/5. At of a consultation pack including links, guided bus flyovers, raised the time, the political will was not a leaflet questionnaire, bus stop surfacing, raised carriageway/ sufficiently strong to take many drops, a telephone contact number, junctions, grade separated of the ‘blue sky‘ schemes forward postal address, on-street posters, carriageway, tidal bus/ traffic although some have re-emerged in advertisement in the local paper, and S lanes, underpasses, bus only later programmes. Elements of these exhibitions on weekdays, week- roads, bus lane colour changing measures, however, were incorporated ends, day and evening. In addition, road studs (lights), central bus in the preferred options to be taken a dedicated webpage was created lane – 24hrs – route corridor, bus forward to preliminary designs. It is for each consultation showing all dedicated junction lanes, diversion possible that opportunities for ‘blue information on the consultation pack of traffic / buses to Royal Parks sky’ measures will arise again in the and giving consultees the opportunity HWAY roads, rail bridge modifications, future, as traffic conditions worsen to comment online. The dedicated and tube station access links; and the political imperatives change G Route 38 webpage was also linked with the need to address the major to the Borough’s web page.

HI environmental concerns facing society. 74 Delivery of bus priority projects - A partnership approach 58

Figure 4 - Piccadilly Circus proposed streetscape

During the early stages of the Tailored consultation approach knowledge available within the local consultation stage, discussions were The public consultation, however, community was captured. In addition, held with the Borough Partners and was also tailored to the requirements the report presented post-consultation the Client regarding the branding of and processes of the individual scheme amendments and additional the consultation material. Should each Partner. This included agreed scheme supporting technical information. delivery Partner’s branding be used proposals and leaflets, the extent of for the schemes in their Borough? the consultation pack distribution Preliminary design Or should the Route 38 Pilot Project areas, stakeholder lists, exhibition Following consultation and approval develop dedicated branding? It was locations, appropriate local papers, to proceed, preliminary designs were soon agreed by all that a dedicated local meetings with associations developed by Atkins in liaison with the Route 38 project branding clearly such as traders and residents, etc. five delivery Partners to demonstrate showing the partnership would In addition, where the Partner had how the schemes operate. The reinforce the identity of this Whole a local scheme in the vicinity of the designs developed for the route were Route project, whilst also speeding Route 38 section being consulted on, presented to stakeholders within up the process by not requiring a and if appropriate, it was included in the relevant Partner for feedback full review and approval by each the Route 38 consultation material in and input, through meetings Partner’s consultation team. order to reduce timescales and costs. and correspondence. Using their This common approach and An example would be the reversal knowledge of the local network and methodology was very well received of Air Street or the pedestrianisation their Council’s strategy and priorities, by all the Partners. In a true of Glasshouse Street in Westminster the delivery Partners commented on partnership way, good practices of the being included in the Route 38 the schemes presented, and advised Partner’s consultations were adopted Piccadilly Circus consultation leaflet. on the elements that needed to be into the Route 38 one; and vice-versa Atkins took the lead on all public included, adjusted, or all together some Partners adopted elements of consultations carried out, however, removed, in order for the scheme the successful Route 38 consultations involving the Borough officers as to be approved by their Council. in their own consultations or much as they wished to be involved. In some cases, the requested notification communications.

Following the public consultation, changes were outside the scope of HIGHWAY An example would be Camden a consultation report was compiled the project. In these cases, a process requesting Atkins to produce their and issued to the Borough for of communication, negotiation, works notification for Bloomsbury each section. This report analysed and compromise with Partners Way, using the easy-to-understand all responses received, in the form was followed until a scheme graphical maps developed. of leaflet questionnaire returns, acceptable to all was developed. comments received by telephone, post or posted online, by proposal type, area, etc to ensure the local S

75 58 Delivery of bus priority projects - A partnership approach

This was an opportunity for Borough Partner approvals Partners to provide improvements At the end of public consultation, on their network, other than bus Borough Partners needed to related. Urban designers were also take the proposals through their involved at this stage to advise Council’s approval process prior on the feasibility of urban realm to commencing detailed design. opportunities and improvements, Assistance was provided to officers such as the creation of open spaces including Committee or Member for pedestrians, etc. Schemes were reports and material, attendance therefore adjusted or redeveloped of and presentations at Committee together with the Partners to reflect meetings, Committee requests and their aspirations as far as possible. follow up actions, additional meetings An example would be the recently with Members and any other implemented Shaftesbury Avenue requirements until scheme approval scheme, which included in addition was granted. The level of assistance to bus measures, the widening of provided varied from Borough to footways, shared surface loading Borough based on the officers’ bays to address local businesses’ requirements, the Council approval Figure 5 - Route 38 map needs, improvements to safety and an process and proposals in question. element of public realm improvements For instance, in the case of Islington, For the Upper Street (TLRN) section, linked to the Theatreland project, in we met with Council Members, for instance, the original proposals line with Westminster’s aspirations. drafted Committee reports, presented included a bus lane leading to Another example would be the schemes at Committee meetings a bus pre-signal. The proposals introduction of an All Red Pedestrian and we dealt with the follow up were subsequently revised as a Phase at the junction of Gray’s Inn actions and Committee requests; result of partnership working with Road/ Rosebery Avenue in Camden, effectively relieving the officers from TfL Road Network Development or the dedicated inset parking the additional workload as far as (RND) to consist of relocated and and loading provision on Essex possible. For Camden, information new pedestrian crossings, signal Road in Islington to accommodate was provided to officers to draft the coordination adjustments, and bus businesses’ and visitors needs. Committee reports but we attended stop layout changes. Finally, they London Congestion Charging the Committee meetings in order to were further revised to widen the and its Western Extension Zone provide any background technical footways, introduce shared surface assistance to officers; the officers took loading bays and widen the bus Whilst the Route 38 project the proposals through the Council lanes to accommodate cyclists. commenced in 2003, the year the London Congestion Charging Zone approvals process and our role was to This iterative process of revising (CCZ) was introduced, the baseline assist them technically. For Hackney, and fine tuning the schemes until data available for use to carry out information was provided to officers an acceptable solution was found feasibility and develop preliminary who met the Lead Member. For set the Partners’ relationships. designs was collected prior to the Westminster, we provided information The relationships developed CCZ. This and the subsequent to officers for the Member reports varied from Partner to Partner, introduction of the Western Extension and we presented the proposals and depending on the officers, their Zone (WEZ) in 2007, altered the their benefits to the Lead Member Council’s priorities and buy-in of baseline conditions in terms of traffic during meetings. Where consultation the project, and the negotiations patterns, and required the schemes was required on sections of the TLRN, held to reach an agreed scheme. to be reassessed and adjusted. similar processes were followed Scope changes: but in liaison and collaboration Conversion from Routemaster with TfL RND and in consultation From an Intensified Bus to Articulated Bus Priority Project to a Corridor with the relevant Borough. A further change was the conversion Management Project of the Route 38 Routemaster buses The designs were initially developed into articulated buses in 2005. whilst Route 38 was an Intensified This required bus stop and scheme Bus Priority project building on the designs to be adjusted once again LBI and LBPN schemes; however

S to reflect the new bus type. subsequently they had to be adjusted Working closely with the Borough to reflect the new focus of a Whole Partners helped considerably to Route Corridor Management manage these changes, and hence and therefore the bus priority minimise as far as possible delays schemes needed to be adjusted to to the project programme, and provide more holistic solutions.

HWAY risk in terms of the feasibility and buildability of the schemes. G HI 76 Delivery of bus priority projects - A partnership approach 58

Furthermore, for all parties involved the transfer of governance translated, amongst other things, into different project and financial reporting requirements and processes to follow. More specifically, we now needed to directly comply with the requirements and processes of the City of London as the new Client, but indirectly also with those of TfL who remained the budget holder. Project instructions and new briefs, for example, were now to be issued by the City of London, instead of TfL. However, elements of TfL’s “Spearmint” project management reports were required, but adapted to the requirements of the City of London and Route 38 project. The transfer of governance to Figure 6 - Shaftesbury Avenue footway widening the City of London seemed to assist positively the relationship The partnership approach followed In a couple of cases, the partnership of the Borough Partners with allowed each Borough Partner to was more difficult to develop and the Client and the project. The request, and receive, the level of maintain. Part of the reason was Partners seemed more willing to assistance they required at that internal Borough politics, where be involved in the development moment in time. This tailor-made perhaps bus priority was not their of the project, and happier that approach lead to a good exchange priority. This caused considerable the project was being moulded to of knowledge and assisted the delays to the schemes. their wishes and requirements. further development of our existing On the contrary, where the The decision making process, relationship with the Partners. In most partnership working was successful however, seemed more complicated cases this approach also minimised the schemes were quickly adjusted in some instances, with the delays to the project programme, and successfully taken through the City of London being the Client since the officers were able to request new Borough Lead Members. but TfL the budget holder. our assistance on any task they were Transfer of project governance too busy to carry out. There was The already in place good relationships only one instance where Borough During the consultation stage, between the Client and the Partners officers did not take advantage of project governance was transferred allowed open communication to our offer to assist in order to alleviate from TfL BPT to the City of London, take place, and assisted greatly in their workload; and this lead to a highlighting the importance of the understanding and dealing with the delay to the programme of a few Borough partnership, although TfL new requirements in an effective way. months. However, it is understood remained the funding authority. Detailed engineering design that part of that delay was due to The transfer of governance created the sensitive nature of the scheme. some confusion as to who was the Following the granting of Council Local Council Elections Client. As relationships became more approval, schemes moved to informal with time, it was very easy Local Council Elections in May detailed (constructional) design. to take instructions from a partner 2006 imposed a purdah restriction This stage possibly involved the who did not have the appropriate in terms of carrying out public greatest variance in the assistance authority. This required a conscious consultation three months prior to provided to the Partners, again effort from all Partners to operate and three months after the elections, responding to their requirements, in line with the responsibilities of hence causing delays to the project preferences and available resources. their new roles, instead of taking programme. In addition to this delay, For Islington and Hackney for

forward those already established in HIGHWAY the elections lead to the change of example, Atkins carried out full the first few years of the project. political leads in certain Borough detailed design in line with the Partners, and hence the change of Borough’s design guidelines, and priorities. In those cases, we worked handed over the designs to the very closely with the Borough Borough for their contractor to Partner officers to understand the implement the schemes. This new priorities as quickly as possible, included preparing the Scheme Brief and try to adjust the schemes in and Traffic Signals Supplementary a way that addresses the new Reports (TSSR) required for London schemes and taking them through priorities, whilst still delivering on S the Route 38 project objectives. to signals approvals by TfL DTO.

77 58 Delivery of bus priority projects - A partnership approach

It also included preparing scheme Change in technical This arrangement assisted greatly to TMA submissions for the Borough approvals processes develop an informal partnership with to submit on the London Works The TMA was introduced in TfL DTO and reflected the importance computer database and liaising with 2004 giving highway authorities of the signals approvals to the Route TfL NAT until approval is granted. the powers and responsibility to 38 project programme. Through this Regarding the works TMA, we preserve and promote uninterrupted partnership DTO engineers were assisted the contractor as required. traffic flow on their roads, i.e. involved in the development of signals Road Safety Audits were carried broadly ensuring that the best schemes as early as possible and out by Atkins auditors independent balance for all users is achieved. assisting in the scheme adjustment to of the scheme engineers. address any concerns they may have. In London, this lead to the creation Under this arrangement work was of the TfL NAT reviewing TMA While the Route 38 Project was carried out in house with regular submissions for both schemes considered a priority project by TfL, it progress meetings with the delivery and works. These submissions also had to “compete” against other Partner. Designs were issued to the have to be made by the highway major projects or resources; in terms Partner for final check and approval authority through the London of both design and implementation at the end, prior to being handed Works computer system. but also in terms availability of over to the Partner’s contractor. highway space. One example would In addition, all schemes involving Beyond that stage Atkins’ involvement be LU’s Tottenham Court Road Station traffic signals in London require mostly consisted of covering the Upgrade, including integration of to be assessed, and the design designer’s responsibilities. In the Crossrail. While partnership working and modelling approved, by TfL. case of Islington, Atkins were also functioned well with the delivery During the Route 38 project, the granted the role of CDM Coordinator; Partners, TfL BPT and DTO, LU and TfL department dealing with signals whilst Hackney appointed one their advisors were rail focused, changed from the Traffic Technology of their own engineers. rather than streets and highways Systems (TTS) to the DTO and with focused. This added a layer of Camden, on the other hand, that changed the signals approvals complexity to the partnership working developed the detailed designs in process which became significantly leading to significantly greater input house with technical assistance from more thorough and time consuming. from the Route 38 team than with Atkins as and when required. This Any scheme requiring alteration to other sections and partnerships. included for example signals design, traffic signals needed to be included traffic modelling, and generally signals in TfL DTO’s Capacity And Resource Implementation approvals, as well as road safety Planning (CARP) programme, and audits and scheme TMA submissions. subsequently submitted to DTO as Implementation was always Regular progress meetings were in part of a Scheme Brief and TSSR intended to be managed by the place to manage progress, deal with for the design and modelling to delivery Partners and carried out queries and decide on the best way be reviewed and approved before by the Partners’ contractor on a forward. The relationship developed submission for scheme approval under section by section basis. Atkins with Camden was a good partnership, the TMA. The submission needed retained the designer responsibilities where all parties contributed their to include, amongst other things, for those sections we developed stronger skills in order to achieve the traffic modelling developed based on detailed design for. In addition, best possible solution, as quickly as DTO’s new Modelling Guidelines. Atkins retained further involvement possible. An example would be, in The DTO scheme approval process and where we were appointed CDM addition to the assistance mentioned modelling guidelines were altered at a Coordinator. In addition to the above, Camden’s request for Atkins time where the modelling assessments above, assistance was provided as to prepare their notification leaflets, of Route 38 schemes were close required by the individual Boroughs. using the easy-to-understand to completion or completed. All To date, implementation of sections graphical format developed for models had to be rebuilt in line with has been progressed as follows: the Route 38 consultation, but the new DTO Modelling Guidelines, adapted to Camden’s branding. • Implementation has and approvals submissions had to been completed for the Lastly, for Westminster, Atkins be in line with the new processes. following sections: developed the designs to an agreed In order to help manage change, level and handed them over to • Gray’s Inn Road/ Rosebery and minimise project delays as much the Borough’s term consultant Avenue, in Camden; as possible, TfL DTO offered to to complete and take forward to • Theobald’s Road, in

S provide a DTO Route 38 coordinator implementation. The partnership Camden; and acting as the main point of contact here was extended to include the for all Route 38 schemes. This • Shaftesbury Avenue, Borough’s term consultant. Regular coordinator attended Route 38 in Westminster. Client/ Partner project meetings progress meetings weekly, and was • Implementation is in progress were in place to ensure appropriate able to advise the Route 38 team for the following section: coordination of works and appropriate regarding signals, but also gave HWAY awareness of progress; in addition, • Bloomsbury Way, in Camden. feedback to the TfL DTO engineers communication and collaboration G working on Route 38 schemes. took place between the two

HI consultants to achieve best outputs. 78 Delivery of bus priority projects - A partnership approach 58

• Implementation is planned Bus travel data was collected Business case to commence imminently for through a versatile GPS device the following sections: which collected speed and position As the Route 38 project was a TfL BPT funded project that started as • Upper Street, TLRN in Islington; data on a second-by-second basis. This frequent ‘logging’ of data an intensified bus priority project, • Essex Road, in Islington; and ensured a large database of route the business case was based on • Graham Road, in Hackney. information was recorded which, in the funding granted by TfL and focussed on bus passenger journey • Implementation will commence turn, allowed for versatility in analysis time savings. Monitoring was in the near future subject to which was not achievable through agreed to be mainly on bus journey final approvals being granted other methods of data collection. times, however, other monitoring for the following section: Post implementation or ‘after’ parameters were also considered • Piccadilly Circus, in journey time data will, hopefully, be where appropriate such as accidents, Westminster. collected at the end of the Route 38 impact on pedestrians, cyclists etc. project in 2010/11 through TfL’s new The remaining sections of the corridor A business case was developed were dropped from the Route 38 iBus system. iBus appears to have similar functionality to the system by TfL BPT, and revised at various programme due to being in an area stages of the project as more of major redevelopment works, used for the Route 38 Project. It is expected that it will be key to the information became available and overlapping with other bus priority accurate. During the consultation schemes more progressed than promotion and delivery of 3GBP, and other bus schemes, through and detailed design stages of the Route 38, following consultation Route 38 project, TfL developed outcome, or for political reasons. providing a good understanding of the existing on street conditions, and issued a new TfL Business Case The remaining schemes will not be Assistant, catering for benefits to all deferred indefinitely and will be and assisting for accurate and representative modelling. modes and safety, rather than just “packaged” to enable the Borough’s journey times. TfL funded schemes to progress and deliver them through Conversion from articulated were now assessed using this new a different funding process. to double deck Bus Assistant. This is a fixed model, It was previously mentioned that requiring though the development Monitoring Route 38 was converted from and interpretation of assumptions. The main monitoring parameter set at Routemasters to articulated buses in The 13 Route 38 schemes were the beginning of the Route 38 project 2005. In 2008 it was announced that now required to be re-assessed was bus journey time, in line with the bus Route 38 would once again be using this Assistant. The urgency to Intensified Bus Priority nature of the converted, from articulated to double complete the assessments would project at the time. ‘Before’ data was deck buses at the end of 2009. have put a considerable strain on the provided by TfL at the beginning of This conversion is expected to impact resources of the TfL BPT and it was the project in 2003. However, network on traffic patterns, hence altering the therefore agreed, that the schemes changes such as the introduction baseline once again. This, and other would be shared with Atkins. of the Congestion Charging and changes yet to take place, will need to In order to achieve consistency in later its Western Extension Zone, be taken into account when analysing developing the business cases of all the conversion of Route 38 from the ‘before’ and ‘after’ journey time schemes and the development of Routemasters to articulated buses, data and concluding on the impact assumptions Atkins worked closely major development works such and benefits of the schemes. with TfL BPT to understand their new as the LU Tottenham Court Road model and develop and interpret Station Upgrade, as well as smaller assumptions in a similar way. schemes such as pedestrian crossings, meant that traffic patterns had changed prior to even commencing implementation of the Route 38 schemes. In order to set a more accurate, updated baseline, additional ‘before’ surveys were carried out. Various methods of collecting journey time data were considered HIGHWAY for the additional ‘before’ data, including TfL’s Automatic Vehicle Location (AVL) system and iBus. Unfortunately the latter was not available at the time. An alternative Global Positioning System (GPS) and Graphical Information System (GIS) was therefore used for undertaking travel time, speed and delay studies. S

79 58 Delivery of bus priority projects - A partnership approach

Conclusions The delivery of the Route 38 Pilot Project was a challenge due to many reasons including the five different delivery Partners and many Stakeholders involved, the length of the route, the difficulty of balancing conflicting user demands with bus benefits in the very constrained road network of Central London, political will, etc. The biggest challenge faced by the project, however, was probably the many changes experienced throughout its lifetime, including management, approvals and network conditions changes. All these changes posed real time and cost control challenges and impacted on the project. The project programme was extended by approximately three years. Whilst the project budget was not increased, scheme costs increased due to price inflation, scope creep and general increase in expectations. Project costs also increased due to the extended design, assessment and approvals processes. So was the Route 38 Pilot Project successful? Considering the complexity and challenges faced, delivering schemes on the ground is a success in itself. The schemes implemented to date seem to have already delivered considerable benefits to bus journey times, pedestrians, cyclists and other corridor users, as well as safety. The exact extent of these benefits, however, will be quantified and assessed when the outcome monitoring is carried out at the end of the project. Furthermore, the project demonstrated that a strong partnership approach assists greatly in managing and controlling change, and delivering a good scheme acceptable to all Partners, minimising time delays and costs. S HWAY G HI 80 59 WATER & ENVIRONMENT

Model extent quality the water work basis for The model of was the Mike 11 hydraulic which was also developed the river, Water as part of the same Strategic Water quality model Water development Modelling software In with the Environment consultation the software Agency (as regulator), Mike 11 (DHI; 2007 Service 1), Package quality water module, with its Ecolab was selected as the most appropriate simulating modelling software for quality and water in the Lower fows simpler models, In to contrast Stour. MIKE 11 includes detailed heat balance reliable and hence calculations, and temperature, calculation of water set of interactions a comprehensive quality water between determinands.

81

Abstract a new water investigate to Water South East Atkins by was commissioned part As scheme in the South of England. East of the water resource quality aspect of this work, a Mike of the tidal Lower model 11 Ecolab observed against model was calibrated The was developed. River Stour and software template on the standard developments data, leading to in the river. quality our understanding of water processes improving of abstraction a number test model was subsequently used to The the wider into feed to anticipated are model results The scenarios. impacthelp shape strategic assessment and ultimately environmental the area. management for resources water decisions regarding part of related environmental impact part environmental of related assessments and planning applications. quality frst stage of the water The an this project involved for work in the appraisal of baseline conditions Automatic catchment. Stour Lower deployed quality were water monitors locations throughout at several and spot sampling the catchment This Atkins. was also carried out by was designed programme monitoring the Environment complement to data collection in the Agency’s the construction into and feed area qualityof the water model. an overview paper will provide This and calibration of the development water River Stour of the Lower it illustrating how quality model, in the wider as a tool was employed impact assessment, environmental Water’s South to East relating scheme. resource water proposed Eliot Simons Consultant Principal & Environment Water Neil Upton Neil Upton Scientist Environmental & Environment Water Brian A. Cox Brian A. Cox Senior Environmental Scientist & Environment Water Vera Jones Vera Environmental Scientist & Environment Water ective planning application

New water resources options are options are resources New water parts in several of being investigated the south east of England in order supplies and meet the secure to The water. demand for increasing Water ‘Strategic South East Water project Investigation’ Resources and a successful deliver aimed to cost-ef increase aiming to a new scheme, for supply and for availability water the security supply in of water a considered work The the area. example wide range of aspects, for aquatic ecology, recreation, heritage, quality. and water hydrology, quality in which water way The of the as a result might be impacted, resources operation of new water for is a major concern schemes, a quality models are Water planners. these impacts in investigating key tool an important can form therefore, and, Introduction The case of the River Stour, Kent Kent Stour, of the River case The water resource management options: options: management resource water Water quality modelling as a tool for assessing new new assessing quality for as a tool modelling Water WATER & ENVIRONMENT 59 feed channelswere alsoincluded. discharges, andmarsh abstractions a dif boundaries, suchasnaturaltributaries, Pegwell Bay, seeFigure 1.Additional at and thedownstream boundary model waslocated atCanterbury ofthe The upstream boundary water motionormixing movement ofmassdueto random in space dispersion isthe to another; movement ofmassfrom onepoint processes. Advection is thesimple through anddispersion advection determinands inthe systemquality determines themovement ofwater module wassubsequentlyadded. This The (AD) advection/dispersion temperatures alongtheriver. wind speeddatato simulate water air temperature, sunshinehoursand which takeinto account humidity, allows for detailedheatcalculations, the hydrodynamic (HD) module. This balance’ and ‘wind’ optionswithin temperatures, by enablingthe ‘heat model wasto simulate water hydraulic modelto awater quality The frst step inmoving from the model water quality From hydraulic modelto potential hydrology andf Resources Investigation, to simulate Calibration pointsSandwichandPlucksGutterabstraction23forthescenariorunsare alsonotedonthemap. The locationsoftheupstream boundary(downstream ofCanterbury)anddownstream boundary(PegwellBay)are shown. Figure 1-MapoftheLower Stourarea. TheSectionoftheLowerStourmodelledishighlightedinblue.

use groundwater inf 5 , seeFigure 2. ow, artif cial ood risks 1 . Water modellingasatool for quality assessing new • for thehealthofanaquaticecosystem: determinands represent keyindicators Together,ortho-phosphate. these demand (BOD), ammonia,nitrate and oxygen (DO), biochemicaloxygen water temperature, salinity, dissolved determinands werequality simulated: project. The following seven water the specif andmodif was selected entitled ‘WQlevel4ColiPhos.Ecolab’, processes. The standard template, determinandsof water and quality containing dif a numberofstandard templates, 11EcolabThe moduleof Mike determinands.of water quality simulates thebiogeochemicalcycling module wasintroduced. This module respectively, theEcolab water quality for water temperature andsalinity simulations were enabledandtested Once the heatbalance andAD to biogeochemical interactions. thataresuch assalinity notsubject for simulating ‘conservative’ substances of allsubstances andisuseddirectly The ADmodulecontrols thetransport ieo 0C Additionally, key a riseof 10°C. ratea doublinginreaction with commonlyreactions experience on aquaticlife, asbiochemical temperature, have acriticalimpact changesinwater particular Water temperature, andin

c requirements ofthis erent combinations 82 water resource managementoptions:

ed to suit The caseoftheRiverStour, Kent ers • • • DO concentrations inthewater water exhibitslow valuesof decaying organic matter. Pristine waters, generally arisingfrom oxygen within demandexerted BOD isameasure ofthepotential or toxic un-ionisedammonia. yield products, suchasmethane zero, anaerobic orsepticconditions levels If persist around pollution. low levels asaresult oforganic when concentrations fallto very are dueto often asphyxia,kills for ahealthy ecosystem, asf system. Suf the overall ofthe water quality provide anexcellent indicator of model.the water quality tool for theADelementof usefulcalibration is hence avery by and biogeochemicalreactions assumed to remain unaf of oxygen inwater. is Salinity alsoafectsthesolubility salinity tolerance.salinity Additionally, narrowaquatic specieshaving very ecosystem, withanumber of have ontheaquatic abigimpact Similarly, can changesinsalinity when temperature changes. (aswithDO) their solubility ionisation ofammonia)oralter change theirform (asinthe constituents ofwater either

cient DOiscrucial

ected ected sh 59 WATER & ENVIRONMENT

ciently cant

ect on model simulations.

Model calibration methodology Calibration of data availability, consideration After the calibrate it was decided to the period qualitywater for model This 30/06/2007. 01/02/2007 to calibration period includes a range and is also suf levels of fow speed (daily) data were set as global speed (daily) data were boundary applied data and hence model. Stour the whole Lower to long to allow any initial model any allow long to no signif have to conditions ef ammonia, nitrate, BOD, Temperature, ortho-phosphate and DO were collected against data at calibrated ce for for ce ed. Ined. most

83 *R2 results from regression analysis **based on the results of analysis of variance, at analysis **based on the results regression from *R2 results carried out for these deter- dence level *** no statistical analyses were the 95% confi minands due to the limited number of observed data points; instead an assessment and model data. between simulated was made based on the visual agreement Figure 2 – Transport of solutes though advection and dispersion processes. After Chapra (1997). and dispersion processes. of solutes though advection 2 – Transport Figure and key issues results 1 - Summary of calibration Table determinand. (associated) with the calibration of each water quality determinand or location, a suitable determinand has been identif proxy a nearby cases this meant data from values from location or representative Where years. of previous a record that this the data set was so limited ‘generic a set of was not possible, based on an was used, values’ data and overview of all the available other similar studies. from experience boundary Meteorological data were the Met Ofobtained from suitable monitoring the nearest location (Manston; 63°24’E 16°61’N; 11.2km Plucks from approximately the heat into and fed Gutter) and wind elements of the balance (hourly), Air temperature model. times a day), humidity (three sunshine hours (daily) and wind

c - uvial uent

sh, causing sh, uent. High cation, the process of cation, the process uvial environments. uvial environments.

sources during high rainfall sources sources. agricultural from events Ortho-phosphate is also an important algal nutrient for in aquatic environments growth be the limiting and likely to nutrient in f primaryThe of ortho sources agriculture are phosphate and human ef of ortho-phosphateconcentrations eutrophication. can lead to is hence essential for algal growth algal growth essential for is hence High concentrations in water. excessive can lead to of nitrate or eutrophication rates growth the preferential in particular, and, toxic of potentially growth blue-green less algae over Due algal species. problematic high the solubilityto of nitrates, f can enter concentrations BOD, whereas sewage ef whereas BOD, typically exhibits very high BOD is measured BOD levels. collecting sample by a water and measuring the decrease the over in DO concentrations of a 5-daycourse incubation. an nitrogen, Ammonia contains algal growth. essential nutrient for It in all natural waters is present in very however, small amounts, of 0.1mg N/l can in excess levels of some sewage be indicative Itor industrial contamination. as total is generally measured for ammonia, which accounts ammonium aqueous forms: two ions (NH4+) and un-ionised ammonia (NH3). relative Their changes with pH abundance High of levels and temperature. to toxic un-ionised ammonia are and f invertebrates such conditions respiratory stress, and reduced hyperplasia as gill and disease. parasites to resistance nitrif Finally, to via which ammonia is converted and can oxygen consumes nitrite, be an important control therefore in the water. levels on oxygen and nitrogen Nitrate also contains • Water quality modelling as a tool for assessing new new assessing quality for a tool as modelling Water options: management resource water Kent the River Stour, case of The Flow, level and water quality and water level Flow, boundary and calibration data were Agency Environment from derived and Atkins spot sampling records Where monitoring. and automatic a specif for not available data were Boundary and calibration data • • WATER & ENVIRONMENT 59 Periods ofpotentially inaccurateboundarydataare noted(boxes1-4) concentration. Chlorophyll concentrations(daily averages)atPlucksGutterare alsoshown. Figure 5-Observedandsimulateddissolvedoxygen(DO;15-min data; fullcalibrationperiod) variability inspotmeasurements; fl owmeasurementslinkedtotidalcycle. toillustratepattern data for2000-2006(Environment Agencyspotsamples)toprovide anindicationofinterannual ortho-phosphate (dailyaverages,minandmax)concentrationsatPlucksGutter. Alsoshown:observed Figure 4-Observed(Environment Agencyspotsamples;2007only)andsimulated Figure 3-Observedandsimulatedtemperature atPlucks Gutter(15-mindata;fullcalibrationperiod). Water modellingasatool for quality assessing new 84 water resource managementoptions: The caseoftheRiverStour, Kent DHI, 2007c)wastakeninto account. DHI manual(DHI,2007a;DHI,2007b; relevant literature andadvice inthe Past experience withsimilarmodels, of acceptable determinand values. while staying withintheconstraints visual agreement data, withobserved carried outaimingto obtainthebest BOD andf followed(salinity), by nutrientsand determinand then theconservative Firstly, temperature wascalibrated, order ofincreasing process complexity. The calibrationprocess followed an the river were correctly simulated. with thetidalmovement ofwater in ensured thatADprocesses associated the salinelimit,seeFigure 1. This also at Sandwich,whichliesjustbelow was calibrated againstdatacollected resolve any calibrationissues, salinity atPlucksGutter istoosalinity low to resource scenarios investigated. As point undersomeofthewater Gutter asanabstraction would serve set wasavailable. Additionally, Plucks data the mostextensivewater quality Plucks Gutter, seeFigure 1,where and soconsidered to betrivialandare much restricted intime and/orspace here wereinstabilities observed very the ADcoef could beavoided by locallyincreasing caused by steep localgradients and DHI informed usthatthiscanbe negative salinitiesorconcentrations. model duringcalibration,leading to instabilities were inthe observed localisedandshort-lived Some models.of pastwater quality sense andcomparing withtheresults other thanby applyingcommon werecycles) accurately simulated, small-scale variations(e.g. diurnal eventsjudge whetherextreme or it wasnotgenerallypossibleto when usingspotdatafor calibration, of conditions inthatmonth. Thus, representative ofthewholereach time andspace ratherthanbeing only representative ofthatpointin thataspotsampleis acknowledges Interpretation of modelperformance samples (01/02/2007-30/06/2007). was carriedoutagainstmonthlyspot nitrate, ammoniaandorthophosphate 30/06/2007); calibrationfor BOD, 15-minute data(01/02/2007- wascarriedoutagainst and salinity Calibration for DO, temperature ofcalibrationresults Overview

nally DO. Calibration was

cient. However, the

59 WATER & ENVIRONMENT uent ow levels. levels. ow

uence of the uence

c length of data record c length of data record

cant impact on temperature, cant impact on temperature,

to 1, 2 and 3, respectively, with 1, 2 and 3, respectively, to taking abstraction, however, at the conf place and the Stour, Wantsum River ~1km of Plucks Gutter. upstream identical Scenarios 7, 8 and 9 are 1, 2 and 3, respectively, to with abstraction, however, taking just downstream place ef where Ferry, of Grove STWs. Herne Bay from enters Overview results scenario of runs of the scenario results The assessed against relevant were quality standards, environmental in particular on the the focusing (WFD), Directive Framework Water the (FFD), Directive Fish Freshwater Abstraction Directive Water Surface Agency and the Environment (SWAD) It General Quality (GQA). Assessment that environmental should be noted generally apply quality standards specif a to the FFD or for annual record (e.g. a three-year sampling cycle for while the model results GQA), a 5-month- represent discussed here long time-series. Environmental used hence were quality standards context, into put the results to rather than assess compliance. the model results In summary, be no would that there suggested signif or ammonia, nitrate BOD, salinity, of the preliminaryDO under any A abstraction considered. scenarios in increase slight, but insignifcant, 2.2% (up to concentrations average and the whole river when considering the 2.8 % when considering up to of channel downstream main river the abstraction point) was simulated ortho-phosphatefor and this was with the under the scenarios greatest highest maximum abstraction rate. • the same run for Scenarios were period as the calibration, so as a range of f include to of the frst six days Results from the from excluded model runs were that the model ensure to analysis had stabilised and the impact of was minimal. initial conditions erent erent 85 . following The 2 . for has allowed This 6,10

Scenario 1 represents abstraction Scenario 1 represents under a 180 at Plucks Gutter Minimum-200 Ml/day Residual (MRF) with restrictions Flow of levels based on permitted Water abstraction under the Directive. Framework abstraction Scenario 2 represents under a seasonal at Plucks Gutter (160 - 220 Ml/day). MRF regime a Scenario 3 represents of abstractionsmaller volume under a at Plucks Gutter MRF. 160 220 Ml/day identical Scenarios 4, 5 and 6 are parameter in the dissolved oxygen oxygen in the dissolved parameter equation Scenario runs Description of scenarios calibration of the model, After carried preliminary runs were scenario make an assessment of water out to quality impacts under dif of abstraction the River scenarios from is based on a scenario Each Stour. modelling resource water separate investigation selected: were nine scenarios • • • • changes in the size of phytoplankton of phytoplankton changes in the size population and subsequent impacts be to levels oxygen on dissolved modelled within the same model of success The 5. run, see Figure in the DO calibration was hindered part boundary by data limitations, 5, and further is work see Figure improve to needed in the future the ft with observed data further. t ect the prediction cantly improved improved cantly

ed template template ed cant improvement improvement cant Water quality modelling as a tool for assessing new new assessing quality for a tool as modelling Water options: management resource water Kent the River Stour, case of The not expected af to of impacts channel. main river in the model these limitations, Despite a very good f simulations showed water with observed all seven data for The quality considered. determinands deemed therefore, calibration was, in producing been successful have to Stour. model of the Lower a robust 1 summarises the calibration Table to noting key points relevant results, the calibration of each determinand. plots are example calibration Two 3 (temperature) included as Figure 4 (ortho-phosphate). and Figure Our understanding of a number of in occurring qualitywater processes was signif the river in the DO simulation, compared applications of the previous to used Mike template 11 Ecolab modif The here. included a chlorophyll produced Other issues raised during process the calibration For the calibration process. through the simulations suggested example, a cyclic in nutrient diurnal pattern Closer examination concentrations. that showed of modelling results inputs the combined this was due to STWs up several swept being from with the diurnal the river and down On a of water. tidal movement of concentrations time scale, larger nutrients at the calibration location peak to appeared (Plucks Gutter) suggesting during spring tide events, of nutrients loads again movement 4. see Figure the tide, by controlled the model calibration Additionally, a signif led to Figure 6 - Longitudinal profi le of average ortho-phosphate concentration Longitudinal profi 6 - Figure river channel. under base run conditions and scenarios 1-3 in the main WATER & ENVIRONMENT 59 10. 9. 8. 7. 6. 5. 4. 3. 2. 1. References statistics from the modelresults. in order to ef tool wasdeveloped for thisstudy 11programme.Mike Abespoke are readily accessible withinthe only maximumandminimumresults scenarios 1,2and3.Note that river channel, for thebaselineand concentrations alongthemain variation inaverage ortho-phosphate 6. shown here asanexample, seeFigure is scenario results for ortho-phosphate standards. Aplotofsomethe againstthose current performance was notlarge enoughto af scenarios under theabstraction result, theincrease inconcentrations even underbaselineconditions. As a environmental standards, quality generally elevated to withrespect concentrations intheStour are nutrient Nevertheless, two STWs. reduced dilutionofdischarges from of PlucksGutter, where there was approximately downstream 4km concentrations become noticeable in average ortho-phosphate These smallsimulated changes This illustrates thelongitudinal DHI, 2007b. Ecolab, WQ templates, Scientif Surface Water-Quality Modelling. InternationalEditions. McGraw-Hill Chapra, S.C.,1997. East Atkins, 2008a.South Water Ltd., Strategic Water Resources Investigation. Lower Stour and model theory anddevelopment. model theory The Scienceofthe Total Environment: 194/195:447-456. Whitehead, P.G., Williams, R.J., Lewis, D.R., simulationalongriver 1997.Quality systems (QUASAR): DHI, 2007c. Ecolab, User Guide. 11.Amodellingsystem Mike for RiversandChannels. User Guide. DHI, 2007a. for theRiver Thames. PhD Thesis, University ofReading. Cox, modellingofdissolved B.A, oxygen. 2002.Dynamic Acasestudy for v6,Draft ClientReview.Draft (5044447/77DG548) Atkins, 2008d. East South Water Ltd, Report_ Model Broad Oak Reservoir for Draft ClientReview.Calibration Report (5044447/77DG436) Atkins, 2008c. East South Water Ltd ,Sarre Penn River Water Model Quality size,of Broad to OakReservoir Memo JonHolland(5044447/72DG429). Atkins, 2008b. runsonDemandProf ResultsofMISERmodelling&Sensitivity les for Optimisation forSarre Draft ClientReview. Penn Report, Hydraulic Modelling (5044447/7502DG395). ciently extract other ect ect Water modellingasatool for quality assessing new subsequent Environmental Statement. Environmental Assessment and Impact information to feed into an approach wasdesigned to provide for thecatchment. The modelling impacts assessment ofwater quality paper to provide acomprehensive River Stour modeldescribedinthis will beusedincombination withthe bybeen constructed Atkins. These to form thereservoir the river proposed to beimpounded models ofaproposed reservoir Stour.points ontheRiver Separate Oak andcreation ofnewabstraction atBroad ofareservoir construction scenarios. This includesthepotential of water resources optionsand as ameansoftesting awiderrange model could beusedinthefuture Water. isanticipated thatthe It available for supplyby East South increasing theamountofwater appraise asuite ofoptionsfor Water Resource Investigation to ofaStrategic has beenusedaspart The Stourmodel River water quality Conclusion c Description. 86 water resource managementoptions: 3 have already The caseoftheRiverStour, Kent 4 and

High accuracy recording for heritage 60 applications - Dover Castle graffiti

Cory D Hope Abstract Principal The 21st century has seen the adoption of digital technologies as the Photogrammetrist standard equipment and methodologies when undertaking close range Water & Environment Photogrammetry for the recording of Cultural Heritage. The use of calibrated digital SLR cameras, REDM total stations and softcopy Photogrammetry has reduced capture time of source imagery and photo control, improved consistency of stereo images, enabled alternative mensuration methods to the traditional individual stereo pair to be adopted and enabled very high accuracies to be achieved for reasonable cost. This paper will describe the method and application of close range digital photogrammetric techniques in order to enable millimetre accuracy measurement and 3D visualisation of historic graffiti at Dover castle. The project was undertaken by Atkins Geospatial on behalf of the English Heritage Metric Survey Team.

Introduction In recent decades there have These methods have proved very photogrammetric surveys for English been many fine examples of successful and the methodology Heritage, Historic Royal Palaces, photogrammetric recording of ancient utilised in this project follows on from CADW and many other institutions rock art, such as carvings, petroglyphs this valuable work. The key difference and conservation architects. and pictographs and the like utilising in this project is that measurement The survey formed part of an ongoing various methods to varying degrees of accuracy not low cost is the key factor English Heritage research project accuracy. One of the earliest examples in determining methodology and at Dover Castle focusing principally was work undertaken to record stone equipment selection. This paper will on the Great Tower (Keep) and carvings at Stonehenge (Atkinson describe the high accuracies achieved Inner Bailey that is contributing to 1968). There have been many using a fully digital workflow for

the current representation works W examples following this undertaken very close range Photogrammetry. within this area of the site. Although all over the world on a variety of

Note: The word accuracy is used in the work commissioned by English ATER & ENVIRONMENT subjects using varying methodologies. this document in general terms and Heritage was limited to just the A description and discussion of relates to the combined precision provision of the stereo photography the variety of examples is a rather and accuracy of any measurements and photo control, Atkins Geospatial large task in itself and hence falls made and statistical results achieved. has undertaken further processing outside of the scope of this paper. of the raw survey data themselves to The results achieved to date The project confirm and describe the accuracies have helped to drive the use achievable using this methodology. and development of close range Atkins Geospatial was commissioned photogrammetry in Heritage and by the English Heritage Metric Survey The subject Archaeological recording. It follows team to undertake a ‘close up’ on that in today’s well established photogrammetric survey of 7 areas The subject of the survey is a variety digital age of Photogrammetry, of historic graffiti at Dover castle. The of inscribed elements (graffiti) on the acceptable accuracies are achievable specification included for the provision internal walls of the Keep at Dover for a variety of subjects, large or of ‘close up’ stereo photography at Castle. The graffiti is of varying age small for reasonable or low cost. a maximum taking distance of 1m and style with much of it dating from the subject and photo control back to the 1700s when the Keep Due to the typically limited funding to be measured to an accuracy of was used as a prison. Some of the available for this type of recording, +/-0.001m. The photogrammetric graffiti is carved relatively deeply there has been a focus on finding the survey was undertaken by Cory into the fabric of the wall (up to most cost effective photogrammetric Hope in November 2008 who a couple of centimetres), which solutions (Chandler and Fryer 2005, has over 10 years experience typically consists of names and Bryan and Chandler 2008). of undertaking close range dates but there are also some areas 87 60 High accuracy recording for heritage applications - Dover Castle graffiti

Figure 1 - Image sample of historic graffi ti at Dover Castle where the graffiti consists of large areas of lightly inscribed patterns and shapes. The total coverage of this initial photogrammetric survey includes 7 separate areas totalling in approximately 70 square metres of historic graffiti.

Project planning Figure 2 - Typical layout of the stereo photography The decision making process as model, utilising the traditional • Small adhesive stickers to be to the most appropriate method image configuration would result used as photo control. and equipment for undertaking in a very large amount of individual the survey was influenced by the stereo pairs. In order to minimise the Placing photo control grid following: (Luhmann et al., 2006) amount of photography required The design of the photo control to cover the subject in stereo, it • The standards required for point layout consists of a roughly was decided that the best option stereo photography as described regular grid of points covering the would be to undertake the stereo in the English Heritage Metric subject area. Sufficient points where photography using an image Survey Specification for Cultural placed such that for each stereo configuration similar to that used for Heritage (Bryan et al., 2008) model there where 4 control points an aerial photogrammetric project. • The required accuracy placed within each stereo model and photo scale of survey The result is a series of horizontal hence satisfying the English Heritage strips of photography with an T products as required by the Metric Survey Specification. approximate side lap of 60% project specification Due to the requirement that the and an end lap of 20-30%. • The nature and location stereo photography was taken at 1 of the subject This image configuration enables metre from the subject, the estimated a block triangulation bundle • The environmental conditions Ground Sampling Distance (GSD), adjustment to be undertaken in order around the subject or the estimated pixel size in reality NMEN to assess and prove the potential is 0.0003m. This small pixel size in • The available equipment measurement accuracies achievable reality offers the potential for very O held by Atkins Geospatial. from the stereo photography. fine measurement, so the photo control points needed to be very small Methodology Equipment to enable accurate measurement The nature of the subject is roughly • Kodak DCS Pro SLR/n with of the photo control points from planar i.e. the subject consists of 28mm lens calibrated by the digital images. The method of inscribed elements on the internal Dr S Robson UCL marking the photo control was to ENVIR use easily available circular sticky walls of the building. The traditional • Tripod with Manfrotto & labels of 8mm diameter. Red and image configuration for close range head with level bubbles stereo photography is individual orange colours where chosen as • Leica TCR703 with stereo models which is the general they would clash with the subject Leica traverse kit requirement of the English Heritage surface making each point easily Metric Survey specification. As • 2x Bowens Esprit 1500W flash identifiable. A small pen mark was

ATER ATER most of the areas to be covered are units with soft boxes attached made in the centre of each label to much larger than a single stereo • Gretag Macbeth colour chart act as the point of measurement. W 88 High accuracy recording for heritage applications - Dover Castle graffiti 60

There are always conservation issues of the resultant images was good when placing photo control targets and only minor colour adjustments and permanent monitoring points were required and in some cases no to any historic fabric. In this case colour adjustment was necessary. the Ancient Monument Inspector Traditional record photography of Judith Roebuck was consulted about carved stone work such as this uses the proposals and once agreement the lighting method of raking light. had been sought the stainless Photography using this method steel markers were located by the typically contains high contrast Properties Curator Roy Porter. The shadows. This helps to highlight stickers used as temporary photo the carved features for visual control points where chosen as no interpretation but is not suitable for glue would be required for fixing and Figure 3 - Example of photo control point ‘A20’. photogrammetric processing as the they are easily removable without shadow areas lack detail and pixel damaging the historic fabric. correlation in the shadows is uncertain or in the worst cases, not possible. Acquisition of stereo photography The exposure setting for all Once the grid of photo control photography was f22 at 1/125 points was placed, the stereo second. This enabled a depth of photography was undertaken. field range of 0.8m to infinity to The two main challenges for this be achieved with the camera lens part of the project were achieving focus locked at approximately 2m. consistent camera position and Sufficient tests where undertaken even lighting of the subject. prior to the survey to ensure that the As each project area is reasonably lens configuration described above small, the lighting was setup such would produce the desired results that the entire subject area could without the need to adjust the focus be photographed under the same on the lens and hence perform a lighting conditions. Wherever possible new camera/lens calibration. the two flash units where placed Where circumstances allowed, the such that all of the subject area was distance of the camera from the covered with as near as possible an subject is maintained by placing a even distribution of light and minimal range pole on the ground at the shadows. Setting the lighting this way taking distance of 1m. Prior to each minimised the need to move the flash exposure, the camera is mounted Figure 4 - Typical layout of the grid of photo units when undertaking the stereo on the tripod and levelled using the control. The points in red are the photo control photography and in many cases the spirit bubbles on the tripod head. points and the points in black are the steel stereo photography was completed Using the grid lines visible in the monitoring studs for Area 6_1 of the project without the need to move the flash camera viewfinder, the camera is units at all. This resulted in reasonably rotated to near orthogonal to the W To aid in the identification of each consistent colour and image quality subject and the photo is taken. photo control point, both for ATER & ENVIRONMENT across the majority of the project Each strip of photography is taken the photo control survey and for areas which helped to ensure the subsequent office processing, each from left to right, starting with high quality of the products of the the uppermost strip. The camera horizontal line of control was given survey. However, due to the nature a representative letter and each height is adjusted using the tripod of the subject and site limitations for each strip as required. vertical line of photo control was there were some instances where given a representative number. Thus surface reflectance created image For some of the subject areas each photo control point had a variations within some stereo models. the position for taking the stereo grid reference for eg. A3, B2 etc. photography was limited by walls Before undertaking the stereo opposite the subject, proximity Easily available stickers with photography the colour chart was numbers and letters on them to the floor or other unavoidable placed near the centre of the project obstructions. In these instances there where then placed adjacent to area and photographed under the the control points but outside of was no alternative but to undertake same lighting conditions that where to the stereo photography hand held. the area of interest to aid in the be used for the stereo photography. identification of these points. The colour chart was used to check the colour balance of the images during processing and act as a colour reference for any future processing of the photography. Due to the fact that great care was taken in the lighting and camera setup, the colour balance

89 60 High accuracy recording for heritage applications - Dover Castle graffiti

Figure 5 - Lighting position for recording the graffi ti above fi replace Figure 6 - Setting camera position Photo control survey Further processing by The photo control survey was as a check. The results of the english heritage undertaken using the technique survey calculations were very good Following the photogrammetric of intersection including REDM (as confirmed by the results of survey, English Heritage has measurements to each control point. the triangulation and bundle undertaken the production of various Most of the project areas allowed for adjustment) enabling very accurate photogrammetric data products. The all points to be measured from two measuring to be achievable form software package used to process the survey stations. The survey stations the resultant stereo models. stereo photography supplied by Atkins where placed such that the best was Topcon PI 3000. Topcon PI 3000 is possible intersection angles could Triangulation and a relatively low cost software package be achieved within the limitations block adjustment which enables stereo viewing and of the site. There where some can be used to produce 3D models, instances where it was not possible Before submitting the stereo VRML models and Orthophotos. to measure the photo control by photography and photo control The processing of selected areas intersection, in these cases the REDM to English Heritage for further commenced in January 2009 and measurements where relied upon processing, some of the project at the time of writing is ongoing. to calculate the 3D coordinates. areas were triangulated using The images were imported into During the photo control survey steel Intergraph ISDM to check the quality Topcon PI3000, the photo control monitoring studs, placed previously by of the collected data and assess the was observed and stereo models the English Heritage Property Curator, accuracy potential for measurement were created. The next task was were measured. These points are to from the resultant stereo models. to create an overall Digital Terrain be used for future referencing and The control points placed where Model (DTM)of the subject area at monitoring of the subject and were measured and included in the bundle a grid density of 5mm. From this also used as check points for the block adjustment solution and the steel DTM an overall VRML 3D model and triangulation bundle adjustment. monitoring studs where measured an overall orthophoto was created T as check points and hence not The survey observations where included in the bundle adjustment for each subject area. Each of the 7 processed in Atkins in-house survey solution. The summary results of the subject areas covered by the stereo calculation software in order to bundle adjustment for area 6_1 of photography consisted of many produce the 3D coordinates of the the project are given in Table 1. separate instances of historic graffiti. photo control points. The REDM measurements were used to calculate NMEN 3D coordinates independently of

O the values calculated by intersection ENVIR &

Figure 7 - Camera ready Figure 8 - Undertaking photo control survey WATER WATER 90 High accuracy recording for heritage applications - Dover Castle graffiti 60

Table 1 - Bundle adjustment key statistics for Area 6_1 (fi gures given are in metres unless otherwise stated)

Parameter X/Omega Y/Phi Z/Kappa XY RMS Control 0.000295 0.000239 0.000174 0.000268 RMS Check (steel 0.000278 0.000222 0.000586 0.000251 mon. studs) Max Ground Residual 0.000545 0.000505 0.000472 Mean Std Dev Object 0.000092 0.000092 0.000201 Mean Std Dev 0.000426 0.000454 0.000173 Photo Position Mean Std Dev 0.022076 0.021363 0.006284 Photo Attitude Figure 9 - 3D VRML model of an individual stone with graffi ti. Conclusion In order for finer analysis to be The results of the survey and bundle The quality and resolution of the undertaken, individual stones adjustment show that accurate 3D VRML and Orthophoto data with graffiti, were recorded in the millimetre or even sub-millimetre products makes it possible for off site same manner, but at the higher measurement is achievable utilising digital analysis of the graffiti to be resolution of 2mm. So for each a well calibrated semi-metric undertaken in very fine detail without subject area an overall DTM, VRML digital camera, modern total the need for high powered computers and Orthophoto at 5mm resolution stations and a robust technique. or expensive software. This ‘close up’ was created as well as individual Undertaking the stereo photography level of recording has provided a very DTMs VRMLs and Orthophotos in a way which replicates the network accurate and detailed archaeological at 2mm for individual stones. design for an aerial survey creates record of the graffiti at Dover Castle. At the time of writing, the production the opportunity to triangulate stage has generated some 207 each project area as a single block Acknowledgments stereo pairs, from which 979 DTMs adjustment. This can be a very The author would like to thank at 2mm and 979 DTMs at 5mm useful and time saving methodology Paul Bryan of English Heritage for have been generated. A total of 979 when using high end digital contributing some text and editing individual stone VRMLs and 979 photogrammetric workstations where this paper prior to submission, individual stone Orthos at 2mm have many aspects of production can be Steve Tovey of English Heritage for been created and 18 overall area automated, such as automatic DEM supplying information and images VRMLs and 18 overall area Orthos creation and orthophoto production. from the production stage of the at 5mm have also been created. It also means that once triangulated, project, Don Martindale for processing there will be seamless joins between the survey data and Dr Stuart Robson individual models making the output

for undertaking the calibration of W from stereo plotting more reliable. the camera/lens combination using

his calibration field at UCL and his ATER & ENVIRONMENT camera calibration software.

Figure 10 - Orthophoto of individual stone with graffi ti. 91 60 High accuracy recording for heritage applications - Dover Castle graffiti

References 1. ATKINSON, K.B., 1968. The recording of some prehistoric carvings at Stonehenge. Photogrammetric Record 6(31): 24-31. 2. CHANDLER, J. H., FRYER, J. G. and KNIEST, H. T., 2005a. Non-invasive three-dimensional recording of Aboriginal rock art using cost-effective digital photogrammetry. Rock Art Research, 22(2): 119-130. 3. CHANDLER, J.H. and FRYER, J.G., 2005. Recording aboriginal rock art using cheap digital cameras and digital photogrammetry. CIPA XX International Symposium, Torino. Pages: 193-8. 4. CHANDLER, J.H., BRYAN, P. and FRYER, J.G., 2007. The development and application of a simple methodology for recording rock art using consumer-grade digital cameras, The Photogrammetric Record, 22(117): 10-21. 5. LUHMANN, T., ROBSON, S., KYLE, S.,HARLEY, I. 2006. Chapter 7, Measurement concepts and solutions in practice. Close Range Photogrammetry, Principles, Methods and applications, page 441-442, ISBN 1-870325-50-8. 6. BRYAN, P. G., BLAKE, B., BEDFORD, J., ANDREWS, D., 2008. Metric Survey Specifications for Cultural Heritage. English Heritage. York, 197 pages. T NMEN O ENVIR & WATER WATER 92 Conjugate heat transfer study of a spin pit rig: 61 Application to the lifing of HP turbine disc firtrees

Naveen Prasad Abstract Gopinathrao Structural life and durability of gas turbine engine discs and blade firtree have always been key design issues for designers and technologists. This is Engineer II particularly important for gas turbine components which operate at elevated Energy temperatures under high loads. The testing for these components needs to be undertaken at operating conditions, particularly when Low Cycle Fatigue (LCF) lives of components are evaluated. The spin pit and bi-axial rig Christophe tests, the former of which is the concern of this study, are used in the aero Mabilat gas turbine industry for the lifing analysis of these critical engine parts. Engineer I In this study a three dimensional conjugate heat transfer analysis of Energy the spin pit rig with the HP turbine disc and blade assembly for a large turbofan engine was performed. The scope of the Computational Fluid Dynamics (CFD) analysis has allowed a detailed understanding Dr Sohail of the flow and thermal field within the domain to be obtained. Alizadeh Flow within the rig was found to be complex, with multiple flow mechanisms simultaneously at play in several distinct areas of the solution domain. All Chief Engineer - heat transfer processes, radiation, convection and conduction were found Thermofluid & CFD Technologies to be important in the overall heat. Radiation heat transfer was found Energy to be dominant in terms of temperature levels achieved in components, whilst convection heat transfer greatly influenced temperature gradients Dougal in solids. The appropriate capturing of flow leakages in the rig proved to have a significant impact on rig performance and component thermal Jackson fields. The component temperatures predicted by the tuned CFD model Aerothermal compared well against thermocouple measurements and predictions of Technologist the conjugate CFD analyses were directly used as boundary conditions in Turbines, Rolls-Royce follow-on stress analyses for lifing purposes. The methodology adopted has resulted in a very satisfactory outcome on declared disc life. Dr Rory It is understood that this is the first time where a full conjugate heat Clarkson transfer CFD study of the spin pit rig without any separate thermal analyses, has been applied directly to the lifing of HP disc firtree and Engineering Manager the eventual engine certification, in the aero gas turbine industry. - Thermals Combustion & Casings, Rolls-Royce

Introduction It is understood that this is the first the lack of air/oxidation. This groups of gas turbine compressor time where a full conjugate heat problem is eliminated by operating disks using probabilistic methods. transfer CFD study of the spin pit the spin pit at absolute pressures Predictions were made using rig without any separate thermal of 1/3 of an atmosphere level. finite element (FE) and analytical analyses, has been applied directly Over the years, intense research methods and the predicted failure to the lifing of HP disc firtree and efforts have been devoted to address location was correlated with the the eventual engine certification, this issue both experimentally and LCF tests. Similarly gas turbine disk in the aero gas turbine industry. computationally. Mahorter et al.6 life predictions were carried out by 5 A spin pit is a test rig designed to discussed the use of cyclic spin pit Sattar and Sundt . They emphasised spin turbo-machinery components testing of gas turbine disks as a on analytical stress methods and up to design speeds in a controlled means of predicting the life and life related that to experimental testing A 8 environment to evaluate the structural predictions were made using statistical of the specimen. Hernan et al. integrity of components. Spin pit tests analysis of a limited number of tests. attempted to predict the fretting EROSPAC failure of gas turbine components are typically conducted in vacuum to Advancements in analytical and using FE methods. They focused on reduce the power required to drive computational methods have brought identifying the driving factors for the test article and to reduce hazards significant developments in the fretting failure and 2-dimensional FE when testing at high temperatures. life assessment of turbomachinery analyses was carried out to quantify Under these conditions inserted blade components. Matthew E. Melis and the influence of the key factors on turbines often show shorter life than Erwin V. Zaretsky7 compared two stress and strain. The predictions they do in engine operation. The series of LCF test data reported by correlated well with the experimental source of the problem is accelerated Mahorter et al6 for two different E fretting at the firtree caused by observations on the spin pit tests. 93 61 Conjugate heat transfer study of a spin pit rig: Application to the lifing of HP turbine disc firtrees

Drive shaft The test disc and blade assembly were heated through three banks of heaters. The heater power Top levels were adjusted during the plate Heat test along with the air bleed just Shield sufficient to achieve the required Pyrometers operating conditions representative Test (top) Chamber of engine operations. The rig was Doors instrumented with thermocouples Heater Pyrometers and pyrometers to measure the (bottom) metal and air temperatures in the pit. Heat Figure 2 shows the locations of the Shield Bottom plate thermocouples (M01 –M06) around the test disc and blade assembly. This provided test data to validate CFD Figure 1 - Spin test chamber geometry models at steady state conditions.

Gas turbine components are This also ensured that the final Modelling approach subjected to thermal and mechanical component temperatures exported stresses during their operation from the CFD model benefited Methodology which are transient in nature from the detailed physics, taken The 3D conjugate heat transfer of and these are generally modelled into account in the conjugate the HP turbine spin pit was carried using FE for certain flight cycles. heat transfer analysis, instead of out on a Linux cluster (3.2 GHz, Management of the thermal loading making simplifying assumptions 3GB, Intel Xeon) using Fluent 6.3.26 in these components plays a key in thermal modelling about metal- for running the simulation and role in component lifing and hence fluid boundary conditions. ICEM-CFD for geometry healing, predictions of detailed and accurate manipulation and mesh generation. temperature profiles for disk and Test rig setup Three CFD models were generated blade assemblies are highly important. namely 2D, 3D sector and a full Given the complex geometries Figure 1 and Figure 2 show the 3D model to understand the rig involved and the challenging physics geometry of the spin-test chamber. behaviour as well as to analyse the associated with engine operation, CFD This test chamber is placed inside effect of asymmetric blockage on flow is a good tool of choice for detailed a pit which is maintained at 1/3 field and heat transfer characteristics. component performance assessment. atmospheres and the test disc In recent developments, CFD is and blade assembly was driven used more for deriving appropriate at 13,500rpm from a backing heat transfer boundary conditions disc bolted to the drive shaft. to thermo-mechanical analyses. The test disc was fitted with 66 Approached vary from fully coupled9 dummy blades connected to disc by to loosely coupled11 approaches. In firtree type attachment geometry. further developments, full conjugate The dummies do not resemble the heat transfer CFD applications for real airfoil geometry, but had a bulk detailed disc temperatures predictions mass equal to airfoil mass to exert can be found for turbojet disc cavity10 the equivalent rim load on the disc. and Bi-axial rig enclosure12 studies. In the present study, the 3-D conjugate heat transfer predictions of the spin pit have been used for the lifing assessment of the HP disc firtree for a large turbofan engine. CFD was used to get a better understanding of the overall rig behaviour and heat transfer mechanisms. Moreover it was intended to tune the CFD CE model to the rig thermocouples measurements and upon satisfactory PA model matching, the detailed disc-blade assembly temperature predictions could be exported and OS used directly for follow-on stress

analysis, circumventing the need Figure 2 - 2D Schematic of Spin pit test rig chamber geometry for separate thermal models. AER 94 Conjugate heat transfer study of a spin pit rig: 61 Application to the lifing of HP turbine disc firtrees

The 3D sector model was 1/66 in size and importantly it captured the blade and firtree geometry accurately. The modelling strategy adopted maintained a level of compromise between accuracy and model size. The solid domains were restricted only to test disc, dummy blades, seal plates and retaining rings which were of prime importance in this study. The flow regime inside the rig is largely dominated by the rotational motion of the blade disc assembly and the entrainment of flow onto the spinning disc, which pumps the flow radially outwards. The Figure 3 Simplifi ed geometry (doors and top plate not shown) temperature field in the domain is reference frames. With this approach Due to a lack of information on the largely impacted by radiation heat the flow around the moving part full operating conditions of the rig, transfer between the heaters and can be modelled as steady state. For setup of the boundary conditions was component surfaces, convection heat the spin pit there are many moving difficult and assumptions had to be transfer between the components parts and stationary surfaces which made on the flow rates and heater and the fluid and the conduction in were not surfaces of revolution. settings. In order to get fastest turn the components. Additionally there Such problems could be modelled around times for the calculations the is significant heat generated through using the Multiple Reference Frames 2D axisymmetric model was used. This the viscous heating generated by the (MRF), Mixing Plane or Sliding Mesh route was taken with the objective of shearing action of the disc. In order approaches. Owing to the limitations gaining a quick understanding of the to predict the thermal environment of the Mixing Plane model and the flow field and thermal behaviour of and heat transfer mechanism at computational cost associated with the model and the sensitivity of the the walls accurately, appropriate the Sliding Mesh Model, it was solution to the operating variables. wall y+ values were maintained, deemed appropriate to use MRF for The amount of leakage present in particularly for the rotating surfaces the present problem. A significant the domain was estimated using and radiation exchange between drawback of using the MRF approach simple heat balance calculations. surfaces were calculated. The is that it does not account for the standard turbulence model with Later a 3D sector model of the rig relative motion of a moving zone with was run with boundary conditions wall functions was employed with respect to adjacent zones, the grid viscous heating option enabled. The imposed from the 2D study, primarily remaining fixed for the computation. to be able to transfer the thermal Discrete Ordinate radiation model In full 3D model because of the above was used to model the radiation. field for follow on FE analysis. The mentioned limitation non-uniform sector model was finely tuned to By default, the equations of fluid thermal gradients are setup in the predict the metal temperatures to an flow and heat transfer are solved solid components. This was overcome accuracy of +/-10K with respect to in a stationary frame of reference. by averaging the temperature thermocouple data. Finally the full 3D However, for cases involving moving field in the solids circumferentially model was run with settings derived zones, it is advantageous to solve using a custom FORTRAN code. from the sector model to investigate the equation of motion in moving the impacts of asymmetries in rig setups on its behaviour and operation. The temperature field in the solids were circumferentially averaged and compared with the sector model predictions. A EROSPAC E

Figure 4 - Mesh-blade and disc assembly 95 61 Conjugate heat transfer study of a spin pit rig: Application to the lifing of HP turbine disc firtrees

Figure 5 - Computational mesh of the 3D sector model Figure 6 - Computational mesh of the full 3D model The 3D sector model calculations were In the MRF implementation the Boundary conditions run on one compute node, whereas calculation domain is divided into the full 3D model was run on 14 sub-domains each of which may The air bleed was defined to be compute nodes over 7 machines. be rotating with respect to the a fixed mass flow rate boundary inertial frame. Hence, the domain with prescribed total temperature Geometry and mesh was divided into two zones namely and turbulence parameters. rotating and stationary, with interface The heaters used to heat the disc Figure 1 shows the three dimensional boundaries separating the zones. and the blade assemblies to engine CAD model of the spin pit test rig. For operating temperatures were CFD modelling, minor modifications Different cell generation techniques defined to be a constant heat flux and simplifications were applied to were used to keep the overall cell boundary with a prescribed output. the CAD geometry, resulting in the count manageable. Moreover a domain presented in Figure 3. The significant degree of modularity was Although the test chamber was simplifications adopted are as follows: used for ease of model handling and well insulated, gaps existed in the manipulation. This approach allowed domain. This was clear from the • The extent of the CFD the meshes to be generated separately digital photographs of the rig. Some domain was limited to test and any potential change in the of the key gaps through which flow chamber assembly only geometry could be easily replicated. can either leave or enter the domain • The turbine disc drum, top The mesh generated had the highest were those between the top plate and bottom lids and vacuum degree of cell resolution in the fluid and the drive shaft, annular gap at chamber doors were modelled about the disc and blade assembly the bottom and the gap between as baffles with a thickness for and a reasonable core mesh for the test chamber doors. The test conductive heat transfer the remainder of the domain. This chamber is placed inside a pit which • Removal of features and achieved an appropriate mesh was maintained at 1/3 atmospheres. components within the rig resolution to capture the complex Hence, a fixed pressure boundary with a resolution deemed too flow features around the disc and was applied to these regions with small to resolve explicitly blade assembly whilst keeping the respective backflow temperatures • Cooling coils were not number of computational cells and turbulence parameters. In the full 3D model the gaps between the CE modelled explicitly. within reasonable and practical bounds. Cell counts for each of the test chamber doors were discrete meshes are summarised in Table 1. as in the test rig (see Figure 8), PA whereas in the 3D sector model it Table 1 - Cell counts was approximated by defining the upper portion of the test chamber OS Mesh Cell Count door. Cyclic boundary conditions were 3D Sector Model 300x103 applied to the angular extents (Sides coloured in green, Figure 5) of the Full 3D Model 20x106

AER domain for the 3D sector model. 96 Conjugate heat transfer study of a spin pit rig: 61 Application to the lifing of HP turbine disc firtrees

Table 2 - Boundary conditions Hence the solution domain included both fluid and solids: Zone Boundary condition • The fluid inside the solution Air Bleed Mass Flow Inlet domain was defined as air. Top Leakage Pressure Outlet Density was varied as a function of temperature and pressure Bottom Leakage Pressure Outlet using the ideal gas law. Leakage through test chamber doors Pressure Outlet Whereas specific heat capacity, dynamic viscosity and thermal- Dummy Blades Coupled conductivity were varied as a Blade-Disc Firtree Coupled, ε = 0.9 function of temperature using the piecewise-linear option. Disc surface Coupled, ε = 0.9 • The solids inside the solution Heater assembly Constant Heat flux domain were modelled based All other rotating surfaces Adiabatic, ε = 0.5 on their respective materials. Blades, HPT disc, seal plates and Top & Bottom plates Constant Heat flux retaining ring conductivities and Test chamber doors Constant Heat flux specific heats were modelled using data provided by Rolls-Royce. All other stationary surfaces Adiabatic, ε = 0.5 • The small gaps between As conjugate heat transfer is of specified. Likewise, all walls that are solid components were importance in the present study a stationary in the moving frame of modelled as solid-air with wall will have a fluid or solid region reference are given a velocity of zero thermal conductivity of air. on each side. In the present case the in the absolute reference frame. Convergence two sides of the wall were coupled Most surfaces away from the blade- hence no additional thermal boundary disc assembly, where conjugate heat In order to see the evolution of conditions were required. However, transfer was not of interest were the thermal field and also to appropriate material type, wall considered adiabatic. Wall regions like tune the CFD model against the thickness and radiation properties test chamber doors, top and bottom temperatures obtained from thermal were specified. The disc and blade lid of the test chamber were applied a survey, monitoring points were assembly was coated with constant constant heat flux boundary condition defined at the exact locations of emisivity paint with a value of 0.9. based on simple heat transfer the thermocouples around the In the Multiple Reference Frames correlations to account for the heat HPT disc and blade assembly. (MRF) model used in the present loss through these boundaries. The assessment of run convergence case definitions of velocity boundary All surfaces were also assumed was based on the magnitude and conditions at the walls are important. to be machined with a significant histories of the global residuals. For each fluid cell zone in the degree of finishing, allowing The levels of convergence of domain, the angular velocity (rpm) the surface roughness to be residuals were of the order of and the axis about which it rotates specified using the default smooth 10-3 for continuity, 10-5 for the is specified and wall velocity is roughness coefficient of 0.5. momentum, 10-6 for the energy made relative to the velocity of the equation and 10−6 for the radiation. adjacent cell zone. Hence, all solid Material properties Convergence was also based on walls that are moving are given the the evolution of the temperature speed of the moving frame and a The analysis included 2D and 3D monitors/histories. The steady relative angular velocity of zero was conjugate heat transfer calculations. A EROSPAC E

Figure 7 - Contours of y+ (disc & blade assembly) Figure 8 - Gap between the test chamber doors 97 61 Conjugate heat transfer study of a spin pit rig: Application to the lifing of HP turbine disc firtrees

here as it only served to provide a rapid understanding of the rig setup. In each case the results have been illustrated using contours and vector plots of velocity and temperature. The illustrated flow field within the spin pit rig corresponds to the results obtained at the end of the tuning process. Flow features The large-scale flow patterns inside the domain can be identified from the velocity vector and contour plots of the computations. The main flow features were observed to be the following: • The disc performs work on the fluid. The intense sheering at the faces of the disc contributes to Figure 9 - Thermocouple data for steady-state test the heat rise in the fluid through solution was thought to be converged of +/-10K. Along with the rotational viscous dissipation of turbulence when the residuals were as low as speed, heater power settings were • There is a clear evidence of indicated above and the monitor also measured during the test. The radial pumping of the flow mean temperatures stable. spin pit test was run for a variety of over the rotor surface. The different running conditions and it stationary enclosure walls around Spin pit tests included the steady state run. The the disk and blade assembly steady state test was run at 9000rpm promote recirculation fields The test data 2 used for the validation for approximately 2 hours and then above and below the blades of the CFD model was a steady for 15 minutes at 13,500rpm. During • In both the sector and the full 3D state spin pit test performed on the test the top sets of heaters model the jet off the blade is at an HP turbine disc-blade assembly were operating at 100%, while the an inclination (Figure 10 location by Rolls-Royce. This test had been bottom ones were operating at 33% 4). This is due to the geometrical scheduled to demonstrate the life of the maximum. Figure 9 shows feature shown in Figure 12 of the disc firtree attachments. The the thermocouple data obtained (the line joining the two blades test was conducted to optimise at the end of the spin pit test. being slightly at an angle). With the thermal gradients within rotation, this 3D geometrical the HP turbine disc and achieve Results and discussion feature exerts an upwards force engine operating conditions. on the fluid, bringing about This section of the paper contains The instrumented test rig provided the observed jet inclination. metal and air temperatures in the results from the 3D sector model and pit. All thermocouples recorded the full 3D model. Results from the metal temperatures to an accuracy initial 2D model are not presented CE PA OS

Figure 10 - Vectors of velocity magnitude (m/s) (full 3D) AER 98 Conjugate heat transfer study of a spin pit rig: 61 Application to the lifing of HP turbine disc firtrees

Temperature (C) • Cool inflow from the top pressure boundary (Figure 10 location 3) is largely entrained by the rotating disk surfaces • Figure 10 shows the cool air entering from the air bleed at the bore (location 1). The flow entrained from the bottom pressure boundary mixes with the flow coming from the bore bleed at the bottom of the stub shaft (location 2) and leaks through the small gap between the stub shaft and the small enclosure at the bottom. Figure 12 - Direction of force exerted on the fl uid This flow mixes with the flow entrained and pumped on the underside of the disc (location 2a) • Significant difference can be noticed in the flow separating from the tip of the blade. In the full 3D model the flow leaves with less of an incline and impinges on the side set of heaters whereas, in the sector model the flow escapes through the gap between the top and side heaters (location 4, Figure 10 & Figure 11). It is thought that the difference is due to the restriction of the sector model domain to one blade (1/66). • The variations in the flow field due to asymmetric blockages were captured in the full 3D model. The impact of such blockages on the flow field came from the heater components and pyrometers mainly. Due to the frozen rotor assumption these effects were not transported around the domain and averaging had to be Figure 13 - Contours of velocity magnitude (m/s) (3D Sector) performed using a custom code • Comparison of flow features inside the disc drum and bore regions are very similar and the flow is mainly rotational (Figure 13 and Figure 14). Thermal field

Figure 15 to Figure 17 show a A

selection of temperature plots through EROSPAC the domains for the sector and the full models. The hot fluid region to the left of the plot is completely separated from the other zones due to the presence of the heat shields. Because of this the transfer of heat to the disc is primarily via conduction Figure 14 - Contours of velocity magnitude (m/s) (Full 3D) through from the dummy blades. E

99 61 Conjugate heat transfer study of a spin pit rig: Application to the lifing of HP turbine disc firtrees

This sets up a top to bottom temperature gradient across the disc as demonstrated in Figure 15 and Figure 16. It is important to note that the cooling flow entrained through the top boundary bathes the top surfaces of the disc and blade, explaining the necessity for the high heater settings and the closer position of the top heaters. It is clear that for large parts of the disc blade assembly surfaces, the temperature of the fluid close to the solids is cooler when compared to the metal temperatures. This shows a clear evidence of radiation impact on the thermal field within the domain. The radiation was observed to be the main driver of the heating whilst convection heat transfer impacts top to bottom temperature gradients. Figure 17(a) and (b) illustrates the results from the full 3D model before circumferential averaging. Figure 17(b) shows the temperature distribution opposite to the heater Figure 15 - Contours of temperature overlaid by velocity vectors side. The temperature distribution coloured by temperature (OC) (3D Sector) in the disc and blade assembly is cooler and more inclined as the influence of top to bottom variation in convection heat transfer on this side in the absence of radiation from the heaters becomes more pronounced.

Averaged results Use of the MRF model establishes non uniform thermal gradients in the solids which is not the case in reality. This problem was overcome by circumferentially averaging the temperature field in the solids using a custom written FORTRAN code. It is well appreciated that circumferential averaging is not the same as the rotor receiving effectively a time-averaged heat flux at its wall boundaries, which would be the case in reality. Given the time constraints on this work circumferential averaging of the thermal field was the best option available. The averaged results Figure 16 Contours of temperature (OC) (3D Sector) presented in this section are compared The temperatures in the solid This is mainly due to the fact that against the 3D sector model.

CE components reduce gradually from large amounts of fresh cool air are Identical plots of domains are blades down into the disc and entrained from the top opening, generated from both full 3D and sector models to give a direct PA there is a significant temperature bringing about cooling on the disc difference between the top topside. On the underside of the comparison. The temperature field and the bottom of the disc. disc in the heat shield, there exists from the 3D sector model was

OS a warm recirculation field and copied 66 times for comparison. therefore its impact on the cooling Figure 18(a) and Figure 18(b) show of the disc is less significant there. the temperature distributions within the disc and blade assembly. AER 100 Conjugate heat transfer study of a spin pit rig: 61 Application to the lifing of HP turbine disc firtrees

From the plots it is clear that the circumferentially averaged temperatures within the disc and blade assemblies show generally similar distribution when compared against the 3D sector model. The key findings of the comparisons were the following: • The dummy blades are slightly hotter in the full 3D calculation • The difference in temperature distributions around the (a) Heater side (b) Non-heater side firtree and the disc rim Figure 17 - Contours of temperature (OC) (Full 3D) regions are less significant • The temperature difference between the top and bottom of the disc firtree is slightly higher when compared to the sector model • Temperature differences between the sector model and full 3D model are shown in Figure 19. Significant difference can be seen in the disc diaphragm and bore regions, where the sector model is appreciably hotter than the full model. The temperature gradient for the latter in this region is significantly steeper.

Tuning of the CFD model (a) Sector model (b) Full 3D Figure 18 - Contours of temperature (OC) (Full 3D) There was very little information available concerning the spin pit rig operation conditions. In particular, there was no information on the heater settings and flow rates. At the end of the experiments the maximum power input used for each of the heaters was provided by Rolls Royce. These are shown in Table 3. The 2D model was initially set-up with the estimates of the operating conditions and our best assessment of the radiation properties. A good first assessment of the rig behaviour was obtained from the 2D model. The boundary conditions derived from the initial running of the 2D axisymmetric model were applied to Figure 19 - Temperature differences between sector and full 3D model the 3D sector model. Further heater settings obtained from the tests A

were applied. Upon analysing the EROSPAC Table 3 - Heater power settings thermocouple data it was observed that the temperature on the top Percentage of maximum power applied side of the disc firtree was higher Top 100% compared to bottom. It was observed that there was a significant amount Bottom 33% of cool flow being entrained from the top which was responsible for this feature. This also explained the necessity of high heater settings E on the top set of heaters. 101 61 Conjugate heat transfer study of a spin pit rig: Application to the lifing of HP turbine disc firtrees

Conclusions and recommendations A full 3D conjugate heat transfer analysis of the spin pit rig was performed. The scope of the CFD analysis has enabled a detailed understanding of the flow field inside the domain to be obtained. By conducting a steady-state calculation the variation in flow patterns and temperature distributions inside the domain have been successfully captured. The main findings from the analysis are as follows: • Flow within the rig was found to Figure 20 - Comparisons between tests and CFD be quite complex, with multiple flow mechanisms simultaneously at play in several distinct areas of the solution domain • From both experiments and simulations it was clear that a large amount of flow leakage was present and this had a significant impact on the flow features and the thermal gradients • Radiation heat transfer was found to be dominant in terms of the bulk temperature levels achieved in components, whilst convection heat transfer had a major influence on temperature gradients in solid components

Figure 21 - Difference between CFD and tests • The finely tuned sector It was clear that the leakage flow It is anticipated that this is due to the model predictions matched temperatures (top and bottom) differences in the manner of leakage well with the thermocouple were important drivers and in flow representations. Owing to the readings (+/-10K limit) the overall heat balance of the computational cost associated, the • A full 3D model simulation using assembly and hence the leakage results obtained from the full 3D the setup of the tuned sector flow temperatures were fine tuned model were deemed appropriate and model was run to assess the in order to get the best possible further tuning was not undertaken. impact of blockage and any match to the thermocouples asymmetric flow features present readings. Later, the final settings The steady state MRF attempted obtained from the sector model to capture wakes behind were applied to the full 3D model to blockages such as pyrometers examine impacts of asymmetries. in the domain. However, due For the sector model it can be seen to the frozen rotor assumption that a good match was obtained these effects were static and not for all the thermocouples located transported around the domain around the test piece which was • Comparison of the flow features within the accuracy limits of the and thermal field between thermocouples. The largest difference the sector model and the full is seen in M05 which is at the 3D model showed similar CE bottom of the disc firtree. This features with few exceptions region is in a relatively warm zone • Distributions in the fluid were PA and hence received a large amount less comparable, particularly in of heat. Further tuning was difficult the outer annular enclosure, due as any small variation in settings

OS to the more accurate treatment had impacts on all thermocouples. of the flow leakages in between Comparison between the experiment enclosure door for the full model and the full 3D model is not within

AER the accuracy limit of +/-10K. 102 Conjugate heat transfer study of a spin pit rig: 61 Application to the lifing of HP turbine disc firtrees

• Temperature field from the steady-state MRF simulation was circumferentially averaged using a custom written FORTRAN program. The result from the averaging program was encouraging and showed a reasonable match with the sector model predictions • The circumferentially averaged MRF temperatures used for the full model, carried with them the implicit limitation of MRF approach (frozen rotor assumption) • A recommendation from this study would be to evaluate the changes between circumferential averaging and sliding mesh calculation, even if this is for a test case problem • The asymmetric blockages seemed to have small impacts on the temperature distributions in the solid components • The sector model captured the dominant physics for this task and achieved a large saving on the computational effort required • The study carried out here demonstrates that CFD can be used as a tool for critical parts lifing, accounting for more realistic physics and in combination with testing circumventing the need for additional thermal models. The predicted component temperature fields from the sector model were directly imported into in-house FE package in order to carry out the stress analysis of the HP turbine disc firtree and lifing of this component. The Low Cycle Fatigue (LCF) tests undertaken provided successful and satisfactory firtree life for the HPT of the engine. The results from the eddy current probe checks on the disc and blades showed no cracks. It was understood that the running of the rig was altogether very satisfactory. It is understood that this has been the first instance, together with the A Bi-Axial study12 also undertaken EROSPAC concurrently where CFD conjugate heat transfer analysis predictions have been directly used for lifing of the aero engine critical parts leading to engine certification. Acknowledgments The authors acknowledge Rolls-Royce E for permission to publish this paper.

103 61 Conjugate heat transfer study of a spin pit rig: Application to the lifing of HP turbine disc firtrees

References 1. Andrew Jopson and Dougal Jackson. HP Turbine Disc Firtree Thermal Survey Test. DNS111494. Rolls-Royce, 2006. 2. Dougal Jackson. HP Disc Spin Pit Thermal Survey Test Data Report. DNS132979. Rolls-Royce. June 2007. 3. Naveen Gopinathrao, Christophe Mabilat and Sohail Alizadeh, HP Turbine Disc Firtree – Conjugate Heat Transfer Analysis of Spin Pit Rig, DNS132072, Oct 2007. 4. Christophe Mabilat and Sohail Alizadeh, HP Turbine Disc Firtree – Conjugate Heat Transfer Analysis of the Bi Axial Rig Enclosure, May 2007. 5. Sattar, S. A., Sundt, C. V., Gas turbine engine disk cyclic life prediction, Journal of Aircraft, 0021-8669 vol.12 no.4 (360-365), 1975. 6. Mahorter, R.London, G. Fowler, Life prediction methodology for aircraft gas turbine engine disks, AIAA- 1985-1141 SAE, ASME, and ASEE, Joint Propulsion Conference, 21st, Monterey, CA, July 8-10, 1985. 7. Matthew E. Melis; Erwin V. Zaretsky; Richard August, Probabilistic Analysis of Aircraft Gas Turbine Disk Life and Reliability, Journal of Propulsion and Power, 0748-4658 vol.15 no.5 (658-666), 1999. 8. Hernan V. Arrieta., Patrick Wackers., Ky Dang Van., Andrei Constantinescu., Habibou Maitournam, Modelling attempts to predict Fretting-Fatigue life on turbine components, RTO-MP-AVT-109. June 2004. 9. J A Dixon, J A Verdicchio, D Bentio, A Karl and K M Tham, Recent developments in gas turbine component temperature prediction methods, using computational fluid dynamics and optimization tools, in conjunction with more conventional finite element analysis techniques, Proceedings of the I MECH E Part A Journal of Power and Energy, Volume 218, Number 4, pp. 241-255(15), Aug 2004. 10. Glen Snedden, Thomas Roos and Kavendra Naidoo, Detailed Disc Assembly Temperature Prediction: Comparison between CFD and Simplified Engineering Methods, ISABE-2005-1130, 2005. 11. K Saunders, S Alizadeh, L V Lewis and J Provins, The use of CFD to generate heat transfer boundary conditions for a rotor-stator cavity in a compressor drum thermal model, Proceedings of the GT2007, ASME Turbo Expo Power for Land, Sea and Air, Montreal, May 2007. 12. C Mabilat, S Alizadeh, D Jackson and R Clarkson, Conjugate Heat Transfer Study of a Biaxial Rig: Application to the Lifing of HP Turbine Disc Firtrees, ASME Turbo Expo 2008 Power for Land, Sea and Air, June9-13 Berlin, GT2008-51297. 13. John .H Lienhard IV and John .H Lienhard V, A Heat transfer Textbook, Phlogiston Press, Cambridge, Massachusetts, USA, 3rd Edition, 2007. 14. Frank .P Incopera and David .P .De Witt, Fundamentals of Heat and Mass Transfer, John Wiley and Sons, 3rd Edition, 1990. 15. Fluent 6.3.26 Help Manual.

104 Non-deterministic thermo-fluid analysis of a 62 compressor rotor-stator cavity

Naveen Prasad Abstract Gopinathrao Aero engines are subjected to a number of competing requirements. Small variations in geometry or operating conditions can deteriorate blade tip Engineer II clearance which directly influence engine performance, efficiency and life. Energy Significant effort has been invested in modelling the propagation of input uncertainties in Computational Fluid Dynamics (CFD) simulations using stochastic methods. In this paper we focus on uncertainty propagation in Christophe gas turbine cavity using a Non-Intrusive Polynomial Chaos method. This is an inexpensive method based on the spectral representation of the Mabilat uncertainty parameters. The polynomial coefficients are evaluated using Engineer I Probabilistic Collocation method providing an exponential convergence Energy for arbitrary probability distributions. This method has been applied to a rotor-stator cavity at the front of an IP compressor for a large bypass turbofan aero engine. The application is presented for a number of operating parameters as the uncertainty sources and tested for two Dr Sohail different input probability distributions. The calculations are carried Alizadeh through from the CFD to the thermo-mechanical modelling to obtain non-deterministic component temperatures and blade tip clearances. Chief Engineer - Thermofluid & CFD Technologies Energy

Introduction In an engine design process, Given that the thermal boundary which cannot be solved using existing thermo-mechanical analyses of conditions in compressor drum and CFD codes. This attracted researchers compressor drums and casings are casing components are largely driven to develop an efficient method to undertaken, to predict component by complex flow physics, variations model and propagate uncertainty temperatures and displacements, in physical input parameters and which enables engineers to use their which are ultimately used for material their associated uncertainty can existing CFD codes for uncertainty selection, component lifing and lead to a significant deviation in the analysis. Recently, several papers blade-tip clearance control. The latter, results. Therefore the quantification have appeared in the literature on directly impacting component and of such uncertainty becomes an implementation of PC methods engine efficiencies is a particularly important aspect in the modelling that are non-intrusive in nature. challenging issue, influenced by of the physical processes. Mathelin et al.4 investigated turbulent, both the thermal and mechanical A significant amount of work on compressible flow in a quasi 1D loads of the rotating and stationary uncertainty quantification in CFD nozzle using a non-intrusive method components. Of particular concern applications applied to external called Stochastic Collocation and are the running clearances at engine aerodynamic problems has been demonstrated a significant decrease in high power condition. The blade published.5,6,7,10 However, as yet there computational time when compared centrifugal forces as well as the have not been many publications to intrusive methods. Walters et 5,7 high thermal loading of discs and on uncertainty quantification in al. developed the Non-Intrusive casing can induce undesirable blade/ gas turbine CFD applications. Polynomial Chaos and applied it to segment rubs, with the potential to three basic fluid dynamics problems. In this paper we focus on applying an adversely impact engine performance. The method starts with replacing inexpensive Non-Intrusive Polynomial the uncertain variable of interest The component thermal boundary Chaos (PC) method for uncertainty with their polynomial expansions. conditions are usually obtained from propagation. The PC method is one Then, for a given PC of given order, a a small number of standard flow field among several methods to model and random variable vector is chosen and and heat transfer simulations which propagate uncertainty in stochastic the deterministic code is evaluated are usually undertaken using input CFD simulations. Xiu and Karniadakis2 at these points. The method is parameters which are deterministic in applied a generalized PC method, non-intrusive however, and the nature. In practice, however, they are which is an extension of the original polynomial coefficients estimated are subject to inherent physical variability. PC method first introduced by Wiener1 not unique, since one can choose to incompressible flow simulations the random vectors arbitrarily. around circular cylinder. This method was intrusive, meaning a coupled system of equations are obtained

105 62 Non-deterministic thermo-fluid analysis of a compressor rotor-stator cavity

In 2006, Loeven et al.6 introduced Probabilistic This projection is approximated another non-intrusive method called using Gaussian quadrature, with Probabilistic Collocation for efficient Collocation method optimal collocation points and propagation of arbitrarily distributed The PC method results in a spectral corresponding weights based on the uncertainties. The method combines representation of the uncertainty input distribution. This results in a the idea of PC method and the parameters. An important concept system of equations which is fully Stochastic Collocation approach. of this approach is the division of decoupled. The mean and variance Exponential convergence is obtained uncertain variable into deterministic of the solution are calculated using with respect to the polynomial order and stochastic parts. The polynomial for arbitrary probability distributions chaos expansion of the solution since the optimal collocation points is generally expressed as, are computed based on the input distribution. The probabilistic collocation method computes the polynomial coefficients using Gaussian quadrature which results where, is the determinstic where wi are the weights in a decoupled set of equations, part and is the stochastic corresponding to the making the method non-intrusive. part of the random variable. The collocation points i. total number of expansion terms The efficiency of Probabilistic is P + 1, which is determined by Probabilistic CFD analysis Collocation method was the dimension of the vector of demonstrated on a linear piston of a rotor-stator cavity random variables and the order problem. Furthermore, the non- of of the polynomial . . The non-intrusive polynomial chaos intrusiveness of the method was (Probabilistic Collocation) approach demonstrated by simulating has been applied to a rotor stator steady flow around NACA0012 cavity in front of an IP compressor airfoil with uncertainty in free of a large bypass turbofan aero stream velocity using the FineTM/ In Probabilistic Collocation method engine.8, 9 Figure 1 shows the Hexa solver of Numeca Int. the random trial basis is a arrangement of the IP Compressor In this paper, the Probabilistic Lagrange interpolating polynomial (IPC) stage 1 and the Front Bearing Collocation method is applied to an denoted by . Each variable Housing (FBH) structure at the industrial rotor-stator cavity in front depending on the uncertain input front of a large civil aero engine. parameter is expanded as follows: of an IP compressor of a large bypass Air is drawn forward through the turbofan aero engine with uncertainty compressor drum, and emerges from in the operating conditions. the front stub shaft, via a low radius The deterministic CFD simulations labyrinth seal (RBS), into the large are carried out using FLUENTTM cavity in front of the stage 1 disc The Lagrange interpolating polynomial and thermo-mechanical modelling diaphragm. There it mixes with the is a function in terms of the random using the finite element program cooler air emerging from the large variable . It is a polynomial SC03. The background of the PC holes in the rear panel of the FBH chaos of order that approach including the probabilistic structure. The whole flow exits the passes through the p collocation collocation method is outlined, the cavity at the high radius labyrinth seal points. The collocation points are results obtained from the investigation in the front of the IPC 1 rotor (FAS). chosen such that they correspond to of the rotor stator cavity and the Gaussian quadrature points used to The deterministic CFD simulations assessment of impacts of uncertainty integrate the function in are carried out using the FLUENT on component temperatures and the domain. Computation of CFD code. The model was 2D blade-tip clearances is presented Gaussian quadrature points and axisymmetric, with swirl transport and the conclusions drawn. corresponding weights is detailed equations enabled. The flow in Loeven et al.6 The Probabilistic inside the cavity is modelled by the Collocation method is a generalization Reynolds-Averaged Navier-Stokes of Stochastic Collocation method. equations using the 2-layer, realisable turbulence model. Figure The inner product in the stochastic 2 shows the grid resolution in the space is expressed as, vicinity of the RBS and FAS seals. The

CE mesh contained a structured boundary layer mesh and an unstructured core with a smooth transition in between. PA OS AER 106 Non-deterministic thermo-fluid analysis of a 62 compressor rotor-stator cavity

The RBS and FBH holes were The stochastic rotor-stator The mean, variance and standard defined to be fixed mass flow rate cavity problem was formulated deviation of adiabatic wall boundaries with prescribed values by introducing uncertainties temperature and heat transfer of total temperature, hydraulic through the operating conditions coefficients are computed. The diameter and turbulent intensity. (boundary conditions). The input propagation of uncertainty is The FAS was prescribed using fixed uncertainties considered are also represented by plotting pressure boundary condition. The flow The results obtained from the standard deviation of the properties at the RBS, FBH and FAS Probablisitic Collocation method temperature field in the cavity. are summarised in Table 1 is for Max for the various input uncertainties Figure 3 to Figure 5 illustrates the take Off (MTO) condition. Rotational detailed in Table 2 are illustrated here. standard deviation of the temperature speed of the IPC 1 disc was 7824rpm. A fifth order approximation is field obtained for the various input The surface boundary of the CFD used, with six collocation points uncertainties using beta and uniform cavity was split-up into 10 different for each uncertain parameter. The probability distributions. Furthermore, wall sections as shown in Figure 1. deterministic CFD simulation was the influence of different input performed at each collocation point. uncertainties on mean and standard deviation of wall temperature and heat transfer coefficient is also given. From the standard deviation plots it is clear that, • The inlet (T0)FBH−Holes results in maximum deviations close to walls 1 and 2. Significant deviations are seen in the core due to the entrainment of this flow in the core recirculation (Figure 3). • The influence of inlet m˙ FBH−Holes uncertainty had similar levels of effect on the temperature deviations in the cavity (Figure 4). • In case of multiple uncertain parameters (m˙ RBS and m˙ FBH−Holes) the standard deviations differ from the results obtained from the uncertainty parameters considered individually (Figure 5). • The influence of inlet (T0)RBS and m˙ RBS uncertainties had a weak effect on the temperature deviations in the cavity. • In all the cases it was observed that the uniform probability Figure 1 - IPC 1 and FBH Cavity A EROSPAC

(a) Rear buffer seal (RBS) (b) Front air seal (FAS) E Figure 2 - Mesh resolution 107 62 Non-deterministic thermo-fluid analysis of a compressor rotor-stator cavity

Table 1 - Infl ow and outfl ow boundary conditions

Property Value m˙ (kg/s) 0.3 T0(K) 578.9 RBS Whirl Fraction 0.5 Hydraulic Diameter (mm) 0.8 Turbulent Intensity 10% m˙ (kg/s) 0.327 T0(K) 483.0 FBH Holes Whirl Fraction 0.0 Hydraulic Diameter (mm) 37.0 Turbulent Intensity 10% P(kPa) 401.5 T0(K) 522.9 FAS Hydraulic Diameter (mm) 0.2 Turbulent Intensity 10%

Table 2 - Uncertain parameters

Beta PDF Uniform PDF Extent Most Likely Mean (μ) SD () RBS Inlet Total [558.9, 598.9] 578.9 578.9 11.5 Temperature (K) FBH-Holes Inlet Total [433.0, 533.0] 483.0 483.0 28.8 Temperature (K) RBS Inlet m˙ (kg/s) [0.06, 0.36] 0.3 0.210 0.08 FBH-Holes Inlet [0.06, 0.39] 0.33 0.230 0.094 m˙ (kg/s) RBS and FBH * See 3rd and 4th rows respectively. Inlet m˙ (kg/s)

distribution brought about larger For other cases the deviations in the Probabilistic thermo- deviations to the temperatures wall temperatures were observed to in the cavity compared to be very modest. According to Lewis mechanical analysis of the beta distribution. et al.3 experience a 15K change in a rotor-stator cavity the cavity temperature can require The uncertainty on wall temperature The thermo-mechanical modelling a change in the material from and heat transfer coefficient due to was carried out using the Rolls-Royce which the component is made. the various uncertain parameters proprietary finite element analysis are illustrated in the charts of Figure Comparison of mean heat transfer code SC03. The code is capable of 6 and 7. Mean wall temperatures coefficients obtained for the different performing fully transient analysis obtained for uncertain parameters input uncertain parameters showed of the engine running conditions. (T0)RBS and (T0)FBH−Holes were no significant difference hence are The analysis program has the almost identical to deterministic not presented. The influence of capability to represent a variety of values and hence not presented. different uncertain parameters on thermal and mechanical boundary However, m˙ FBHHoles and multiple wall heat transfer coefficients is CE conditions found in gas turbine uncertain parameters with uniform clearly evident from the standard type applications. The capability distribution had significant influence deviations plots in Figure 7. to define the cooling flows to PA on the mean wall temperatures. represent the heat exchange is of Comparison of standard deviations particular importance. The thermal in Figure 6(b) shows significant

OS boundary condition for example deviations about the mean for (T0) contains mass flow, flow direction, FBH−Holes, m˙ FBH−Holes and fluid temperature, heat transfer multiple uncertain parameters with coefficient and pressure. Heat transfer uniform input probability distribution. AER 108 Non-deterministic thermo-fluid analysis of a 62 compressor rotor-stator cavity

Figure 3 - Standard deviation of the temperature fi eld (K) - Uncertain inlet

Figure 4 - Standard deviation of the temperature fi eld (K) - Uncertain inlet m˙ FBH−Holes. A EROSPAC E

Figure 5 - Standard deviation of the temperature fi eld (K) - Uncertain inlet m˙ RBS and m˙ FBH−Holes.

109 62 Non-deterministic thermo-fluid analysis of a compressor rotor-stator cavity

590

580

570

(K) 560 mean )

wall 550 (T

540

530

520 1 2 3 4 5 6 7 8 9 10 Walls

(m)FBH Beta (m)RBS Beta (m)2 Var Beta Deterministic (m)FBH Unif (m)RBS Unif (m)2 Var Unif

(a) Mean wall temperature

25

20

15 (K) σ ) wall

(T 10

5

0 1 2 3 4 5 6 7 8 9 10 Walls

(T0)RBS Beta (T0)FBH Unif (m)RBS Beta (m)2 Var Unif (T0)RBS Unif (m)FBH Beta (m)RBS Unif (T0)FBH Beta (m)FBH Unif (m)2 Var Beta

(b) Standard deviation wall temperature Figure 6 - Infl uence of uncertainties on wall temperatures uncertain parameter with uniform probability distribution. Figure 9(a) coefficients are generally applied using carried out using a stabilised slam shows temperature contours for the standard convective heat transfer accel to MTO and a typical flight fully stabilised MTO case and Figure correlations. CFD is increasingly cycle as shown in Figure 8. The 9(b) at the end of the acceleration being seen as a tool to define outputs from the thermo-mechanical to MTO for a typical flight cycle the set of heat transfer boundary analyses were transient temperatures (Figure 8). Plots clearly show the conditions in complex secondary and radial displacements. propagation of heat into the disc air system cavities. In this study, the The heat transfer results obtained for and blade assembly from the cavity Temperature Influence Coefficient the (T0)FBH−Holes with a uniform fluid. Maximum temperatures and

CE (TIC) method proposed by Lewis and probability distribution exhibited the stresses occur during max take- Provins3 is used for transferring the largest variations. Hence, results from off. At fully stabilised conditions heat transfer information from CFD this case was used to investigate

PA the disc and blade regions are to thermo-mechanical analysis. the influence of input uncertainty significantly hotter compared to end The thermo-mechanical models were on component temperatures of accel due to thermal inertia.

OS 2D axisymmetric and comprised and blade-tip clearances. approximately 9000 six-noded Figure 9 and 10 shows the mean and triangular elements. The transient standard deviation of the temperature thermo-mechanical analyses were field obtained for the (T0)FBH−Holes AER 110 Non-deterministic thermo-fluid analysis of a 62 compressor rotor-stator cavity

30

25

20 K) 2 15 (W/m σ h

10

5

0 1 2 3 4 5 6 Walls

(T0)RBS Beta (T0)FBH Unif (m)RBS Beta (m)2 Var Unif (T0)RBS Unif (m)FBH Beta (m)RBS Unif (T0)FBH Beta (m)FBH Unif (m)2 Var Beta

(a) Standard deviation HTCs, Stator walls

200

150 K) 2 100 (W/m σ h

50

0 7 8 9 10 Walls

(T0)RBS Beta (T0)FBH Unif (m)RBS Beta (m)2 Var Unif (T0)RBS Unif (m)FBH Beta (m)RBS Unif (T0)FBH Beta (m)FBH Unif (m)2 Var Beta

(b) Standard deviation HTCs, Rotor walls Figure 7 - Infl uence of uncertainties on heat transfer coeffi cients A EROSPAC

Figure 8 - Typical fl ight cycle E

111 62 Non-deterministic thermo-fluid analysis of a compressor rotor-stator cavity

Figure 9 - Mean temperature contours (K)

Figure 10 - Standard deviation of the temperature fi eld (K)

0 0 Probabilistic Probabilistic Deterministic Deterministic -0.2

-0.5 -0.4

-0.6

-1 -0.8

-1 Radial Closure (mm) Radial Closure (mm)

-1.5 -1.2

-1.4 CE

-2 -1.6 0 5000 10000 15000 20000 25000 30000 0 5000 10000 15000 20000 25000 30000 Time (s) Time (s) PA

(a) IPC1 Blade Tip Front (b) IPC1 Blade Tip Rear

OS Figure 11 - Comparison of probabilistic and deterministic radial tip closures AER 112 Non-deterministic thermo-fluid analysis of a 62 compressor rotor-stator cavity

0.045 0.025

0.04

0.02 0.035

0.03

0.015 0.025 (mm) (mm)

closure 0.02 closure σ σ 0.01

0.015

0.01 0.005

0.005

0 0 Time (s) Time (s)

(a) IPC1 Blade Tip Front (b) IPC1 Blade Tip Rear Figure 12 - Standard deviation - IPC1 blade radial tip closure Conclusion Acknowledgments It is interesting to note that the A non-deterministic thermo fluid This work was supported by the standard deviations about the mean simulation based on Probabilistic project NODESIM-CFD ”Non- for both the stabilised and flight cycle Collocation method has been applied Determinstic Simulations for CFD- are similar. Temperature deviations to a rotor stator cavity in front of based Design Methodologies” of about 14K and 10K are observed an IP compressor of a large bypass funded by the European Community around the front air seal and disc turbofan aero engine. The impacts on represented by CEC, Research diaphragm regions respectively, wall temperatures and heat transfer, Directorate-General, in the 6th whilst, the lower portion of the static have been assessed with different Framework Program, under Contract panel containing FBH holes shows sources of uncertainty in operating No. AST5-CT-2006-030959. the largest deviations. This latter conditions. The (T0)FBH−Holes The authors would also like to observation is not surprising given and m˙ FBH−Holes as independent thank Rolls-Royce for the provision the proximity of the source of the uncertain parameters showed of the test case application and uncertainty and the jet flow bathing higher deviations. A multivariable operating conditions data. this panel below the FBH holes. calculation with these two parameters Blade tip clearance has been a could yield much higher deviations challenging problem since the due to the combined effect. The development of the gas turbine results obtained from CFD are engine. They can have a significant carried through a non-deterministic influence on the component thermo-mechanical analysis to efficiency, surge margin and assess the impacts on component specific fuel consumption. temperatures and displacements. Significant deviations are observed Better understanding of the variation in key component parameters at the earliest possible stage in presented. The calculations presented the design cycle helps to improve demonstrate how uncertainty sources engine efficiency by optimising in engine operating conditions, both transient and steady-state tip can be managed in computational clearances. The uncertainty on the flow field and component blade tip clearances due to the (T0) thermo-mechanical calculations, FBH−Holes uncertain parameter providing greater confidence in with uniform probability distribution the design and assessment of is illustrated in Figure 11 and 12. component performance. The comparison of mean radial

tip closures with the determinstic A results show a significant difference for the majority of the flight cycle. EROSPAC However, the standard deviations about the mean were observed to be modest. Such calculations can potentially increase the accuracy of tip-clearance calculation, reducing the ”unknowns” and provide greater confidence in the setting of the

cold-build clearances of the engine. E

113 62 Non-deterministic thermo-fluid analysis of a compressor rotor-stator cavity

References 1. Wiener, N., ”The Homogenous Chaos,” Amer. J. Math., Vol. 60, pp. 897-936, 1938. 2. Xiu, D. and Karniadakis, G. E., ”Modelling Uncertainty in Flow Simulations via Generalized Polynomial Chaos,” Journal of Computational Physics, Vol. 187, No. 1, pp. 137-167, May 2003. 3. Lewis, L. V., Provins, J., ”A Non-coupled CFD-FE Procedure to Evaluate Windage and Heat Transfer in Rotor-stator Cavities”, ” Proceedings of the ASME Turbo Expo 2004: Power for Land, Sea and Air, GT2004-53246, 2004. 4. Mathelin, L., Hussaini, M. Y., and Zang, T. A., ”Stochastic approaches to uncertainty quantification in CFD simulations,” Numerical Algorithms, Vol. 38, pp. 209-236, 2005. 5. Hosder, S., Walters, R. W., and Perez, R., ”A Non-Intrusive Polynomial Chaos Method for Uncertainty Propagation in CFD Simulations,” Proceedings of the 44th AIAA Aerospace Sciences Meeting and Exhibit, AIAA paper 2006-891, Reno, January 2006. 6. Loeven, G. J. A., Witteveen, J. A. S., and Bijl, H., ”Probabilistic Collocation: An Efficient Non- Intrusive Approach For Arbitrarily Distributed Parametric Uncertainties,” Proceedings of the 45th AIAA Aerospace Sciences Meeting and Exhibit, AIAA paper 2007-317, Reno, January 2007. 7. Hosder, S., Walters, R. W., and Balch, M., ”Efficient Sampling for Non-Intrusive Polynomial Chaos Applications with Multiple Uncertain Input Variables,” Proceedings of the 48th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference, AIAA paper 2007-1939, Honolulu, April 2007. 8. Saunders, K., Alizadeh, S., Lewis, L. V. and Provins, J., ”The Use of CFD to Generate Heat Transfer Boundary Conditions for a Rotor-Stator Cavity in a Compressor Drum Thermal Model, ” Proceedings of the ASME Turbo Expo 2007: Power for Land, Sea and Air, GT2007-28333, Montreal, May 2007. 9. Naveen Prasad Gopinathrao, Christophe Mabilat, Sohail Alizadeh, ”Use of Polynomial Chaos to Assess the Impact of Uncertainty on Heat transfer in a Gas Turbine Rotor-Stator cavity”, Proceedings of the 10th AIAA Non-Deterministic Appraoches Conference, AIAA paper 2008-2151, IL, April 2008. 10. Loeven, G. J. A., Witteveen, J. A. S., and Bijl, H., ”Airfoil Ananlysis with Uncertain Geometry using the Probabilistic Collocation method”, Proceedings of the 10th AIAA Non- Deterministic Approaches Conference, AIAA paper 2008-2151, IL, April 2008. CE PA OS AER 114 The optimum position for a tidal power 63 barrage in the Severn Estuary

Rod Rainey Abstract Chief Engineer G.I.Taylor’s approximate analytical solution for the tidal flow in the Severn Energy estuary is extended to find the optimum location for a tidal power barrage, from the power point of view. It appears to be at the lowest point in the estuary, between Ilfracombe and Gower - contrary to earlier computations. The analytical solution shows that barrages radiate tidal waves out to sea, which highlights the important role of the far-field boundary condition in absorbing them. This appears to have been neglected in numerical models, which may explain the difference from the earlier results.

Introduction Tidal power barrages in the Severn Taylor’s model is described in Lamb’s In the case of the Severn estuary, estuary were studied intensively 30 account of the “canal theory of the Taylor observed that the width years ago, by a UK Government tides” (Lamb, 1932, pp 267-278), b(x) and depth h(x) both increase committee chaired by Bondi, see of which it is a special case. The approximately linearly with distance x Bondi et al, 1981. It was concluded canal theory considers tidal flow downstream (referred to henceforth from computer models that the as a longitudinal gravity wave in a as “west”) of the head of the estuary optimum position for a barrage channel. Following Lamb’s notation, at Portishead, see Figure 1 (originally from the power point of view was if the width of the channel is b(x) Table 1 and Figure 1 in Taylor, 1921). approximately half way down the and its depth h(x), both varying with He therefore took x = 0 at Portishead. estuary at Minehead. If the barrage position x along the channel, then All the data in the table have been was moved further downstream the equation for the surface elevation updated, from the latest Admiralty no more power was obtained η(x,t) at time t is (Lamb, 1932, p.274): Charts. Breadth, depths and areas because it was found that the are at mean sea level. The area barrage increasingly attenuated upstream of Sharpness (17km the incoming tides. Although tidal (1) upstream of Chepstow) is excluded, power barrages for the Severn where g is the acceleration due to since it is small and the tidal range have been studied on several more gravity. In an estuary, high tide is is markedly reduced there. The time recent occasions, it appears that no assumed to occur at the same time delays are mean values for neap tides more recent computer modelling t=0 everywhere, since the extent (appropriate since we are considering has been undertaken on this of the estuary, when measured a mean tidal range in Figure 3, which point, see Burrows et al. 2009. in degrees of longitude, is small will be reduced by the barrage as in The problem can be investigated compared with the tidal cycle of Figure 4) based on data for the north using G.I.Taylor’s simple analytical approximately 180 degrees. A solution shore of the estuary (which appear model of the tidal flow in the is therefore sought of the form: more reliable than those from the Severn estuary (Taylor, 1921). This south shore) in the 2009 Admiralty has the advantage of revealing the (2) Tide Tables. Parameters for n=8 are ENERG fundamental features of the problem defined to give the correct averages more clearly than a computer model. where 2π/ω is the tidal period of over the area upstream of Section approximately 12 hours (half a G, when used in equation (9) lunar day). Thus (1) becomes:

(3) Y

115 63 The optimum position for a tidal power barrage in the Severn Estuary

We thus reach the important conclusion that Taylor’s solution (or any solution of the form (2)) is inadmissible west of the barrage, because it transmits no mean power. What is required west of the barrage is a progressive wave, in which there is a power flux, because the surface slope is 90 degrees out of phase with the surface elevation (and thus the velocity is in phase with the pressure). Rather than a solution of the form (2) we can seek a solution of the more general form:

(7)

Taylor’s Distance Mean Breadth Area to Delay of high Loss angle where is now complex. This n Sn tn η0(x) Section from depth (km) next Section tide, relative to (degrees) from xn again leads to (5), which can be Section (m) (sq.km) Section A (min) equation (9) solved in the same way as: G (km)

A 1 114.3 36.9 40.6 800 0 3.8 B 2 92.10 28.7 37.7 585 2 3.9 (8) C 3 77.83 24.4 30.0 695 2 5.2 where H1(1) and H1(2) are a first and D 4 46.33 16.3 22.7 383 6 6.1 second Hankel functions of order 1, and we now have two constants K1 E 5 28.72 16.3 13.2 220 14 5.2 and . The first term is a progressive K2 F 6 14.82 9.5 15.2 166 19 wave travelling east, and the second G 7 0 5.3 7.8 113 29 is a progressive wave travelling west. Far to the west, both resemble tidal 8 0 53 waves in open water of the same Figure 1 - Taylor’s model of the Bristol Channel depth (since ~ -{cos ) H1 (x) (x+π/4 ± isin(x+π/4)}/√x, for large x). and (4) Tidal power - the need East of the barrage, we can extend where β,γ are constants. for progressive waves Taylor’s solution empirically to include the observed delay-times This reduces (3) to: Considered as a function of time, of the tide. These are caused by the horizontal velocity in a tidal the need to transport energy into wave (and indeed in a water wave (5) the estuary, to overcome natural generally) is 90 degrees out of energy losses from turbulence, and phase with the surface slope ∂η/∂x, with may therefore be important in the since the latter is in phase with the context of a tidal power barrage horizontal acceleration. And the (in fact they turn out to be of only pressure variations are in phase with which can be solved exactly minor importance, see Figure 3. the surface elevation . Thus for a as a Bessel function: η standing-wave solution of the form (2), where the surface slope is in An equivalent phase with the surface elevation, electric circuit (6) the velocity and pressure are 90 In his account of waves in channels, where is a constant. Taylor took = degrees out of phase. Therefore the K γ Lighthill (1978, p.104) introduces {25 fathoms}/{80 UK nautical miles} = power flux (= velocity × pressure) has the standard electrical analogy of 0.0003084 ( is immaterial) and the a mean value of zero everywhere. β voltage with pressure, and electric tidal period / as 12.4 hours, so This is of course to be expected, 2π ω current with volume flow rate. If the that k = 0.00655km-1, and found (6) since the tidal energy is nowhere level variation of a reservoir of area S to be a good approximation to the being dissipated in the estuary is written Re{eiωt}, then its pressure observed variation of tidal range in in potential flow, only stored. variation is Re{ρgeiωt] (where ρ is the the Severn estuary, shown in Figure 1 When we extract tidal power with density of water), and the volume

Y (which are close to modern values). a barrage, however, we require an flow rate in and out of the reservoir This paper extends Taylor’s equal mean power flux inwards iωt iωt

G is Sd/dt(Re{e }) = Re{Siωe }. analysis to the case of a tidal at the mouth of the estuary. Thus on the electrical analogy power barrage in the estuary. its impedance is ρg/(Siω) so it is analogous to an electrical capacitance S/ρg (Lighthill, 1978, p.200 (3)). ENER 116 The optimum position for a tidal power 63 barrage in the Severn Estuary

A similar calculation applies in our The water acceleration in this wave, Thus the total (complex) volume case, for a reservoir formed by a in the direction of propagation, is flow rate at this location, in an barrage at one of Taylor’s Sections A-E minus the surface slope times g, eastward direction, can be written: in Figure 1. The reservoir area can be whence we can obtain the water can be discretized into the sub-areas velocity in a westward direction by between the successive Sections, integrating, as the real part of: Sn given in Figure 1. The level variation (12) at the barrage is given by Taylor’s If we write the total (complex) formula (6) with his x-coordinate x pressure at this location as P’’ = P + n (10) given in Figure 1, and this formula P’, then (12) can be re-arranged to: can also be used to find the average The volume flow rate in the direction amplitude of the level variations of propagation is this velocity times of each sub-area. The phases of bh, and the water pressure is ρgη. these level variations is given by the We obtain the impedance by Z2 average delay-times in Figure 1. dividing the latter by the former, tn (13) Thus the reservoir impedance Z which gives this impedance as: 1 On the electrical analogy, this the of a barrages at the th of Taylor’s n same current as would be produced Sections A-E can be written as: by a voltage generator with a source impedance P(Z1 + Z2)/Z1 (11) of . The flow in an eastward Z2 which we can consider as a resistance direction produced by this voltage R in series with an inductance L, generator passes first through the barrage, and then into the reservoir (9) giving a combined impedance of R + iωL. For large x, the wave beyond it. The impedance seen by Evidently (9) is no longer purely resembles a tidal wave in open water, the flow is thus the flow-resistance of imaginary, but has a real part for which the impedance is known the turbines in the barrage, in series analogous to a resistance as well RL with the reservoir impedance . The to be purely a resistance of ρc/(bh) Z1 as an imaginary part analogous to a (Lighthill, 1978, p.104), where c is turbines can be taken for simplicity as capacitance C. The resistance R gives L the open-water wave speed √(gh). allowing flow in both directions. This the natural energy dissipation in the This gives a useful cross-check, is the most common arrangement, reservoir - to continue the electrical when (11) is evaluated numerically. see Baker, 1991, p.31, and also the analogy, it can be expressed as a most efficient, before turbine losses, “loss angle” tan-1( ) , which is In the absence of the barrage, the ωCRL see Prandle, 1984. Also for simplicity, (complex) volume flow rate at the readily calculated from the argument the flow-resistance of the turbines barrage location is P/Z , of (9) and is given in Figure 1. 1 can be taken as a constant R , in an eastward direction. The B West of the barrage, it is convenient because very similar results have been additional wave-making volume flow to consider the water pressure obtained in simpler cases with linear immediately west of the barrage variation (= ρg × level variation) as and quadratic turbine characteristics is P’/Z , in a westward direction. the sum of the pressure variation 2 (Garrett & Cummins 2004). Thus the Re{Peiωt} which would be seen in equivalent circuit of the complete the absence of the barrage, and system is as shown in Figure 2. the additional pressure variation When = 0, it may be seen that RB Re{P’eiωt} caused, immediately west the pressure at the barrage is its of it, by the presence of the barrage. undisturbed value P, as it should be. The additional pressure Re{P’eiωt} at the barrage produces a tidal wave Z2 = R + iωL which propagates out to sea - as far as the flow to the west of the R L RB barrage is concerned, the barrage is acting like a wave-maker. We require its wave-making impedance , i.e. Z2 the ratio of pressure to volume flow rate in the tidal wave it generates.

A unit wave propagating west RL is described by the second term in (8), with = 1. ENERG K2

C Y

Figure 2 - Equivalent electric circuit of barrage 117 63 The optimum position for a tidal power barrage in the Severn Estuary

Similarity to wave power At first sight it may seem curious that to provide the inward power flux needed to power the barrage, we introduce an additional tidal wave travelling in an outward direction. The reason is that from (8) Taylor’s standing-wave solution (6) can be seen (by putting = = in (8) K1 K2 K and noting that (1) + (2) = ) H1 H1 2J1 as the superposition of a tidal wave travelling east and an equal one travelling west. Our additional wave travelling west is cancelling part of his, giving a net inward wave. This situation is familiar in wave power, see for example Mei 1989 Section 7.9. Two-dimensional wave power devices likewise need to radiate waves out to sea, to cancel out wave reflections. Power available at Figure 3 - Average power (GW) for barrages at locations A - E of Figure 1 The horizontal axis is the water level difference across the barrage, divided by the tidal amplitude various locations in (= range/2) in the absence of the barrage. Solid lines are with the outer estuary the Severn estuary model included. Dashed lines are without it. Dotted lines are with it included, but with the delay times in Figure 1 set to zero, to remove natural energy losses. tn We can now calculate the power from the equivalent circuit of Figure Evidently the optimum value for this will be to reflect some of the wave 2. The argument does not rely on the fraction is between 0.4 and 0.6, travelling west considered in section approximations above, but applies and the power increases steadily 3, back up the channel. This reflection equally if accurate values for and as the barrage is moved west. This will be re-reflected from the barrage, Z1 are available. The (complex) volume is of course to be expected - as and then again from the abrupt Z2 flow rate through the barrage is: we move west, the reservoir area transition after section A, in an increases much more than the tidal infinite sequence. We can sum all the range reduces, see Figure 1. waves travelling west into a single (14) wave travelling west between the barrage and Taylor’s section A, and and thus the average power is: Effect of the shape of the estuary west likewise sum all the waves travelling east into a single wave travelling of Taylor’s model east in this region. We can write the (15) Taylor observed that the shape of (complex) volume flow rates in the direction of wave propagation as: This is readily calculated as a function the Severn estuary changes abruptly • and for the wave travelling of , using the expression (9) and west of his outer boundary (section A VO VB RB (11) above for and . It is given in in Figure 1), and ceases to follow his west, at respectively the outer Z1 Z2 Figure 3 for Taylor’s sections A to E of formulae (4), even approximately. The end of the region at section Figure 1. The (complex) tidal pressure width of the estuary approximately A, and at the barrage. in the absence of the barrage is doubles immediately west of section and for the wave P • V’O V’B taken as 4ρg at Watchet, or 8m tidal A, and thereafter follows another travelling east, at respectively range, which is the approximate root- of Taylor’s linearly-tapering profiles, the outer end of the region at mean-square value between the mean with both depth and width increasing section A, and at the barrage. approximately linearly with distance spring range of 10m, and the mean We can first find the ratio of V’ from a notional apex at Abergavenny, O neap range of 5m, and thus gives the to V , which we can express 100km east of section A. The depth of O annual-average power. The values as a reflection coefficient 40m at section A gives a new value of elsewhere are extrapolated from this r where V’ = rV . 8m figure, using Taylor’s formula (6). γ* = 40m/100km = 0.0004 for γ, and O O -1 Y thus a new value k* = 0.00505km In the wave travelling west, the Rather than plotting against R , B for k. We wish to find the effect of G Figure 3 is plotted against the this transition to a new profile on the pressure difference across the barrage barrage wave-making impedance Z . (i.e (14) times ), expressed as a 2 RB fraction of the tidal pressure variation The effect of the abrupt transition /P/ in the absence of the barrage. ENER 118 The optimum position for a tidal power 63 barrage in the Severn Estuary

impedances at the two locations just most recent studies (e.g. Burrows considered are given by (11), we can et al. 2009) of extending computer write them as and . In the wave models out to the limits of the ZO ZB travelling east the impedances can i.e. continental shelf, although the type be seen from (11) to be the complex of boundary conditions applied there conjugates of and (the Hankel are very important (recent studies ZO ZB function H (2) from (8) becomes appear to be subject to the criticism 1 (18) and the –i from (10) that the boundary conditions are becomes +i because the acceleration where we are noting that the zero impedance, see next Section). in the direction of wave propagation argument of is -ωT, The calculations can also be repeated is now plus the surface slope times where T is the wave transit time with the delay-times in Figure tn g). In the region west of section between the barrage and section 1 set to zero, which will remove A, we have only a wave travelling A (readily calculated from (8)). We natural energy dissipation. This west, and the impedance is given can thus obtain the wave-making is done in Figure 3, and reveals by (11) with the new parameter k* impedance at the barrage, as the that natural energy dissipation instead of k, and with x = 100km. sum of the pressures divided by the is only of minor importance. We can write this impedance as Z*. sum of the volume flow rates: Finally, the changes in tidal range The sum of the pressures in the produced by the barrage are two waves immediately east of the important. They are readily calculated transition at section A can now be from the equivalent circuit in Figure equated to that in the single wave (19) 2, using the full expression (19) for immediately west of it. The latter is , and are shown in Figure 4, on When r = 0, there is no reflection Z2 obtained from the volume flow rate at the outer boundary and (19) the same horizontal axis as Figure 3. – in the westward direction: Taking into account the fact that the VO V’O then predicts that the wave- making impedance of the power peak in Figure 3 is further to barrage is Z , as expected. the left for section C, the changes B to the tidal range are very similar for i.e. The barrage powers can be re- all barrage locations. With barrages calculated using this new wave- operated at maximum power, the making barrage impedance Z 2 tidal range is cut to 70% of its former - the results are shown in Figure 3. value east of the barrage, and 90% Evidently the changed shape of the so that of its former value immediately west estuary west of Taylor’s original model of the barrage. A very simple view increases the power considerably, of the barrage is that (from (9) and which is to be expected since the (11)) Z is small compared with Z , (16) increased width of the estuary will 2 1 and R is small compared with C. When * = there is no lower Z and thus from Figure 2 L Z ZO 2 From Figure 2, the optimum power, increase the power. The effect is reflection from the outer boundary, as a matter of elementary electrical and (16) accordingly predicts more pronounced the closer the engineering, is when R has the same that V’ = 0, as expected. barrage is to this increased width. B O impedance as C, This is an existing Thus the conclusion remains We can now find the required result in the tidal power literature, that the power increases steadily wave-making impedance Z of the due to Garrett and Cummins (2004). 2 as the barrage is moved west - barrage, in terms of the reflection It would reduce the tidal range east indeed it now increases more. coefficient r given by (16). From (8): of the barrage by a factor √2, and The question thus arises of the leave the range immediately west of boundary condition even further it unaffected, because Z is small. out, where the second Taylor profile 2 The horizontal axis and colour stops abruptly at the western coding are the same as Figure 3. and extremities of England and Wales. Dashed lines are east of the barrage, This transition can be treated exactly solid lines are just west of it. like the transition at section A. If the (17) impedance is assumed to halve at where and are the x-coordinates this transition, for example, and the xO xB of section A and the barrage. Since calculations repeated, the maximum the RHS of these two powers in Figure 3 all increase, by equations are complex conjugates of between 1% (barrage at section E) ENERG each other. and 7% (barrage at section A). So Thus: again the effect is more pronounced for barrages closer to the transition - it appears that features beyond the UK are relevant to the barrages

furthest down the Severn estuary. Y This supports the practice in the

119 63 The optimum position for a tidal power barrage in the Severn Estuary

Thus the wave-making impedance at the barrage is purely imaginary (i.e. reactive), as we would expect - the barrage can radiate no wave power, because the waves it sends west are perfectly reflected back by the model boundary. Its amplitude is small if the model boundary is close to the barrage, because then θ and φ are nearly equal, and the phase delay ωT of a tidal wave between the barrage and the model boundary is then also small. Thus the change in the results will be small, because is small anyway, as noted at Z2 the end of the previous Section. However, when the model boundary is a long way from the barrage, φ will be small because the tidal wave at the model boundary will resemble an open-water wave. Thus when Figure 4 - Fractional tidal change for barrages at locations A – E of Figure 1. the phase delay ωT reaches 90 degrees, the denominator in (22) Previous computations will drop to zero, and the wave- making impedance of the barrage will The question of the optimum It will produce a total reflection become very large. The power from position for a barrage in the Severn of the outgoing tidal wave – it is the barrage will accordingly drop. estuary, from the power point of equivalent to setting in (16) Z* This condition requires the transit view, was studied 30 years ago, equal to zero. This leads to: time T of a tidal wave between the see Bondi et al. (1981). The power barrage and the model boundary to was computed with various finite- be a quarter of the tidal period, or difference numerical models, some 12.4/4 = 3.1 hours. This is a resonant of which extended out into the Irish i.e. condition, with the natural sloshing Sea. They showed the average power period of the basin between the rising strongly from 0.5 GW to 2.3 barrage and the outer boundary GW as the barrage was moved west (20) equal to the tidal period. With a from Taylor’s section F to D (Bondi et where φ = arg(Z ) is the phase mean tidal wave speed of 25m/s, al. 1981, Vol1 p.18). This is similar O advance of pressure over volume flow say, it corresponds to a distance from to the results in Figure 3, allowing rate, in an outwards-propagating the barrage to the outer boundary for conversion losses. However, tidal wave, at the model boundary. of 25×3600×4 = 360km. This is very little increase was found for For a model boundary at section A, comparable with the size of the larger positions further west. By Taylor’s for example, it can be calculated from models used by Bondi et al. (1981). section C, the power was starting (8) as 45.3 degrees. If we similarly to decline, in marked contrast to It is thus possible that the models write θ = arg(Z ) then Z = ζeiθ where the increase seen in Figure 3 – albeit B B used by Bondi et al. (1981) were ζ is real and θ is the phase advance that significant discrepancies were giving spurious results due to internal of pressure over volume flow rate, resonances, caused by the incorrect found between computer models in an outwards-propagating tidal (Bondi et al. 1981, vol 2 p.57). model boundary condition, in which wave, at the barrage. For a barrage the tidal range was held at the same We now explore a possible reason at section E, for example, it can be value it would have were the barrage for this decline, which is that all the calculated from (8) as 68.9 degrees. absent. The appropriate boundary models simply held the tidal range The wave-making impedance of Z2 condition is an “absorbing” one, fixed on the model boundary, at the barrage (19) thus becomes: which does not reflect waves – these the same value it would have were are standard in naval architecture, the barrage absent. This was then, and familiar in physical model testing and apparently still is, the usual too, as the beach in a wave tank. assumption in tidal modelling (see (21) e.g. Prandle, 1980), although it has Y been recognised as wrong in principle Since eiχ + eiψ = {ei(χ-ψ)/2+ e-i(χ-ψ)/2} i(χ+ψ)/2 = 2cos{( )/2} i(χ+ψ)/2 this G (Garrett and Greenberg, 1977). e χ-ψ e impedance can be written:

(22) ENER 120 The optimum position for a tidal power 63 barrage in the Severn Estuary

Acknowledgement This work was performed under contract to the RSPB, WWF, WWT, the National Trust and the Wye and Usk Foundation. Equations (7) and (8) are due to F.J.M.Farley, who kindly reviewed the manuscript. The author had formerly used an exponential- horn approximation downstream of the barrage, which fits the geometry of the Severn estuary much less well. ENERG Y

121 63 The optimum position for a tidal power barrage in the Severn estuary

References 1. Baker, A.C. 1991 Tidal Power Peter Peregrinus Ltd. on behalf of the Institution of Electrical Engineers. 2. Bondi, Sir Hermann et al. 1981 Tidal Power from the Severn Estuary. Dept. Energy, Energy Paper No.46. HMSO. 3. Burrows, R. et al. 2009. Tidal Energy Potential in UK Waters Proc. ICE - Maritime Engineering, 162, Issue 4 (in press). 4. Garrett, C. & Greenberg, D. 1977 Predicting Changes in Tidal Regime: The Open Boundary Problem. J. Physical Oceanography 7 171-181 5. Garrett, C. & Cummins, P. 2004 Generating Power from Tidal Currents ASCE J. Waterway, Port, Coastal & Ocean Eng. 130 114-118 6. Lamb, Sir Horace 1932 Hydrodynamics 6th Ed. Cambridge University Press. 7. Lighthill, Sir James 1978 Waves in Fluids Cambridge University Press. 8. Mei, C.C. 1989 The Applied Dynamics of Ocean Surface Waves. World Scientific. 9. Prandle, D. 1980 Modelling of Tidal Barrier Schemes: An Analysis of the Open-Boundary Problem by Reference to AC Circuit Theory Estuarine & Coastal Marine Science 11 53-71 10. Prandle, D. 1984 Simple Theory for Designing Tidal Power Schemes Adv. Water Resources 7 21-27 11. Taylor, Sir Geoffrey 1921 Tides in the Bristol Channel in The Scientific Papers of Sir Geoffrey Ingram Taylor, Vol 2. pp 185-192. Cambridge University Press 1960

1800 Y G ENER 122 Greater Plutonio project - Subsea 64 flowline design and performance

David Bruton Abstract Senior Advisor Greater Plutonio is a large subsea development of five fields (BP 50%, Energy Sonangol/Sinopec 50%) in deepwater offshore Angola Block 18. The development comprises a Floating Production Storage and Offloading (FPSO) connected to some 43 subsea wells by some 130km of infield flowlines, which were commissioned and delivered first oil in October 2007. An EPCIC contract was awarded in 2004 to a consortium of Acergy and Technip for the design and installation of the complex subsea systems. The design had to consider not only the flow assurance issues of long distance tie backs, but also the potential for lateral buckling and pipe walking under the anticipated operational cycles. As a starting point, the Atkins Boreas SAFEBUCK design Forbes Sinclair principles were used, supported by specific test programmes that included Head of Pipeline fatigue testing and detailed pipe/soil interaction testing. Following the Engineering, successful completion of the installation programme, detailed monitoring of Atkins Boreas the performance of the flowlines during hydrotest and early operation was Energy carried out to confirm system integrity and validate the assumptions used in design. This paper will cover several aspects of the challenging design for the flowline systems, the development of the test programmes, the installation and survey techniques used to establish the operating performance as well as an assessment of the actual performance against design predictions.

0000 Field layout and overview

Field layout The development is located in 1400m of water some 165km North West of Luanda off the coast of Angola. First oil was achieved on 1st October 2007. Remote Offl oading System The development comprises the following flowline systems N4 Northern Production line which serve 43 subsea wells: Umbilical Single Riser Tower • 2 Off oil gathering networks Northern Water injection line (North and South totalling FPSO mooring approximately 23km and 20km of insulated 12inch flowline) Service line • 2 Off Water Injection (WI) Gas injection line networks (North and South 1350 meters water depth totalling approximately 25km S4 and 30km of 14inch and Southern Production line 12inch flowline respectively) ACERGY Infi eld • 1 off gas injection system Umbilical (South – totalling approximately TECHNIP Southern Water injection line 13km of 12inch flowline) Hydrocarbons are processed onboard an FPSO and offloaded via a CALM buoy. The field produces Figure 1 - Development schematic approximately 250,000 bod. Gas is currently being re-injected into Each of the flowlines utilise a number Schematics of the development are the reservoir, and will continue of flexible and rigid jumpers/spools shown in Figure 1 and Figure 2. until such time as a dedicated to communicate with the subsea exported line to an onshore LNG wells via In Line Tee (ILT) or Flowline terminal is completed. WI rates Termination Assemblies (FTA). will average 750,000 bpd.

123 64 Greater Plutonio project - Subsea flowline design and performance

Basis of design for thermal expansion A key challenge for the subsea flowline design was the likelihood of lateral buckling, pipe walking or route curve instability of the flowlines between riser base and the drill centres. The analytic engineering studies (Phase 1 of the Execution Plan) indicated that all of the flowlines were susceptible to lateral buckling and several were susceptible to flowline walking. Uncontrolled lateral buckling can lead to high strains and cyclic loads at the buckle crown and seriously compromise the integrity of the Figure 2 - Flowlines schematic plan flowline. In addition, the flowlines Execution plan and strategy have ILT structures which require controlled lateral buckling to The issue of thermal expansion and (3) Special “integration” focus prevent unplanned “rogue” buckles the impact on the development was would target specific disciplines forming close to the ILT structures realised in August 2004. First oil was (geotechnical and materials and compromising their integrity. scheduled for 2Q 2007, and the and welding) that were market (supply chain) environment considered key input drivers Flowline walking can occur when a flowline is subjected to steep thermal at this time was overheated. The (4) The soils, materials and welding aforementioned challenge was further transients during start-up/shut-down test programmes would be cycles or when the flowline is laid compounded by the lack of available managed by a third party and data on key input parameters on a seabed slope and subjected funded by BP separately 1 such as soil friction and process to load cycling . Over a number of (5) All FE analysis work would be operating performance data, which cycles, this movement can lead to very 100% independently verified led to a wide range of mitigation large global axial displacement, with possibilities. Finally, it was also (6) All mitigations would be associated overload of the FTA or ILT realised that the analysis itself was scalable, simple, and not connections. The production flowlines going to require complex FE models require new technologies are subject to steep thermal transients and are on moderate slopes. and significant engineering input. (7) A joint (Acergy/Atkins Boreas/ To overcome these challenges BP/Technip) Peer Team would be Flowline design in accordance to an execution strategy was established and a robust peer API1111 was supplemented by the 2 established that encompassed review programme implemented. SAFEBUCK design guideline on lateral buckling and flowline walking seven key elements: The execution plan to deliver issues. The guideline was used to the programme within the (1) A new contractual framework perform the key lateral buckling available schedule was structured (2) A fully integrated team would limit state checks, including local as detailed in Table 1. be established – this would buckling, fatigue and fracture. include BP/Technip/Acergy and Atkins-Boreas personnel Key design data Key design data to address lateral buckling and walking were Table 1 - Execution plan reviewed and established early in the detailed design, including: Phase Primary Parallel Activity • Field layout routes, bathymetry 1 Agree commercial framework Analytic work (Atkins boreas) and flowline configurations 2 Develop robust Procurement/soil data • Design operating conditions and mitigation elements acquisition/fatigue testing operating cycles through life • Thermal transients during Y 3 Develop robust design Procurement/fatigue/ shutdown/restart operations soils/materials testing/ G • Fatigue and fracture design third party verification parameters, SN-curves, 4 Optimise using QRA Testing/fabrication environment effects and Stress 5 Final design Testing/third party Concentration Factor (SCF) verification/fabrication • Pipe material response ENER 124 Greater Plutonio project - Subsea 64 flowline design and performance

• Pipe-soil interaction, including non-linear axial and lateral friction response.

Flowline configurations The configurations are summarised in Table 2, all flowlines are single pipes. The production lines are wet insulated with 5 layer PolyPropylene (PP) coating. All other lines have 3 layer PP coating.

Operating conditions Operating conditions for each flowline are summarised in Table 3, the most onerous co-incident pressures Figure 3 - Hot oiling thermal transient profi les and temperature combinations Table 2 - Flowline data were used for lateral buckling and walking design checks. The Flowline Total Length OD (mm) WT (mm) Coatings table also includes the maximum Production 22.6km (North) 323.9 20.6 90.5mm 5LPP seabed slope along each flowline 19.5km (South, 3 lines) 323.9 19.1 91.5mm 5LPP used in the walking analyses. Service 21.2km 323.9 19.1 2.5mm 3LPP Thermal transients Water 24.4km (North) 355.6 15.9 2.5mm 3LPP Injection 28.8km (South, 3 lines) 323.9 14.3 2.5mm 3LPP On the production flowline thermal transients occur during shutdown/ Gas Injection 12.6km 355.6 15.9 2.5mm 3LPP restart cycles, a full shutdown this Table 3 - Design data includes cold displacement, hot-oiling and well restart. Thermal transient Flowline Temperature Pressure Design Cycles Seabed Slope profiles were determined for all production, service and gas injection Production 95°C (North) 160 bar 92 major, 1.0° flowlines. Figure 3 shows an example 88°C (South) 135 bar 1656 minor 0.96° of hot-oiling thermal transients. Service 51°C 265 bar 184 0.22° Soil friction determination Water Injection 25°C (North) 225 bar 1000 1.05° 23.5°C (South) 225 bar 0.92° Pipe-soil response is the largest Gas Injection 40°C 365 bar 1650 0.81° uncertainty in the design of pipelines susceptible to lateral buckling excellent data on near-surface shear The Greater Plutonio soils testing and pipeline walking. Guidance strength profiles, see Figure 4, and programme included laboratory was provided from the force- soil sensitivity, both fundamental tests to assess axial resistance at displacement models developed to the design of surface-laid the low contact effective stresses for the Safebuck Joint Industry pipelines in very soft clays. that occurs in the soil under typical Project (JIP)3, which included project- As laid pipeline embedment is flowline weights (<10kPa). Specialist specific test data donated to the equipment including the tilt table JIP by BP and other participants. fundamental to the prediction of axial and lateral friction. The values device, see Figure 5, at the University 7 Improving the understanding of pipe- of as-laid embedment predicted of Texas at Austin and the Camshear soil response provided the greatest using published models4,5,6 is device, see Figure 6, at the University scope for optimising the pipeline generally found to be substantially of Cambridge were used to measure design because of the extreme exceeded in practice, due to the pipe-soil interface friction at a sensitivity of design solutions to the load concentration at touchdown range of loads and pipe velocities. axial and lateral resistance imposed by and the dynamic displacement of Large-scale axial and lateral cyclic the soil. The Greater Plutonio Project the pipe catenary during installation tests were also undertaken at NGI, therefore decided to carry out project- causing vertical and lateral see Figure 7, using 4.5m3 of soil specific pipe-soil testing programmes oscillations of the pipe at touchdown. collected from the field and re- ENERG that dramatically improved Calculated levels of embedment consolidated in a large tank8. The understanding, enhancing the pipeline were therefore compared with lateral soils testing at NGI measured design process and significantly observed levels of embedment from lateral first load response and lateral reducing installation costs. similar pipelines in similar soils to cyclic response, including the soil The T-bar test data provided calibrate embedment predictions. resistance at span touchdown

points to each side of a sleeper. Y

125 64 Greater Plutonio project - Subsea flowline design and performance

qT-barqT-bar (kPa) (kPa) NkNk == 10.5 10.5 These tests confirmed that the 0.00 4.00 8.00 12.00 16.00 light production flowline lateral response fitted well with SAFEBUCK 0.00 predictions3, as shown in Figure 8. However, the tests also demonstrated that significantly higher levels of lateral friction could be expected 0.10 0.10 for the heavier WI flowlines. Cyclic tests showed that soil berms are quickly established 0.20 under cyclic loading and provide significantly increased resistance to lateral displacement after the first cycle, see Figure 9. 0.30 This test programme greatly improved the industry’s understanding of

Depth (m) pipe-soil response and the significant resistance provided by the soil berms. 0.40 0.40 If walking predictions are unacceptably high then flowline anchoring would be required. The design had already shown 0.50 0.50 that axial friction was hugely TB1c TB2 TB3 significant to the pipe walking TB4 TB5 TB6 response because, while low levels TB7 TB8 TB11 of friction could be expected, a 0.60 TB12 TB13a TB15 high level of axial resistance can Lower Bound Upper Bound. Bound. Mean reduce walking significantly or Figure 4 - Undrained shear strength profi les defi ned by in situ T-bar tests prevent walking altogether.

Figure 5 - Tilt-table device at the University of Texas Y G

ENER Figure 6 - Cam shear device at the University of Cambridge 126 Greater Plutonio project - Subsea 64 flowline design and performance

Axial test set up

Lateral test set up

Figure 7 - Large-scale axial and lateral cyclic tests at NGI

So as testing progressed BP co- ordinated a high level geotechnics workshop team, including key individuals from within BP, Cambridge University, NGI, Technip, Acergy and Atkins Boreas. This team was instrumental in agreeing an unconventional approach, which relied on the long-term “drained” soil response for slow cyclic axial loading. This approach was supported by test results which demonstrated: (1) Higher levels of axial friction at low contact stresses associated with very light flowlines (particularly wet-insulated flowlines)

Figure 8 - Lateral sweep resistance - Greater Plutonio fl owline tests at NGI (2) A roughened PP pipeline coating compared with SAFEBUCK residual friction model and data adopted for all flowlines was also shown to benefit the design by increasing axial friction (3) Low pipe displacement velocities are more likely to result in a “drained” soil response. The increased level of axial friction resulted in significant project savings by avoiding the need for anchoring on a number of flowlines. ENERG

Figure 9 - Lateral resistance provided by soil berms - NGI tests Y

127 64 Greater Plutonio project - Subsea flowline design and performance

Overview of mitigations

Lateral buckling mitigation If uncontrolled lateral buckling were to occur, the flowlines would have been subjected, at the buckle locations, to extreme loading conditions that could seriously compromise the integrity of the flowline, as outlined earlier. It was decided at the early stages of detailed design to engineer and install mitigation measures to control the locations and shapes of these lateral buckles. The main driving parameters for an efficient mitigation are: a • Determination of the buckle initiation location: The actual spacing between lateral buckles has a significant impact on the loads induced to the flowline. This parameter is linked to the virtual anchor spacing which characterises the load in each buckle. As such, the first engineering step was to determine the planned lateral buckle locations so that the corresponding virtual b anchor spacings would enable a sufficient reduction of the expansion forces in the flowline Figure 11 - Sleeper location plan (a) and sleeper assembly (b) • Buckle initiation under • Flowline integrity over field vertical imperfection. The length of hydrotest/operation induced life needs to be ensured: The each sleeper was determined based loads: It is necessary to create loads in the buckle are impacted on the predicted maximum lateral an artificial imperfection along by the actual pipe/sleeper friction displacement of the flowline. the pipelay route to efficiently coefficient; the cyclic loading The ranges of lateral displacement trigger buckle formation. This was induced by field operations; and were up to 15m and this could occur achieved with the use of sleepers the touchdown conditions within to either side of the sleeper centre. pre-laid perpendicularly to the the soils. For all planned buckle In view of installation tolerances flowline route, these supports locations the flowlines were and safety margin this led to a creating vertical imperfections welded using fatigue class weld maximum sleeper length of 42m. large enough to enable buckle parameters over a 200m distance. The final location of planned firing as illustrated on Figure 10 For the production flowlines local buckling sites, along with the buckle additional buoyancy was used to parameters such as amplitude, was increase the buckle length so as to also determined by the need to achieve acceptable stresses at the minimise the expansion levels at buckle crown. This was achieved in-line structures and at crossing through extra thickness of the locations. Sleeper locations are insulation coating over 100m presented on Figure 11(a). length either side of the sleepers. The coating on top of the sleepers The solutions engineered during the was selected based on tests to ensure detailed phase of the project led to an low friction levels and durability intensive testing programme which Y throughout the flowline life. Figure encompassed friction behaviour of 11(b) shows a typical sleeper assembly G materials, welding and fatigue testing used for the production flowlines. described in subsequent sections. The design of the sleeper assemblies relied on the use of 32inch diameter pipe sections to achieve the required

ENER Figure 10 - Buckle fi ring (Initiation) on a sleeper 128 Greater Plutonio project - Subsea 64 flowline design and performance

a

b

c

Figure 12 - Flowline anchor confi gurations Pipe walking mitigation The susceptibility of all flowlines to global walking and curve pull-out was assessed through extensive studies simulating the global behaviour of each flowline system over its entire d life with simulation of operation shutdown and restart cycles. Sliding structures for With the aid of rough coating, axial displacements applied to the external coating to increase axial friction, it was on Figure 12(a). The difficulty was In order to accommodate the shown that the water and gas to ensure very precise alignment of anticipated levels of expansion at injection and service lines were not the pipe end section, including the flowline ends or at in-line structures, impacted by pipe walking issues. FTA structure with the locking device all piping modules were designed so However, it was shown that all three installed on top of the suction pile as to be capable of sliding relative branches of the South Production to avoid side loads that the chain to the mud mat. This was achieved system were affected and needed may have induced on the FTA if not through low friction rails on either to be anchored at one end to properly aligned. This necessitated side of the structure frame. In view prevent walking. The potential the design of the innovative of the weight of up to 70tonnes retaining loads calculated were chain capture solution shown on and the size of up to 14m length in the range of 100tonnes which Figure 12(b) and Figure 12(c). by 6.5m width for these structures, dictated the use of suction anchors. Using quantified risk assessments this was quite a challenge. Selection In view of the field layout and due it was decided that the northern of the low friction rail and pad to operational reasons, it was not production system would not need materials was made through extensive possible to anchor the flowline ends laboratory testing, as presented

to be anchored to prevent walking. ENERG during flowline lay initiation which However, the residual uncertainty in Section 5. For the production could have allowed the use of the in axial friction predictions meant flowlines the maximum sliding traditional and relatively simple stab that the risk was significant, enough distance required was found to be and hinge over method. Anchoring to warrant the implementation 4.5m for FTA structures and 4m for of the flowline ends needed to be of an in-line anchor attachment ILT structures. The range of expansion done during lay-down with a length arrangement to allow later retrofitting was determined for each of the of chain connecting the pipe end of pile anchors Figure 12(d). structures. For the ILT structures of the Y to the suction anchor as illustrated WI flowlines 1m sliding was required. 129 64 Greater Plutonio project - Subsea flowline design and performance

Figure 13 - Sliding structures assemblies

Pictures of a typical structure and soft soil conditions it was decided displacement of the flowline above assembly and sliding rail details to achieve the separation between the sleeper. The crossing location are presented in Figure 13. both lines with sleepers similar to was adjusted as much as possible Markings on the rails and expansion those used for lateral buckle initiation. in view of the layout constraints recording pads (shown in Figure The vertical imperfections created to correspond to areas of low pipe 13) will allow future monitoring made it necessary to consider the axial compression force to minimise of axial displacements throughout presence of those crossings in the the risk of buckling. However, the

Y the field operational life. lateral buckling studies. It was then risk of having an undesired buckle required to ensure that no lateral or at the crossing could not be totally G Crossing design vertical displacement of the flowline eliminated. It was then decided to could occur at the crossing location in restrain lateral and vertical movements There are 3 locations where the order not to make the crossing design by installing a locking beam after production flowlines cross the WI further complicated. Finally, the laying the pipe on top of the flowlines. In view of the water depth design should ensure free longitudinal sleeper, as illustrated in Figure 14. ENER 130 Greater Plutonio project - Subsea 64 flowline design and performance

Figure 14 (a) - Elevation on Crossing Design

Stoppers Flowline Stopper

Figure 14 (b) - Installed crossing

Testing of sliding FTA and ILT sliding pads and materials rails - friction testing Lateral buckling of flowlines on The aims of the test programme were: Suitable FTA/ILT pad material the Greater Plutonio development (1) To measure the static and combinations tested, see Figure 15, is controlled by the use of sleepers dynamic friction coefficient of the included Polytetrafluoroethylene to raise the pipe off the seabed at various material combinations (PTFE) on PTFE or Epoxy coating, intervals along the length of each under various design loads which achieved a very low friction flowline. The design relies on the and displacement speeds coefficient (up to 0.04) and Ultra- flowlines sliding across the top of High Molecular Weight Polyethylene (2) To confirm material wear and the sleepers. The friction and wear (UHMWPE) on Epoxy (up to creep rate over the equivalent performance of the pipe-sleeper 0.08). The PTFE/PTFE combination design life, to ensure that interface is important to system exhibited the best wear and friction the loss of thickness would performance for the life of the field. properties but field experience has not compromise the wear shown it is difficult to manufacture The ILTs along the flowlines and surface performance FTAs are supported on mudmat for this application. These tests (3) To confirm the full scale foundations with the flowline also showed that the increase in performance of the pad design, components sliding on rails with friction that occurs after a long including the attachment method. low friction support pads. The dwell period or with seabed soil foundation loads and the flowline end The test programme proved of at the interface was marginal. loads are affected by the frictional great value in eliminating unsuitable resistance of the pad-rail interface, materials and enabling the project which is important to system to confirm suitable material performance for the life of the field. combinations with acceptable friction and wear properties for To understand the behaviour and all required applications. long-term performance of these sliding interfaces a test programme

managed by Atkins Boreas on ENERG behalf of BP and the Greater Plutonio Contractors was carried at Doosan Babcock (Scotland). Y

131 64 Greater Plutonio project - Subsea flowline design and performance

Figure 15 - Back-to-back bearing pad/rail confi guration prior to seawater submersion Flowlines on sleepers Fatigue testing - friction testing and welding The pipe-sleeper interface tests, However, friction coefficients for In-air tests of welds confirmed see Figure 16, measured consistent the Polyurethane (PU) field joint acceptable in-air fatigue performance friction coefficients 0.14 to 0.36 were higher than expected at 0.54 for all welds. Specimens tested for under operational loads for insulated to 0.76. Although not considered a the WI flowlines were subjected flowlines with PP coating, and 0.09 long-term concern, some wear of to prestraining representative of to 0.17 for the heavier WI flowlines the field joint coating was observed. the reeling cycle before testing. with PP coating. These were lower The Glass-Flake Epoxy (GFE) These tests confirmed that there than expected and within design sleeper coating was shown to be was no clear influence of prior limits. The tests demonstrated susceptible to wear over many reeling strain on the subsequent the expected reduction in friction cycles under high vertical loads. fatigue performance of the welds. coefficient with increasing load While this is not expected to be a A desk study carried out by TWI and showed that increasing dwell concern for the Greater Plutonio confirmed that little data existed on times between sweeps did not flowlines, this coating is unlikely to the fatigue performance of pipeline significantly influence results. be acceptable for heavier pipelines. girth welds in a corrosive environment, under the high stress/low-cycle fatigue conditions applicable to the Greater Plutonio production flowlines. A programme of sour service endurance fatigue testing in simulated pipeline operating conditions was therefore initiated at TWI to support the flowline design process. Testing compared the fatigue life of welds in air with that in a sour environment. A particular concern was the increase in fatigue damage due to low frequency cycling, associated with cyclic lateral buckle loading. Results of tests 3L PP coating on sleeper carried out at 0.2Hz suggested that the fatigue performance, for the three pipeline welding procedures tested, is similar and that the fatigue life reduction factor relative to air is approximately 10. The use in design of an ‘E’ curve divided by ten encompasses the 0.2Hz and 0.01Hz results, shown in Figure 17. In parallel testing, the effect of low frequency Fatigue Crack Growth (FCG) rate in the same simulated sour environment was studied using

Y frequency scanning at constant Stress Intensity Range (DK) as well G as the more usual variable ΔK Insulation coating on sleeper approach. The FCG rate was shown to increase with decreasing fatigue load frequency, to reach a plateau Figure 16 - Back-to-back pipe-sleeper test arrange- at around 0.1Hz, see Figure 18.

ENER ment prior to seawater submersion 132 Greater Plutonio project - Subsea 64 flowline design and performance

Figure 17 - Fatigue endurance test results in sour environment These results proved useful in guiding Lower bound data on toughness Experience from low-frequency testing the test frequencies employed in was measured in terms of the stress in seawater for the SAFEBUCK JIP the endurance testing, discussed intensity at which crack extension had highlighted a similar concern earlier. The crack growth rate data occurred by stress corrosion cracking about a significantly increased obtained was used as one of the (Kiscc) at test temperatures in the “fatigue life reduction factor” under required inputs to Engineering Critical range 20°C to 60°C. Other tests low frequency loading in seawater Assessment (ECA) calculations from conducted on hydrogen charged with Cathodic Protection (CP). As which weld acceptance criteria were samples were used to measure the a result, particular attention was derived for flowline girth welds, effect of hydrogen embrittlement given to field joint integrity on all prepared onshore and offshore. on the overall toughness of pipeline flowlines to ensure that welds would A further test programme by TWI, welds at ambient temperature. not be exposed to seawater. established the fracture toughness These results were then used in of welds in the simulated sour the flowline ECA, to calculate the Plutonio environment under operating tolerable flaw sizes for surface and shutdown conditions. breaking and embedded flaws. ENERG Y

Figure 18 - Increase in fatigue crack growth with reducing frequency 133 64 Greater Plutonio project - Subsea flowline design and performance

Fabrication and installation An overview of the fabrication The subsea structures piping was • Rigid line pipe SCR type and installation scope which was fabricated in Europe, and then welding for fatigue sensitive split between Technip and Acergy assembled to the structural beams areas: more than 270 manual is presented in this section. and mud mats in South Africa. It is GTAW with only 1-off repair worth noting that all piping within • HDPE liner fusion welding into Technip scope these structures was internally 1000m stalks, see Figure 19(b) clad with Inconel 625 alloy. Technip installation scope • HDPE insertion into line pipe was as follows: The rigid pipelines were fabricated stalks, see Figure 19(c) at the Dande Spoolbase jointly • 55km of WI flowlines • HDPE liner energization, owned by Technip and Sonangol. (plastic lined pipe) see Figure 19(d) Over 500 000 man hours were • Sleepers for buckle initiation spent on the following activities: • Stalks assembly into circa 3km sections, see Figure 20 • 16 WI FTAs and ILTs • Rigid line pipe jointing into • 12 WI flexible spurs (20km 1000m stalks: circa 2000 • Pipeline spooling onto the length in total). automated welds were installation vessel. See Figure 21. performed, with no recordable time loss, see Figure 19(a)

b

a Y G

d c

ENER Figure 19 - Rigid pipeline fabrication at the Dande Spoolbase 134 Greater Plutonio project - Subsea 64 flowline design and performance

The sleepers were assembled at the Dande Spoolbase, and then loaded out on the Geofjord vessel, to be installed prior to the pipelay activities, see Figure 22. The 56km of WI rigid flowlines were spooled onto the Deep Blue, and then reel-laid. The 15-off subsea structures inserted in the rigid WI Flowlines were also installed by the Deep Blue, while the flexible spurs and the associated subsea structure were installed using the Constructor, see Figure 23. The Geofjord also Figure 20 - Stalk assembly providing field assistance to the Deep Blue, retrieving the buoyancy modules used to land the subsea structures.

Figure 21 - Pipeline spooling onto the Deep Blue installation vessel ENERG

a b Y Figure 22 - Sleepers loaded out Figure 23 - Installation of subsea structures (a) and fl exible spurs (b) at the Dande Spoolbase 135 64 Greater Plutonio project - Subsea flowline design and performance

Acergy flowline installation scope Acergy installed the production, The Acergy installation Flowline initiation was service and gas injection flowline scope was as follows: performed as follows: systems using the Acergy Polaris • 34km of Production flowlines (1) FTA inserted in tower from (flowline and large spools) and (with insulation coating) bottom by crane, see Figure 26 with the Acergy Eagle for the • 42km of Gas Injection and (2) Top end collar landed on bushing anchor piles. The flowline lay and Service flowlines (non insulated) structure installation was carried (3) First weld performed, out using the new J-lay tower • 24 sleepers for buckle initiation see Figure 27 installed on the Acergy Polaris • 3 flowline anchors and 2 (4) Initiation structure linked as illustrated on Figure 24. devices for retrofit anchors to initiation wire/chain for The new tower enabled a maximum to control pipe walking seabed connection. tension capacity 950tonnes and • 18 FTAs and ILTs Two welding procedures were Gimbaling ability of 15° pitch • 17 rigid spools (including spools qualified for the production flowlines and roll, see Figure 25. Two work in excess of 100m length). to cover the welding of double stations allowed for welding joints onshore and the offshore and field joint coating. installation by the J-lay method.

Figure 24 - Acergy Polaris installation vessel

• Pipelay angle from 0o to 180o • Installation vessel bow/stern to main swell (+/-20o allowance) • Increase installation window (less weather downtime) Y G

Figure 25 - J-Lay Tower Gimballing ENER 136 Greater Plutonio project - Subsea 64 flowline design and performance

a b

Figure 26 - Flowline initiation - FTA installation The welding technique qualified by the allowable root misalignment on The fatigue sensitivity of the the double joint contractor involved fatigue welds was more restrictive. buckle zone welds meant that a 60° included angle, single “V” The welding procedure qualified for weld acceptance criteria based on preparation with manual GMAW the J-lay installation used a standard workmanship standards were not root and intermediate runs, sufficient narrow gap weld preparation. appropriate. Automated ultrasonic to support the subsequent fill and The Surface Tension Transfer (STT) testing (AUT) was essential to permit capping by SAW. This work was GMAW technique was used for the the through thickness extent of performed at a double jointing facility root run, as an alternative to the flaws to be determined and achieve set up for the project in Brazil. The more traditional copper backing the necessary reliability of flaw same welding processes were used technique. Fill and cap runs utilised detection and sizing. The selected for double joints destined for both mechanised pulsed GMAW. AUT Contractor was able to provide fatigue sensitive buckle zones and limited previous qualification data for other areas in the flowline, although the double joint weld configuration, hence a more extensive AUT validation programme was undertaken. This involved the preparation of test welds with artificial defects and metallographic sizing of defects to compare with the flaw sizes given by the AUT system. For the offshore J lay welds, previously qualified data was available, but again was supported by a validation test programme. For both the double joint and J-lay welds the degree of AUT under-sizing was less <1.0mm ENERG and a probability of detection in accordance with the requirements of DNV OS-F101 was achieved. Y

Figure 27 - First weld performed on fl owline 137 64 Greater Plutonio project - Subsea flowline design and performance

a b

Figure 28 - Installation of FTA and ILT structures

The sizing accuracy values from Laydown with FTA and ILT structure On the south production system the validation testing were used installation were performed by 3 FTAs required to be connected to modify the tolerable flaw sizes inserting the structures in the J-lay to anchor piles for pipe walking obtained in the flowline ECAs, tower according to similar steps as prevention. Suction anchors needed allowing the development of the initiation structures, see Figure 28. to be installed prior to laydown with final weld acceptance criteria the Acergy Polaris. The top of pile, as applied to production welds. installed, is shown in Figure 29 (a).

a b Y G

c d

Figure 29 - FTA connection to anchor piles ENER 138 Greater Plutonio project - Subsea 64 flowline design and performance

a b

Figure 30 - Sleeper deployment and pipelay over a sleeper

The steps for connection of the pipe end to the anchor pile with a chain during laydown are shown in Figure 29 (b) to (d). Sleepers were deployed with the barge crane while pipe-laying, since the touch down point was significantly behind the barge location, see Figure 30 (a). Transponders installed on the sleepers and Remote Operated Vehicle (ROV) observation ensured that the flowline was later laid as close as possible to the centre of the sleeper, see Figure 30 (b), but with an offset on lead in. Rigid spools were installed with the Acergy Polaris. A specific spreader frame was fabricated allowing deployment of the whole range of shapes and sizes as a illustrated on Figure 31. The longest spools measured 80 x 40m. All spools between FTAs or ILTs and the manifolds were connected using FMC horizontal connectors with the CAT tools whereas on the riser tower side, the spools were connected using the Matis flanged connection tool illustrated in Figure 32 (a) while Figures 32 (b) and (c) show the subsea connection steps. ENERG

b

Figure 31 - Rigid spool installation Y

139 64 Greater Plutonio project - Subsea flowline design and performance

Operational monitoring

Overview Operational monitoring for each of the systems relies on an integrated process that re-evaluates the remaining fatigue and allowable expansion available within the structures on a periodic basis. A highly accurate, as-laid survey was carried out to establish the initial condition of the flowlines including out-of-straightness (OOS) along the full length of each flowline. Additional surveys are being undertaken during hydrotesting and post-startup, to verify buckle formation identify rogue (unintended) a buckles and confirm operational loading at each lateral buckle site.

Deflected position survey requirements The survey requirements for successful monitoring required accuracies of better than 0.2m deflection over 50m length in a water depth of 1400m without using an LBL array. To enable this, the following survey equipment was integrated to the ROV: (a) Kongsberg HiPaP 500 transponder/responder in Ultra Short Base Line (USBL) mode (b) Kongsberg Hydroacoustic Aided Inertial Navigation System (HAIN), fitted with CDL miniRLG 2 or Ixsea IMU90 subsea unit, and NavLab post processing software b (c) Doppler Velocity Log (DVL) RDI WorkHorse Navigator 600KHz or 1200KHz (d) Subsea gyrocompass CDL miniRLG2 and/or Ixsea Octans (e) Paroscientic Digiquartz Depth Sensor (DQ) (f) ROV mounted Valeport or SAIV CTD and mini SVS (g) Dual Heads Multibeam Echosounder (MBE) Reson Seabat 7125 (h) Starboard and portside boom cameras, with central camera. All three videos were logged using

Y VisualSoft digital video system

G (i) Wheeled undercarriage system, adjustable for 300mm to 500mm flowline OD. c

Figure 32 - Matis fl anged spool connection ENER 140 Greater Plutonio project - Subsea 64 flowline design and performance

Digiquartz

Starboard SeaBat Head

IMU90

Digiquartz mounted CDL RLG2 Laser behind Light

a DVL b

Figure 33 - HiRov 3000 (a) and SCV 32 (b) The above configuration enabled Hain IMU two OOS survey options: DVL HAIN HAIN Digiquartz Online Data Navlab Post Contact Method: With the ROV APOS_HiPaP Acquisition Processing continuously rolling along the 1 pps GPS DO Charting flowlines. This was the primary NaviEdit method as it allows direct derivation Fledermaus Nave Sensors NaviScan or Point Rejection of the XYZ position. The HAIN INS Rov Sensors Data Logging DeepOcean NaviPac results are directly used as flowline Data Merging DBMS Data Cleaning reference. They can be compared Data Processing Data Editing Data Editing Data Display Extraction of Data Flagging with ROV tracks computed using Data Correction 5-point Profi le alternative processing methods such MBES for Tide, Motion Data Reporting QC and Refraction as Gyro+DVL and/or HiPaP+DVL, Data Data Processing for quality control and removal of ArcGIS possible positioning artefacts. Digital Event / Event / Non Contact method: With the Video Navigation X-Profi le ROV flying 0.5m to 1.5m over the Eventing Merge Merge flowlines. The XY values of the Top of Pipe (TOP) are then determined Post Processing from the processing of the swath bathymetry data in relation with Data Collection Quality Control the ROV Track (HAIN INS aided Figure 34 - Data fl ow/quality control schematic USBL HiPaP). The Z values of the TOP are determined from the Digiquartz combined with the MBE. This method was used when the flowlines were buried or too deeply embedded and also for spools. A very detailed HAIN, DVL and MBE calibration procedure was performed offshore prior to the survey start up. This involved an onshore analysis at Kongsberg Horten for fine tuning of the Inertial Measurement Unit (IMU), to be reset within the Navlab post- processing parameters. The following ENERG deliverables were produced onboard for immediate engineering analysis: (a) Processed ASCII Flowline 5 points files at 1m spacing (b) Cross Profiles at 1m spacing

(c) Digital video logs, combined Y with flowline events Figure 35 (a) - Thermal expansion 141 64 Greater Plutonio project - Subsea flowline design and performance

Occasionally, some lines were also surveyed using SCV 32, see Figure 33 (b), deployed from M/V Acergy Legend, for example during hydrotest OOS surveys along the south production system. During the remaining field life, ongoing surveys will be performed using a dedicated infield support vessel.

Use of GIS data base as an IMR tool Operational monitoring uses geographically referenced information that enables engineering analysis to be undertaken on a holistic basis. The use of GIS also allows ease of recognition of apparent areas of concern, and rapid Figure 35 (b) - OOS survey/project data fl ow prioritisation of ongoing inspection (d) OOS check list visual observations of the soil berm or intervention activities. Table 4 provides an (e) Raw and processed MBE data created to one side of the line. example of some of the data that for Fledermaus import and Comparison made between pre- is contained and regularly updated 3D visualisation by NetSurvey hydrotest, hydrotest and post start within GIS for the flowline system. (0.15m grid for detailed buckles up surveys have allowed high quality Integrated data display examples are areas and 0.25m grid over the data to be extracted and used shown in Figure 36 and Figure 37. complete flowline length). within engineering FE analyses. The red dot on the field map in All available information was Data is currently being imported Figure 37, indicates position of ROV combined during analysis of the into Geographical Information when acquiring these images. OOS survey results. In particular, System (GIS) for use during the very small rogue buckles identified operational monitoring of the field. during hydrotest with an amplitude The majority of the surveys conducted of 0.2m to 0.5m, over a 50m long to date were undertaken using a section, were confirmed from DeepOcean WorkRov HiRov 3000, positioning track comparison, and see Figure 33 (a), deployed from differences in the left/right differential the M/V Normand Tonjer, which embedment level, combined with is a DP2 ROV/Survey vessel.

Table 4 - Example of GIS fl owline data

Item Description When Comments 1 Deflected position of At installation Provides key input data for fatigue life flowline/ Spool (geo calculations and monitors issues such as During hydrotest referenced video/ MBES/ curve pull out / rogue buckle growth etc. 5 point cross profiles/ Post start up embedment etc) Intervals during field life 2 Linear expansions on As for 1 above Provides data on walking ILT/FTA assemblies 3 External coating integrity Post start up Provides data on whether knock on deflected flowline areas down factors require adjusting (i.e. Intervals during field life welds exposed to seawater) 4 Internal crack detection on Intervals during field life Provides data on fatigue weld integrity. deflected flowline welds

Y 5 CP field gradient Intervals during field life Data on coating integrity

G 6 Internal and external Intervals during field life As above corrosion on deflected flowline areas 7 Start up and Continuous Allows subsea and topsides teams to shutdown history evaluate possible contingency plans ENER 142 Greater Plutonio project - Subsea 64 flowline design and performance

Performance Detailed Out of Straightness (OOS) and end expansion surveys were carried out along each flowline post installation, during hydrotest and after first operation. OOS surveys were also performed for the spools. The data from these surveys was then used to determine the performance of each flowline and allow it to be compared to design. The hydrotest and first operation surveys were also used to determine whether any unplanned (rogue) buckles had formed. Buckle formation The primary function of the OOS surveys during hydrotest and first operation was to determine whether buckles had formed at the planned Figure 36 - Cross profi les and combined MBES 3-D imaging locations. Detailed design analyses concluded that the hydrotest would generate enough axial force to initiate planned buckles on all lines. This effectively meant that the hydrotest could be used as a “dummy run” for buckle initiation before flowlines were subjected to the more onerous operating conditions. It also gave the project scope to force buckle initiation during start-up, if required. A summary of the buckle formation at hydrotest and during first operation is included in Table 5. Buckle formation was very reliable on the production, service and gas injection flowlines during hydrotest, with all 24 planned buckles forming; in addition only 4 unplanned “rogue” buckles formed. Each rogue buckle formed between planned initiators and did not hinder planned buckle formation. The flowline design accounts for on-bottom rogue Figure 37 - As built digital video images buckles forming between initiators. A number of the WI Buckle Initiators (BI) did not trigger buckles during hydrotest. Several of the initiators are expected to form during Table 5 - Buckle formation normal operation. The reason that some WI buckles did not initiate Flowline Planned Hydrotest buckles formed Operation buckles formed during hydrotest has yet to be fully Buckles Planned Rogue Planned Rogue explained, although there are a Production North 9 9 1 TBC TBC number of potential causes, including potentially high residual tension in ENERG South 6 6 0 6 0 the flowline, low out-of-straightness Service 4 4 2 TBC TBC level, the designed pipe-sleeper Water North 7 5 3 TBC TBC configuration being less conservative Injection than those on the production or gas South 11 5 5 Injection flowlines, or possibly the

Gas Injection 5 5 1 TBC TBC fact that the WI lines are heavier Y than the production flowlines. 143 64 Greater Plutonio project - Subsea flowline design and performance

Buckle shapes Detailed OOS surveys were carried out along each flowline prior to hydrotest, during hydrotest and after first operation. Each survey provided a very accurate understanding of the behaviour of each buckle under different loading regimes. An example of typical survey data is presented in Figure 38 (a) includes an overlay of all three flowline OOS surveys. Survey video was also used to cross check the buckle a amplitudes on each sleeper; lateral measurement markings on each sleeper allow estimation of buckle movement. An example pipe on sleeper is presented in Figure 38 (b). The survey data clearly shows the formation of a buckle on the initiator during hydrotest which grows in amplitude at first operation. For each flowline the surveyed buckle shapes were used in FEA analyses to determine the loads in each buckle and also estimate the actual pipe-soil response. Buckle shapes from first operation, see Figure 39, were within the bounds expected in design. As part of the ongoing integrity monitoring b programme the buckle shapes were fitted using finite element models, the Figure 38 - OOS surveys – (a) Survey Data comparison, (b) Pipe movement on initiator majority of buckles achieved excellent shape fits with lateral friction at large displacements well within the design range and close to the mean defined by pipe-soil interaction testing.

End expansion Each FTA and ILT was designed with measurement bars on the sliding frame so that accurate readings of axial movement could be taken during each survey. Readings were taken at each FTA during hydrotest and during first operation. Figure 40 presents typical video-grabs of slider movement on an FTA for hydrotest and operation. The observed end expansions in operation for each of the southern Figure 39 - Buckle shape in operation production flowlines were within the bounds set in detailed design, most of the expansion values lay Y reasonably close to the mean design

G predictions, typically within 0.1m. ENER 144 Greater Plutonio project - Subsea 64 flowline design and performance

a b

Figure 40 - End Expansion Readings – (a) Hydrotest, (b) First Operation Ongoing flowline Abbreviations and monitoring nomenclature The flowline integrity management Key lessons learnt include: AUT Automated plan calls for regular flowline (i) The necessity to perform Ultrasonic Testing surveys to check buckle shapes analytic modelling at the BI Buckle initiator and end expansion readings. The Front End Engineering Design COF Coefficient of Friction survey data will be combined with (FEED) stage of any design regular monitoring of pressure CP Cathodic Protection (ii) Early integration with and temperature conditions in DK Stress intensity range topsides design teams each flowline so that models can FCG Fatigue Crack Growth be updated regularly to monitor (iii) Early implementation of GIS FEED Frontend Engineering buckle shapes, predict fatigue life (iv) Recognition that soils and Design consumption and also determine fatigue behaviours will require whether flowline walking is occurring. an integrated test programme FPSO Floating Production Storage and Offloading (v) The realisation that the modelling Conclusions complexity would benefit from FTA Flowline Termination Assembly The flowline system was installed off independent third party analysis the critical path with first oil being (vi) Evaluation of the effects of pipelay ECA Engineering Critical achieved on October 1st 2007. systems and methodology on Assessment The performance of the system embedment, residual lay tension, GFE Glass-Flake Epoxy was within design limits and no and the provision of initial offsets GIS Geographic remedial works have been required. to assist buckle formation. Information System Several world firsts were accomplished ILT In-line Tee along the way. These include the Acknowledgements JIP Joint Industry Project soil testing and analysis, sour service The authors would like to thank all OOS Out of straightness low frequency fatigue testing, the those individuals and companies use of integrated buoyancy and that have contributed to the success PP Polypropylene sleepers to control buckles, the use of the Greater Plutonio Project and PTFE Polytetrafluoroethylene of rough coated pipe to increase axial give a special mention to those PU Polyurethane friction, a novel installation scenario individuals who have also contributed for connection and tensioning of to the writing of this paper: R Ratio of minimum load piles to flowlines, and the provision to maximum load Peter Bowden (BP) – Fatigue of a future (contingency) mid- ROV Remote Operated Vehicle line anchor pile connection. Testing & Welding SCF Stress concentration factor In addition, there were several design Andrew Cosham (Atkins Boreas) UHMWPE Ultra-high molecular details which should receive mention: – Fatigue Testing & Welding weight polyethylene these include the use of high integrity field joints, machine profiled J-lay WI Water Injection ENERG collars, the elimination of anodes on the flowlines to assist with overall coating integrity, and the use of wet lay to enhance embedment and reduce operational loads (less effective force in the operation). Y

145 64 Greater Plutonio project - Subsea flowline design and performance

References 1. Carr, M., Sinclair, F., Bruton, D., “Flowline Walking – Understanding the Field Layout Challenges, and Analytical Solutions developed for the SAFEBUCK JIP”, OTC-17945 Offshore Technology Conference, Houston (2006). 2. “Safe Design of Flowlines with Lateral Buckling - Design Guideline” Report BR02050/ SAFEBUCK/ Rev B, SAFEBUCK JIP (2004). Confidential to JIP Participants 3. Bruton, D., White, D., Cheuk, C., Bolton, M., Carr, M., “Pipe-Soil Interaction Behaviour during Lateral Buckling, Including Large Amplitude Cyclic Displacement Tests by the Safebuck JIP” OTC 17944 (2006). 4. Murff, J.D., Wagner, D.A., & Randolph, M.F. “Pipe penetration in cohesive soil”. Géotechnique, 39(2):213-229 (1989). 5. Lund, K.M.. Effect of Increase in Pipeline Penetration from Installation. Proc. of the Conference on Offshore, Marine and Arctic Engineering. New Orleans, USA, (February 2000) 6. Cheuk, C.Y., White, D.J., “Centrifuge Modelling of Pipe Penetration due to Dynamic Lay Effects” Proc. Conf. on Offshore Mechanics and Arctic Engineering (2008). 7. Najjar, S.N., Gilbert, R.B., Liedtke, E.A., McCarron, W. Tilt Table Test for Interface Shear Resistance Between Flowlines and Soils. Proc. Conf. on Offshore Mechanics and Arctic Engineering (2003). 8. Langford, T.E., Dyvik, R. & Cleave, R., Offshore pipeline and riser geotechnical model testing: practice and interpretation OMAE2007-29458 (June 2007) Y G ENER 146 Development of an armour mass 65 estimation tool for land vehicles

Matt Geary Abstract Graduate Engineer The Future Rapid Effects System (FRES) is the British army’s largest Defence procurement programme; aiming to deliver a family of medium weight armoured vehicles. This paper describes the Atkins survivability team’s development of a spreadsheet tool to model the weight and performance of conceptual armour systems for FRES vehicles. The selection of armour materials is complicated by the wide variety of impact mechanisms, which have conflicting demands on armour material properties. This paper first discusses some of the more common types of impact mechanism and armour type. The general armour concepts of Whittaker weighting and mass efficiency are then described. The development of the ‘Weight stack’ spreadsheet tool is split into the three main stages of; describing vehicle geometry, storing material information and calculating armour weight.

Background Threat types The Future Rapid Effects System (1) The protection levels which There a huge variety of threats project covers a wide variety of are achievable within a given in existence, but the type of vehicles, including protected troop weight budget, to ensure that impact can largely be broken carriers, reconnaissance and direct protection requirements given to down into four categories: fire roles. Despite their different vehicle providers are achievable. purposes, the vehicles all have (2) The likely weight of a defined Low kinetic energy projectiles the common goals of providing protection level, in order to This is a broad category encompassing high levels of protection and assess a vehicle provider’s any projectile travelling below mobility. In addition, the vehicles armour proposals approximately 1200m/s, with the are all to be air transportable so (3) The problem of assessing armour threats ranging from blast fragments as to provide the British army weight and performance is to heavy machine gun projectiles. with a rapid reaction capability. complicated by the variety of Even within this category there are Atkins has worked alongside threats, impact mechanisms, many different impact mechanisms the Ministry of Defence (MoD) armour types and other variables. at work, several of which can occur Defence Equipment and Support at different stages of the impact. (DE&S) group as a customer friend The specific impact mechanisms and provides a variety of systems of the projectile will be dependent engineering and technical services. on many factors including the velocity, material properties, Statement of problem geometry and angle of incidence. In armoured vehicle design there has always been a tension between the requirements for protection and for mobility. In order to provide higher levels of protection, additional armour DEFENC is inevitably added, decreasing the power to weight ratio of the vehicle. In order to effectively balance the protection requirements for an armour vehicle, the Survivability team requires a way of assessing: Figure 1 - Sample range of low KE projectiles E

147 65 Development of an armour mass estimation tool for land vehicles

To provide effective protection Therefore an armour material At such high speed the penetration against a projectile, an armour should be specified to have high is dominated by material density material must be capable of shear yield strength and ductility, and so the strength of materials both resisting penetration and while also having high resistance to can be largely ignored. However, absorbing the projectile’s energy. shear instabilities through having while the penetration is inversely To resist the initial penetration of a high work hardening index and proportional to the square root of the projectile, it is desirable for the low thermal softening rate. the armour density, the armour material to have high hardness and Taking low K.E. threats alone, the thickness is linearly proportional to high yield strength. A particularly ideal armour material will therefore the density. A low density material hard material is also likely to deform have high ductility, high hardness, will therefore be more effective per the nose of a projectile, or in some high yield strength, high shear yield kg of material at preventing complete cases shatter it. This distributes strength, high work hardening index penetration by long-rod penetrators. the impact and reduces the kinetic and low thermal softening rate. energy density of the projectile, Clearly there are many competing Shaped charge so reducing the penetration. and in some case mutually exclusive Shaped charge devices go by a variety Obviously, for a projectile to pass demands. In practice a balance must of names, but broadly speaking a through a surface it must move be struck in order to ensure an armour shaped charge is any device using material out of its path. This material provides protection against a an explosive lined cavity to deform a deformation is particularly dependant variety of projectiles. The competing material into a fast moving projectile. on the geometry of the projectile. demands have lead to the use of The devices make use of two related The impact from a blunt projectile a composite of material types in effects; the Munroe effect and will cause the armour to form a dish, specific layers. The most common of the Misznay-Schardin effect. The while a sharp nosed projectile will these is the placing of a hard ceramic Munroe effect, discovered in the force material to flow radially from onto a ductile composites backing. 1880s by Charles Munroe at the U.S. the point of impact. This movement of Naval Torpedo Station at Newport3, material through plastic deformation High kinetic energy projectiles describes the focussing of a blast by absorbs a greater amount of energy The impact load of a projectile is the shape of an explosive. Munroe than if brittle failure of the armour proportional to the square of its famously demonstrated the effect material occurs1. An armour material velocity. As the projectile approaches by noting that the serial number is therefore also required to have high approximately 1200m/s the impact written on the explosive was etched ductility to effectively absorb energy. load becomes so great that the onto the surface of the target by Thin sheets of ductile armour material, materials begin to behave in a hydro- the explosion. The Misznay-Schardin particularly when impacted by a dynamic fashion. These high levels of effect states that when a sheet of sharp projectile, are likely to form K.E. are achieved in ballistic projectiles explosive is detonated, the majority radial cracks from the site of impact through the use of a casing known as of the blast energy is dissipated as the material is less constrained. a sabot. As the projectile leaves the normal to the larger surfaces. This reduces the amount of energy barrel the sabot is discarded leaving absorbed in the impact and goes a projectile with an extremely high some way to explaining why a K.E. density. In order to maximise single thick block of armour is more the K.E. density a large length to efficient than several thin layers. diameter ratio is required, hence Particularly blunt projectiles, with the collective name of long-rod low kinetic energy (K.E.) density, are penetrators. This has led to the use of known to produce a ‘plugging’ effect. fins to stabilise the projectile in flight In these cases large shear stresses once the sabot has been discarded. form at the edges of the blunt impact, leading to localised heating and causing thermal-shear instability2. This leads to the separation of a plug of material from the armour, forming a secondary projectile. It has been suggested that this occurs when the thermal softening effect generated by impact overcomes the heat capacity and work hardening of the material. CE

Figure 2 – Armour Piercing Fin Stabilised Discarding Sabot (APFSDS) projectile in fl ight DEFEN 148 Development of an armour mass 65 estimation tool for land vehicles

A precision manufactured shaped penetration. This effect is seen in Aluminium can also be more charge, such as that found in a Rocket projectile impacts as well as blast. susceptible to plugging, due to Propelled Grenade (RPG), consists By far the most effective method of its lower heat capacity per unit of an explosive with a conical cavity protecting against blast is to increase volume than steel. Titanium lined with a ductile material. The the stand-off distance from the point is also occasionally used as an explosion collapses the liner and of detonation. This is seen in vehicles armour material, but its cost has focuses the material into a fine fast purchased for recent conflicts, which prohibited its widespread use. moving jet. Typical velocities are in raise the floor of the vehicle off the All of the metal armour solutions the order of 8500m/s at the jet tip ground as far as possible. The recent so far are based on an energy 3 and 1500m/s at the jet rear . This vehicles also tend to incorporate a V absorption approach. In practice velocity difference stretches the shaped hull to direct the blast away absorbers are often coupled with a jet, eventually destabilising it into a from the vehicle. In addition there is disruptive outer layer. This aims to particulate stream. The relationship great potential for the use of false disperse the energy of the projectile between penetration and armour floors to prevent damage to the crew either by disrupting its motion so density for this threat is largely the through buckling of the belly plate. that it impacts with a greater surface same as for a high K.E. round. area, or deforming the projectile. Not all shaped charges are made to Armour types The most commonly used metallic the same precision as an RPG. Many disruptor/absorber system uses a of the Improvised Explosive Devices Metal sheet of High-Hardness Steel (HHS) (IEDs) found in current operations perforated at regular intervals, Rolled Homogenous Armour (RHA) make use of the shaped charge backed after an air gap by a block is the base-line steel armour for concept in a more basic way. of aluminium. The perforations in land vehicle. As the name suggests the steel add extra edge effects and the material has uniform properties Blast encourage the projectile to tumble. through its section. While armour The high hardness of the disruptor While the initial pressure wave from materials are often capable of deforms or shatters the projectile, a blast is capable of causing rapid superior protection levels, their reducing its K.E. density. This armour buckling of a vehicle’s facets, blast can effectiveness is still judged against system is aimed at addressing the cause damage to a vehicle through RHA. The obvious advantage of competing material property demands a variety of other mechanisms. For a RHA is that it is a relatively cheap described in the previous section. start, the blast itself is also generally and well understood material. As accompanied by fragments and a steel it also has the advantage Ceramic hot gases. If the integrity of the of experiencing work hardening vehicle fails due to either fracture around the site of an impact. This Ceramics are known to make effective or fragment impact, the hot gasses can result in a multi-hit performance disruptor materials due to their high cause significant damage to the better than other materials which hardness and low density. Their brittle interior of the vehicle. The impulse outperform it in single hit scenarios. nature means that they require a from the blast can result in the vehicle backing absorber material, with a overturning, particularly in cases of Aluminium is another commonly used monolithic metallic armour. composite often used to maintain partially buried explosives, where the low mass of the system. The the blast will be focussed upwards. Against most K.E. threats it has a better performance level than RHA, fragmentation of the ceramic due to A blast in very close proximity to however, due to its lower density an impact can significantly reduce the vehicle will form a compressive a greater thickness is required, its ability to withstand multiple hits. shock wave travelling through the increasing the size of the armoured This has lead to the design of smaller vehicle hull material. Due to the vehicle. The use of aluminium armour ceramic segments, in order to limit lower acoustic impedance of the air has been held back by instances the damage to the surrounding at the back face the wave is reflected of stress corrosion cracking until material. Unfortunately ceramic to form a tensile wave. The tensile improvements in the metallurgy solutions tend to be expensive, in stress can cause a scab of material and welding were developed. the region of 2-12 times the cost of to be thrown from the back face, an equivalent protection using RHA, producing a secondary projectile depending on the grade of ceramic4. within the vehicle without actual DEFENC E Figure 3 – Anti-tank weapon showing shaped charge in nose 149 65 Development of an armour mass estimation tool for land vehicles

to build an impenetrable tank. The Panzerkampfwagen VIII Maus weighed approximately 180tonnes, but unfortunately the power to weight ratio was such that it could only reach a top speed of 13km/h. As it has been shown impractical to fully protect a vehicle, an effective distribution of armour must be sought. This has led to the use of Directional Probability Variation (DPV) to determine the most likely directions of attack. The most famous of the DPVs is that developed by Lt Col. J M Whittaker in 19434, which assumed a vehicle travelling towards a line of enemy vehicles at a constant velocity. Whittaker’s assessment showed the majority of shots would impact on the front segments of the vehicle, Figure 4 – Warrior armoured vehicle fi tted with a mixture of bar armour and ERA up to an angle of approximately 60˚ Bar armour from normal. Such an approach is useful in conventional warfare, where Bar armour consists of a cage like Against a standard RPG an ERA the direction of the enemy is fairly set of bars with defined spacing, cartridge can perform as well as certain. However, its applicability in surrounding the vehicle at an offset. a steel plate many times heavier. more modern scenarios is less certain. The armour disrupts the correct However, ERA is still too heavy functioning of the rocket propelled for many light or medium weight Areal density grenade (RPG) rather than simply vehicles to use it extensively. Areal density, defined as the mass putting more material in its path. The threat defeat mechanism of material per square metre, is the This has the advantage of being an described above does mean that the term generally used to describe an extremely lightweight and cheap ERA performance is very dependent amount of armour. This is found method of protecting against a highly on the angle of incidence and the to be more appropriate than using penetrative threat, whilst being position of impact on the cartridge. A the overall mass, as it allows the highly resilient to multiple strikes. jet impacting the cartridge at normal protection levels of different sized Unfortunately the bar armour requires will find the ERA little more effective facets to be compared, while a relatively large offset from the than an equivalent mass of RHA. giving a greater appreciation of vehicle hull. This greatly increases both This is due to the plates’ directions the mass than using thicknesses. the space envelope and visibility of the of motion now being in line with the vehicle. Additionally, there are RPGs jet, no longer causing a slot to be Mass efficiency in existence specifically designed to cut or having a destabilising effect. bypass the bar armour’s effectiveness. Armour effectiveness is most To counter ERA, RPGs have been commonly considered using the developed which incorporate two Explosive Reactive Armour (ERA) general term of ‘mass efficiency’. shaped charges. The device is This is defined as the ratio of As a concept ERA is a counterintuitive carefully timed so that the first charge the areal density of RHA method of protecting a vehicle. The detonates the ERA, giving the second required to stop a particular threat, armour consists of two metal plates charge a clear path to the vehicle to the areal density of another sandwiching a layer of explosive, hull. Fortunately these threats are armour material required with the whole cartridge positioned currently not particularly prevalent. to stop the same threat. at an angle to the likely direction of impact. When a jet of copper from Armour concepts an RPG impacts the cartridge, the explosive is detonated driving the Whittaker weighting While it is convenient to use RHA plates apart. The plates’ movements as a baseline, the performance of at an angle to the jet cause the jet It is unsurprising that as the threats RHA and other armour materials to cut a slot where before it would to armoured vehicles have evolved, depends on an assortment of the amount of armour a vehicle variables including; angle of incidence, CE only have had to make a small hole. This increases the effective is expected to carry has increased material thickness and threat type. dramatically. Unfortunately, there thickness of material through which Taking angle of incidence as an is a limit to the weight a vehicle the jet must pass. The opposite example; for low K.E threats when a can carry before its functionality directions of travel of the plates also critical angle is reached, the geometry is impaired. This is illustrated well cause the jet to destabilise, greatly of the round will play a large part in reducing its penetration capability. by the German World War 2 effort DEFEN the impact. As the tip of the projectile 150 Development of an armour mass 65 estimation tool for land vehicles

contacts the surface a moment will required by the Survivability team. Division of problem be imparted causing the projectile to As described previously, the specific The problem of developing the tumble and so impact with a larger projectile impact mechanism is surface area. Obviously this leads to weight stack tool was divided dependent on a great many variables. into the three main areas of: a reduced penetration level but by However, empirical relationships, different amounts according to the being derived from test results, are (1) Defining the geometry of a armour material and threat geometry. specific to particular impact types. complex vehicle shape Armour performance is also known As such they will only function (2) Storing and accessing the to vary with the thickness of material, over a limited range of variables. mass efficiency information so that penetration through a number For instance Recht produced the (3) Combining the geometry of layers of material is not equal to separate equations for the residual and mass efficiency data the penetration which would be velocity of blast fragments and sharp to determine the mass of a achieved into a semi-infinite block. penetrators5. For the Survivability selected armour system. Put another way, a bullet with 15mm team’s purposes, many different RHA penetration capability might empirical relationships would have to Geometry take 4 closely spaced layers of 5mm be used according to the projectile RHA to stop it. Even taking this into type, impact angle, material type, and To define the vehicle geometry the account, the change in mass efficiency so on. This would require the precise user of the tool starts by entering the may still be hard to predict; when a detail of the impact to be understood 3D co-ordinates of the corner points thin plate is hit by a low K.E. projectile before a new projectile type or of the vehicle in an alphabetical list. at a high angle of incidence, the armour material could be modelled. In The geometry of the vehicle is then projectile would not be expected this way empirical relationships were defined in a facet by a facet basis, by to take a linear path through the considered too case specific for the selecting the pre-defined coordinates material. At some point the projectile team’s purposes. Recht’s equations of the corner points from drop-down would pass too close to the back face are by no means the only relationships lists. Each facet is defined by its corner for it to be contained. At this point produced. There are a great many points, assuming that straight lines the projectile would rotate towards conflicting empirical equations, each connect the points and that the facet the back face and burst through. with benefits and disadvantages. is flat, as tends to be the case with military vehicles. The normal vector of As discussed previously, there As with the empirical relationships the facet is determined by taking the are a wide variety of penetration there are a great many case specific cross product of two vectors formed mechanisms which vary with the competing analytical methods from the first three co-ordinates of threat and so the mass efficiency for assessing impacts. Analytical the facet. From the angle of this of a material is particularly threat methods are generally used to normal vector, the elevation and specific. Low K.E. projectiles can understand the specifics of the impact azimuth of the vehicle facet is then be expected to ricochet or gouge a mechanism at work and identify the determined. By applying a rotation scoop of material when impacting important variables when numerical matrix to the facet coordinates; the metal at a steep angle. However, a simulation is unavailable. For the facet is rotated so that it is only in one copper jet from an RPG is travelling purposes of assessing the weight plane. Now a 2D problem, Green’s at such speed when it hits the metal, or performance of concept armours theorem, which relates a closed area that its impact has been likened to a on a vehicle, standard empirical and to the path surrounding it, is applied hot knitting needle passing through analytical methods were considered to determine the area of the shape. butter. Clearly the angle is of no too case specific to be of use. consequence in the latter case. Weight stack Modelling This method does not limit the The solution arrived at by the number of sides which a facet can When attempting to predict Survivability team was to base the have, allowing complex shapes to the penetration of a threat, the analysis of armour solely on the be entered and analysed using their approaches generally fall into the experimental results determined corner point coordinates. While categories of numerical simulation by the Defence Science & this method determines the area, or empirical/analytical assessments. Technology Laboratory (DSTL). azimuth and elevation of a facet Numerical simulation is undertaken The team developed a ‘weight it does not indicate which side of using a dynamic analysis software stack’ spreadsheet tool to make the facet faces the inside or outside package such as Autodyne. While use of Dstl’s extensive knowledge of the vehicle. In other words the able to give a great understanding of mass efficiencies for armour method does not distinguish between of the impact mechanism and materials against the required the roof and floor of a vehicle. This provide accurate results, the process threats; effectively a fundamental poses a problem when threats are is computationally intensive and form of empirical relationship. defined as only impacting over extremely time consuming both to specific angles, as is the case with set-up and run. Such an approach mine blast, which should not be would take far too long to assess considered as impacting the roof. the number of combinations of To address this, the tool estimates armour, threat and impact angle

151 65 Development of an armour mass estimation tool for land vehicles

As the mass efficiency values are taken from experimental results they automatically take account of a variety of physical effects whether or not the precise mechanism is fully understood. For instance a bullet impacting steel at a steep angle may be subject to ricochet, may break through or may deform enough to cause plugging. It need not be understood which the dominant mechanism is, as the mass efficiency values are experimentally derived and so take account of all factors.

Applying information The first step in the process of determining the mass of an armour system obviously requires the user to specify the armour material. The vehicle geometry data is imported into the calculation sheet and the user selects a single armour material and a threat to be applied for each facet. With the threat selected, the calculation sheet imports the associated angles of incidence and RHA penetration levels from the Figure 5 – Complex vehicle geometry is built up from simple shapes threat data sheet. The worst case angle of impact is determined as the centre point of the vehicle as DSTL provided mass efficiency values the angle closest to normal at which the midpoint of the maximum and dependent on threat type, armour the threat is capable of impacting minimum coordinates in each plane. type, impact angle and material a facet. The worst case impact The position of the facet relative to thickness based on experimental elevation ( χ ) is therefore defined as: the centre point then defines which results. The information is stored side faces outwards. This works in the tool as a series of lookup Where: adequately as long as the vehicle is tables displaying mass efficiency a standard shape. A manual override values varying with angle of attack function is included for particularly and thickness of material. Each unusual vehicle shapes to be analysed. combination of threat and material The importance of the Whittaker type has an individual lookup table. weighting system was discussed Bilinear interpolation is used to With the worst case impact earlier. In assessing concept armour estimate the mass efficiency at a azimuth ( β ) similarly defined, the systems it is desirable to be able to particular angle and thickness. While vector of the impact at the worst move the line defining the frontal arc taking a relatively large space to case elevation and azimuth is division of the vehicle. For example, store the information, this method determined by trigonometry as: the question might be posed as offers a great deal of flexibility. New to how much weight is saved by materials can be included with rough moving the crew section forward. estimates of effectiveness and as A feature was included to allow the more information becomes available From this the angle from normal user to define a moveable front/ through testing the mass efficiency of the worst case impact, rear dividing line, rather than force data can be updated accordingly. combining elevation and the user to specify two sets of facets azimuth, is determined as: for front and rear. The geometry sheet automatically forms separate front and rear facets from each facet intersected by the dividing This impact angle is used to determine line. This is done simply by linear the thickness of RHA ( ) required tRHA interpolation in three dimensions, to stop a threat with penetration allowing the front and rear of a facet P(mm of RHA) using the cosine rule. to be armoured independently.

152 Development of an armour mass 65 estimation tool for land vehicles

In practice, the cosine rule does The mass efficiency of each pre- Conclusion not apply exactly to calculating existing layer must therefore be penetration at an angle, due to the recalculated for subsequent threats. The weight stack tool described in this effects, such as ricochet, mentioned Once the penetration reduction paper provides the Survivability team previously. However, the mass due to the pre-existing layers is with a rapid method of assessing the efficiency values are set relative to the subtracted from the penetration mass of complex concept armour performance of RHA at normal and so of the threat, the mass of armour systems and evaluating survivability these effects are inherently accounted for the new layer can be calculated requirements, incorporating a variety for. The calculated thickness of RHA as before. An adjustment factor is of impact mechanisms through and the impact angle are then cross added to the penetration capability the use of experimental data and a referenced with the material type and of the threat according to the layer function to compare modular threat type to determine an effective number of extra layers in place. armour to monolithic systems. mass efficiency value from the stored This accounts for uncertainty in the The tool has been used to explore mass efficiency data. The thickness residual penetration capability, once how the FRES protection requirements of RHA is converted into an areal a threat has passed through a layer. could be achieved and what the density of RHA by multiplying by the The thickness( ) of the new layer impact of the necessary weight of tχ density. Finally the areal density of of armour is therefore found by: armour would have on the vehicles armour required to stop the projectile mobility. It has been concluded that impacting a facet is determined by it is impractical to provide an armour dividing the Areal density of RHA solution that could defeat all the by the determined mass efficiency. Where: identified threats simultaneously. An overall value of armour mass This has subsequently driven a can be determined by multiplying great deal of analysis to see how the areal density by the facet area. the threat distribution changes with the type of operation. In turn it has resulted in “mission tailored” protection requirements which Once this calculation has been The capabilities required in order to reflect the changing balance of performed for each facet subject assess modular armour, reassessing threats and better represent the to the defined threat, a new threat mass efficiency for each layer against user’s needs. Recently the tool is selected by the user and the each threat, also provide the capability has been used in conjunction calculations repeated. The tool to use layers of different material with armour cost information to compares the mass of material types. This is a useful function when investigate options for “value for required for each threat on a defining ‘top-up’ armour, where money” protection solutions within facet by facet basis, showing the a different material which is more the constraints of the vehicle. increase in armour mass required efficient against a particular threat is for protection against increasing placed on top of the standard armour. levels of threat severity and giving Despite the flexibility of this model, a cumulative armour mass. some complex forms of armour do However, the description above is not lend themselves to the mass only covering the most basic case efficiency approach. This includes of single module, single material bar armour, which due to its very solutions. In this case it would not be threat specific nature provides no possible to alter the level of protection protection against other threat types. offered to the vehicle unless the ERA also poses a problem due to entire armour pack was removed its varying angles and thicknesses. from the facet and replaced with a Fortunately these armours are new one. For a modular approach generally only available in standard where the protection level can be forms, where their capabilities are altered, the armour must consist of well understood. They are therefore a stacked set of layers of material. applied as Areal densities without The tool therefore allows the user to assessing their effectiveness and so define at which threat levels there are the tool requires a certain amount divisions between armour layers. To of armour experience from the user. enable this, the contribution of the By using a spreadsheet as a basis, lower armour layers to reducing the the tool can be easily modified DEFENC penetration of a more sever threat or expanded and has an inbuilt needs to be assessed. Unfortunately capacity for growth, as the armour the existing thickness of the lower material and threat data sheets can level cannot simply be subtracted be either added to or updated with from the total thickness of armour more recent experimental results. required to stop the more severe

threat, as the mass efficiency is E known to vary with layer thickness. 153 65 Development of an armour mass estimation tool for land vehicles

References 1. Energy and momentum changes during ballistic perforation. Hetherington, J. G. 3, s.l. : International Journal of Impact Engineering, 1996, Vol. 18. 2. Penetration: Mechanical aspects and mathematical modelling (review). Aptukov, V. N. 2, s.l. : Strength of Materials, 1990, Vol. 22. 3. Payne, C. M. Principles of Naval Weapon Systems. s.l. : Naval Institute Press, 2006 . 4. Hazell, P. J. Ceramic Armour: Design, and Defeat Mechanisms. s.l. : Argos Press Pty, Limited, 2006. 5. The effect of projectile nose shape upon ballistic limit velocity, residual velocity and ricochet obliquity. Recht, R. F. s.l. : Weapons Development Department, Naval Weapons Center, China Lake, California, 1973. CE DEFEN 154 Explosives safety - An oxymoron? 66

Chris Marriott Abstract Senior Safety A history of the development of explosives, and related accidents, provides Reliability Engineer a background to the practical application of safety assessment techniques Defence for projects involving explosives systems. The analysis techniques include ‘fault tree’ and ‘event tree’ analysis, which have been used qualitatively to help develop ‘lines of defence’ arguments. These have been subsequently developed using a graphical technique termed ‘bow tie’ analysis. The development of ‘safety functional requirements’ and ‘safety integrity levels’ is also described for equipment which will be used in explosives processing. Finally, the difficulty in determining the consequences of an insult to explosives is discussed, together with the implications for ‘quantitative’ risk assessments. This paper aims to demystify the terms used above by demonstrating how they have been used in practical applications. The opinions herein are those of the author and not necessarily those of Atkins Limited.

Introduction A history of explosives If a material is said to be ‘explosive’ and related accidents then it can be considered to be The history of explosives is closely The first recorded use of gunpowder chemically or otherwise energetically linked with tales of tragedy, either by English soldiers was at the battle of unstable and liable to release from the misery it can cause in Crecy in 1346. However, even by the energy in a sudden and often its intentional application or from 15th Century, early weapons could violent manner. This is likely to be accidents during its development, cause as much harm to the user as accompanied with the generation manufacture, transportation and the intended target, as James II of of high temperature, large changes storage. Ironically, the discovery of Scotland discovered when he was of pressure, a flash and loud noise. the ingredients for the earliest known blown up by his own cannon whilst The level of such an energy release forms of explosive originated from a besieging Roxburgh castle in 14604. can be quite staggering: In 1917 the Chinese attempt to find an elixir of The manufacture of such a lethal City of Halifax, in Nova Scotia, was life! In 850 AD, a Chinese book warns devastated when a French cargo substance was also a risky business. of a particular elixir: “Some have The three components had to be ship exploded in its harbour. Over heated together sulphur, realgar and 1600 people were killed instantly ground together intimately to be saltpetre with honey; smoke & flames effective and at first this was done and all buildings and structures result, so that their hands and faces within a vast area were devastated1. with a mortar and hand-operated have been burnt, and even the whole pestle. Later, ‘pestle mills’ were Safety, on the other hand, is typically house where they were working used, the mixture being pounded 2 defined as “the condition of being burnt down” . The ingredients of mechanically in wooden mortars by safe; freedom from danger, risk this potent mixture include two of wooden stamps. Although the use or injury”. The term ‘explosives the three ingredients of what is now of wood avoided the generation of safety’ therefore fits quite neatly known as ‘gunpowder’, namely sparks, to which the powder is very into the definition of an oxymoron, sulphur and saltpetre, with the sensitive, this was still a hazardous i.e. “a figure of speech in which third ingredient being charcoal. The operation because the impact and incongruous or contradictory terms earliest known weapons employing friction from the pestle often caused are combined”. However, the need this mixture were probably Chinese the gunpowder to explode. In Britain, for safety when handling explosives gunpowder fire arrows reported stamp mills started to be replaced 3 is far from contradictory and it is rare in 969 AD , but it took a couple by ‘incorporating’ (or rolling) mills in for the word ‘explosive’ to appear of centuries before knowledge of around 1740. An incorporating mill DEFENC without some reference to the gunpowder was transferred to Europe consisted of two heavy wheels which word ‘safety’ in close proximity. This with a description by Franciscan monk ran over a bed of partially mixed paper provides a brief history of the Roger Bacon appearing in 1242. gunpowder and ground it together development of explosives and their for a period of up to 8 hours, without attendant lessons in safety, and then subjecting it to any severe shocks5. goes on to show how modern safety cases are developed for processes

and equipment which, by their very E nature are designed to be explosive. 155 66 Explosives safety - An oxymoron?

The requirement for such relatively large scale, ‘industrial’ processes, was driven not only by military requirements but also more peaceful applications in the field of civil engineering. The earliest record of the use of gunpowder for blasting occurred in 1627 and it was being used in the Derbyshire copper mines in 16386. Its first use in large scale engineering was the construction of the French Languedoc Canal in 1681 and over the next couple of centuries many of the canals, road and railways were all constructed with the aid of explosives. This was not without the attendant risks associated with explosives: by the time Isambard Brunel had completed the “monstrous and extraordinary, most dangerous and impracticable tunnel at Box”7 it had cost over a hundred lives. Even more perilously, explosives started to be used for underground blasting in around 1820. This was short-lived however, due to numerous catastrophes, and a Royal Commission on Accidents in Mines recommended that from 1887 it be replaced by specially blended and tested explosives.

The pressure for special types Figure 1: Nitroglycerine manufacture – the operator sat on a one-legged stool of gunpowder to meet the ever so that he would not fall asleep and let the mixture overheat. increasing demands of miners, civil engineers and the military led in the 1867 he patented the idea of taming The next major development in what 19th century to replacements for liquid NG by absorbing it into a solid were to be termed ‘high explosives’ gunpowder as an explosive. Christian (diatomaceous earth or ‘Keiselguhr’) was the introduction of ‘Lyddite’ Schönbein was credited with the to make a dough. He named the (based on picric acid) as a filling for invention of ‘guncotton’ (made from substance ‘dynamite’, from the Greek explosive shells in the British Army, nitrocellulose and cotton wool) in word for ‘power’. Production of first used at Omdurman in 1898. 1846. It was more powerful than gun the NG itself remained a hazardous It was used up until 1916, despite powder and was one of the first of operation, as shown by the operator’s its relatively poor performance the so called ‘smokeless powders’. seating arrangements in Figure 19. and the many accidents associated Also, in the words of Schönbein, “The Whilst dynamite was a major with its manufacture, including: manufacture is not attended with the step forward, it suffered from the 16 deaths at Woolwich Arsenal 8 least danger…” . However, in 1847 dangerous problem of ‘sweating’, i.e. in 1903 following the forcible the first ever guncotton factory blew the leaking of unstable nitroglycerine insertion of a former in 9.2” shells; up and 21 people were killed, possibly from the solid matrix. To resolve 10 deaths in Heckmondwike due due to overheating in the drying this, Nobel invented ‘gelignite’ to a foreign body in the grinding process. An even more powerful (or ‘blasting gelatin’) in 1875. mill; and 38 deaths in 1916 at Low explosive, nitroglycerine (NG), was Gelignite is an explosive material Moor munitions factory, when a fire first made by Ascanio Sobrero in 1847 consisting of ‘collodion-cotton’ (a in drum of picric acid spread to a 10 by treating glycerine with sulphuric type of nitrocellulose or gun cotton) packing building . These deficiencies and nitric acids. He discovered its dissolved in nitroglycerine and mixed accelerated the introduction of the physiological effects when he tasted with wood pulp and sodium or eponymous ‘TNT’ (trinitrotoluene), a sample and it gave him palpitations potassium nitrate. Its composition a substance considered so stable and weakness in his limbs, he during its manufacture and storage CE makes it easily moldable, and safe discovered its explosive effects to handle without protection, as that it was exempted from the 6 when a small sample exploded glass long as it is not near anything 1875 Explosives Act . TNT was, fragments into his hands and face. Its capable of detonating it. One of however, very poisonous and led to exploitation as an important explosive the cheapest explosives, it burns the deaths (from jaundice) of many only arose when Alfred Nobel found slowly and cannot explode without a female workers in the munitions a way to bring it under control. In factories of the Great War.

DEFEN detonator, so it can be stored safely. 156 Explosives safety - An oxymoron? 66

Between the wars, more powerful Explosives - definition of terms Low explosives, such as those used explosives were developed such in fireworks, can also detonate as RDX, which was first used in When explosive materials explode under certain circumstances, the 1890’s as a medicine by Hans they produce a large volume of hot sometimes in tragic circumstances: Henning. When combined with gases and a consequent large increase In May 2000, a fireworks factory in TNT it provided a stable, relatively in pressure. However, the effects of a Enschede exploded, destroying an ‘insensitive’ munition. When ‘low explosive’, such as gunpowder entire neighbourhood and killing combined with jellies, waxes and compared to a ‘high explosive’ such 23 people. This was considered to plasticizers the RDX forms what are as TNT are on different scales: Low be due to the self-confining effect known as ‘plastic explosives’, which explosive can produce pressures up of a large mass of material. In the are malleable, very stable and store to 600MPa in milliseconds; a high same month, an accident at a public well (for example, the notorious explosive can produce nearly 50 fireworks display in Australia led to Czech material known as ‘Semtex’). times the pressure in a thousandth of the death of a young girl when a The most ‘powerful’ explosive the time (i.e. at supersonic speeds). roman candle detonated in the steel currently in large scale use is HMX, Technically, a low explosive is said to tube in which it was standing, again which was discovered in the 1930’s as ‘deflagrate’, or burn rapidly, whereas due to the effects of confinement. a by-product from the manufacture a high explosive is said to ‘detonate’ This led to the introduction of the of RDX. HMX is a common ingredient (from the Latin ‘to thunder’). This Queensland Code of Practice in 2002, in Plastic Bonded Explosives (PBXs), rapid shattering effect, for which the which had specific aims to minimise which are powdered explosives to term ‘brisance’ is employed, makes risks at all public firework displays. which plastic binders have been high explosives useful in mines, added. The manufacturing process for bombs and shells, whereas low Explosives regulation a typical PBX explosive in a modern explosives are more suitable where a In Britain, the first Act of Parliament weapon system requires the powder steady thrust, or ‘heave’, is required to deal with public safety in relation to to be oven dried at temperatures in as in gun propellants or mine blasting. gunpowder was that of 1719, which excess of 100ºC (the powders are Explosives are also commonly divided regulated the carriage of gunpowder wetted for transport and storage into ‘primary’ and ‘secondary’ types, in London. This was followed in to prevent explosive dusts clouds where ‘primary’ refers to a detonating 1772 by the Gunpowder Act, which from forming), then isostatically explosive which is extremely sensitive covered manufacturing processes pressed (i.e. subject to equal pressure to stimuli (e.g. heat, friction, shock), (prohibiting pestle mills and limiting from every side) at pressures up to and ‘secondary’, which refers the quantities of powder)5. In the mid 140MPa11, before the final shape is to a detonating explosive with 1800’s, amidst the discovery of new machined using standard cutting, relatively low sensitivity. Generally, explosive chemicals, almost anyone milling and drilling machines. Thus an ‘explosive train’ would consist could make and sell explosives and the explosive is exposed to every type of a primary explosive as initiator this inevitably meant that there were of hazard and is expected to survive (i.e. a ‘detonator’), a ‘booster’ and increasing numbers of accidents; a with the least sign of reaction and yet a secondary explosive as the main particularly severe explosion occurred will detonate reliably on demand. charge. The ‘booster’ translates the weak detonating wave from the at a Birmingham factory in 1875, More than 8000 people are initiator and amplifies it to initiate the killing 52 people. This incident, reported to have been killed in the main charge. This is important for amongst the many others, led to the 20th Century due to the result of modern munitions, where the main passing of the Explosive Act in 1875 explosives responding to unplanned charge is designed to be as insensitive (extended in 1923) and established stimuli such as heat or shock. As a as possible to increase safety in the Explosives Inspectorate. From result of this, and recent advances in manufacturing, transport and use. 1847 to 1974, every explosives related explosives chemistry, there is an ever incident was published in the Annual Low explosives are generally referred increasing drive for the employment Report of Her Majesty's Inspectorate to as ‘pyrotechnics’ or ‘propellants.’ of ‘Insensitive Munitions’, defined of Explosives. The Health and Safety The former is derived from two Greek as: “Those munitions which reliably Executive maintain a database (EIDAS words: ‘fire’ and ‘art’ and fireworks fulfil their performance, readiness -Explosives Incidents Database are a popular example. A pyrotechnic and operational requirements on Advisory Service) which includes these produces a self-sustaining combustion demand but will minimise the records, as well as those reported to reaction which may or may not be violence of reaction and subsequent HSE since 1974 or drawn from the accompanied by a visible flame and collateral damage when subjected public domain. There are currently the production of gas. Compositions to unplanned stimuli”12. It is to be nearly 9500 incidents recorded are mixtures of a fuel (e.g. carbon) 10 hoped that this particular directive on the database . Figure 2 shows DEFENC and an oxidising agent (e.g. is pursued with due vigour. particular information drawn from this potassium nitrate). Propellants can database, relating to major explosives be pure materials (e.g. nitrocellulose) incidents in which 5 or more people or mixtures of chemicals designed died. Note the series of catastrophic to burn at a controlled rate and events in the Great War period due to provide a predefined thrust to explosions in factories producing NG, the system containing them. TNT or picric acid, and the reduction E in such incidents in modern times.

157 66 Explosives safety - An oxymoron?

Figure 2 - Explosives related incidences with >5 fatalities. Explosives - safety testing The database contains such In 1887 the Coal Mines Act The manual provides a compilation of tragedies as the incident in 1909 established a series of tests for those sensitiveness and explosiveness where attempts to demolish a explosives to be used in coal mines. tests that are used to evaluate the wreck by explosives means led to a These tests included the ‘Trauzl test’ risks from military explosives during tremendous explosion and caused (later the ballistic pendulum) to development, manufacture, handling, 6 deaths. It later transpired that the measure the power of the explosive filling, transport, storage and use, wreck was that of a ketch loaded and a test to measure the explosive and to provide definitive methods with gelignite! The report by the effect on typical coal mine gases. for the tests. Traditionally it has been inspector suggested that “…before The Sensitiveness Collaboration assumed that explosives intended to commencing the work of destroying Committee (SCC) was established at fulfil a specific function possessed a wreck by means of explosive it is Waltham Abbey in 1962 to produce broadly similar safety characteristics, desirable to ascertain, so far as may procedures for the safety tests i.e. the designations ‘propellant’, be, the nature of the cargo.”13 applied to energetic materials, and ‘high explosive’, ‘pyrotechnic’, Legislation such as the Gunpowder this resulted in the SCC Manual of ‘initiator’ or ‘booster’ could imply Act & Explosives Act have been Tests. Ownership was transferred to various degrees of potential hazard. replaced by the Manufacture and the Defence Ordnance Safety Group However, the search for ever Storage of Explosives Regulations (DOSG) in 2005 and the manual improved performance has blurred (MSER) 2005 and, for UK military was re-issued under the Energetic the distinction, in hazard terms, applications, by the Joint Service Materials Testing and Assessment between explosive materials. This is 14 Publications (JSPs). A key requirement Policy Committee (EMTAP) . shown by the overlapping regions of MSER is that premises must in the figure below, which shows be licenced and that separation typical regions in the sensitiveness/ distances between explosive sites explosiveness plane occupied by (also known as ‘Quantity Distances’) the various types of explosives. must be maintained, according to the ‘hazard type’ (ranging from hazard type 1 for mass explosive hazard to hazard type 4 for fire or slight explosion). This aligns with the UN recommended system for the transportation of dangerous goods, i.e. class 1.1 is equivalent to hazard type 1. Other key regulations include the Control of Major Accident Hazards regulations (COMAH) 1999, made under the Health and Safety at Work Act (HWSA) 1974, which aims to prevent major accidents involving dangerous substances and limit the consequences should they occur. The military equivalent is the Major Accident Control Regulations (MACR).

Figure 3 - Explosiveness and sensitiveness 158 Explosives safety - An oxymoron? 66

Explosives risk assessment It may appear that explosive safety is concerned with solely technical issues and is achieved by compliance with rules and regulations. Lessons from the many disasters involving explosives, however, suggest this is not the whole picture. Prescriptive regulations are accepted for routine Figure 4 - Possible nodes for a fi rework display situations and well-established practices; it is very difficult to write What can go wrong? rules for unpredictable explosive materials and impracticable to be able The start point for any assessment Thus for a major public fireworks to cover every eventuality. Thus, in the of risks is the hazard identification display on a barge in Bristol docks, formation of a risk assessment, some process, or ‘HAZID’, which will also the nodes may look something basic questions need to be answered: define responsibilities (duty of care) like that shown in Figure 4. and interfaces with other safety (1) What can go wrong? The associated guidewords may assessments. There are many ways be generic (e.g. ‘more of ‘, ‘less (2) What is the likelihood & of conducting the identification than’, etc.) as would be used in a who can be harmed? process, including ‘Structured What-If HAZOP, or more topic specific, as in (3) How can we reduce this? Techniques’ (SWIFT) and ‘Hazard a SWIFT analysis. A typical hazard The results of this analysis are then and Operability Studies’ (HAZOP), to log will consist of a unique hazard tested by assessing whether all name two of the more recognised reference number, a description of practicable control measures have methods. However, essentially the cause and consequence and been taken (and that industry practice they rely on the same principle of a list of any existing safeguards. has been followed) and that further ‘structured brainstorming’ with an Safeguards relevant to fireworks action is grossly disproportionate appropriate team of knowledgeable include ‘engineering safeguards’ to any benefits gained from the people. Regardless of the method, (e.g. separation distances, racks for risk reduction. This test is generally the schedule of hazards identified mortars, and lightning arrestors) referred to in the UK as ‘As Low As should be as comprehensive as and ‘procedural safeguards’ (e.g. Reasonable Practicable’ (ALARP). possible. Structuring consists of firework quality assurances and laying out the ‘nodes’ or key break operator training). An example points in a process and working spreadsheet is shown in Figure 5. through a list of guidewords to prompt the team to identify any hazards associated with that node and guideword. This data is then recorded in a ‘hazard log’, which may be in a spreadsheet or database format. DEFENC E Figure 5 - Example hazard log sheet 159 66 Explosives safety - An oxymoron?

What is the likelihood & who can be harmed? The next stage is to carry out risk For a more quantitative risk FTA is used to graphically display the estimation, which entails assessing assessment (termed ‘QRA’), other means by which faults can lead to a the severity and frequency of the techniques are applied to provide a failure of the system of interest (the hazardous event. For a qualitative more systematic analysis of the risks. ‘top event’). Thus, in the development assessment, a risk matrix is a useful These include: Failure Modes Effects of an explosives manufacturing plant, way of ranking and displaying and Consequences Analysis (FMECA), a series of trees would be developed the results. This is generally a which is generally more suitable to the for each of the different operations grid with frequency down one analysis of complicated component (e.g. receipt, storage, drying, side and severity along the top, failures; Fault Tree Analysis (FTA) pressing, machining and inspection). ranked from high to low. and Event Tree Analysis (ETA). A fault tree describing explosive pressing could look like Figure 6. Taking fault 3.4 ‘over heating’ further leads to the fault tree shown in Figure 7, where the bottom level events derive from the hazard log. In this example, the gate with descriptor ‘N3_F2’ represents a ‘Line of Defence’ (LOD). To qualify as a LOD, robust justification is required to show that the safeguards (engineered features and/or managerial controls) provide adequate protection against the event under consideration from occurring15. These safeguards then become formal Safety Functional Requirements (SFRs). The SFRs identify Figure 6 - Fault tree for explosive pressing specific requirements on systems and structures which make up the LOD, e.g. “The thermal trip shall isolate the heating medium should the heat exceed 102°C”. There may be one or more LODs required to terminate the fault sequence arising from a hazard, depending on its severity and the level of protection afforded by the SFRs within the LOD. If these systems are programmable, then a ‘Safety Integrity Level’ (SIL) may also need to be assigned. A SIL is a relative level of risk-reduction provided by a safety function. Four SIL levels are defined, with SIL 4 being the most dependable and SIL 1 the least. The international standard IEC 61508 defines SIL using requirements and IEC 61511 is an application specific adaptation for the process industry sector (including explosives manufacturing). This defines the ‘Layer of Protection Analysis’ (LOPA) methodology to be used in the assessments. This is a quantitative method (as opposed to a LOD, which is descriptive) which requires CE an understanding of the probability of failure of a system and, crucially, an understanding of the potential consequences in the event of failure.

DEFEN Figure 7 - Fault tree for overheating during explosive pressing 160 Explosives safety - An oxymoron? 66

How can we reduce this? Once the risks have been ranked and prioritised, they can be systematically assessed and risk reduction measures put in place, with the objective of meeting the previously set safety level criteria. Risk reduction is generally approached in a logical sequence, as shown in Figure 10. Clearly a duty of care is owed to the public in the siting and management of any explosives facility and this is achieved through the enforcement of Quantity-Distance (Q-D) rules. These relate the quantity of explosives and the distance necessary to limit the severity of effects, in the event of an accidental ignition. Considering Figure 8 - Event tree explosive initiation during pressing workers on-site or actually involved in explosives manufacturing, elimination of the hazard may only Where hazards are significant Finally, the results from FTA and ETA be achieved by a complete change and SIL requirements too can be usefully displayed on a ‘bow- to the process and may require some onerous, remote operation may tie’ diagram, which is a representation lateral thinking (e.g. remove the need be the only viable option. of all the initiators and consequences for explosive machining by pressing Explosive pressing is deemed to of a particular scenario and the LODs directly to the final shape). Reducing 16 be sufficiently hazardous that it is in place to prevent or mitigate them . the likelihood of an explosive event required to be carried out remotely17, Figure 9 considers the controls in could be achieved by reducing thus the consequences for other place to prevent hazards identified process temperatures or pressures, worker of an explosive event in the in the press cell leading to fatalities. but these are generally already at a cell are related to the effectiveness minimum. Reducing the effect can of interlocks, as shown in Figure 8. be achieved by reducing quantities processed, but this will have an effect on throughput. Isolation can be achieved by removing the operator to a safe place, as is recommended for machining and pressing operations17. The design of controls (making up the LODs) will depend on an analysis of the consequences of ignition8 or initiation9. A clear understanding of the nature of the explosive hazard in terms of its likelihood and its potential for harm to people is required so that any engineering provisions are robust and can tolerate a wide range of faults. DEFENC

Figure 9 - Example bow-tie E

161 66 Explosives safety - An oxymoron?

Figure 10 - Risk reduction hierarchy Consequence analysis By the very nature of explosives, However, an understanding of Intrusion into the explosive by the effects of accidental ignition or the consequences of the response ‘spigots’ or pinching between hard initiation are largely immediate and to an insult is necessary so that surfaces can exacerbate this heating entail severe consequences. The proportionate effort is applied in the by providing additional confinement, magnitude is largely determined by risk assessment. For example, impacts enabling the hot spots to coalesce the quantity of explosives material to an explosive can occur at any time and initiate a runaway reaction involved, while the type of effect is during processing operations, either within the explosive. The violence largely determined by the hazard intentionally as when machining, of the response depends on: type. A quantitative assessment or accidentally because of drops or • Severity of impact (i.e. of the probabilities of reactions of mechanical/ control system failures. energy available). explosive to various stimuli is very Depending on the severity of the • Transient confinement (i.e. as a difficult to achieve. Usually there impact, this may lead to a localised result of the accident, such as are insufficient tests results (for temperature rise and the production crushing from a blast door). the particular stimuli/ explosive of ‘hot spots’ as the mechanical combination) to give statistically valid energy is converted into heat. • Degree of damage of estimates, due to the cost of testing. the explosive. • Reactivity of explosive. • Mechanical strength of explosive. • Close fitting case materials if present (providing confinement and a source of fragments). Thus, for hazard assessments involving conventional explosives there is little benefit in carrying out probabilistic risk assessments when such great uncertainties exist in the data. It is considered that a rigorous assessment based on strong qualitative arguments will yield a better understanding of the hazards and their control15. It is acceptable to use a ‘relative ranking’ rather than a specific probability figure, grouping consequences into, for example, ‘partial reactions’, ‘violent reactions’ and detonations, against which the consequences for operators can be assessed. Some guidance is provided in the form of ‘typical’ effects from given quantities of explosive, as shown in Figure 1119. CE

Figure 11 - Explosive effects within 6x6m building DEFEN 162 Explosives safety - An oxymoron? 66

Conclusions This paper has shown how the standard risk assessment methodologies developed in the process industries are applied to those processes involving explosives. A number of the terms used, both in risk assessments and in explosives processing, have been demystified by demonstrating their application in a number of practical applications. The effect of applying such a systematic process has had a significant effect on the safety of the industry, as evidenced by the reduction in incidences involving explosives. Thus it is concluded that whilst ‘explosives safety’ appears to be an oxymoron, it is nevertheless an achievable goal. Acknowledgements Thanks are due to Richard Gough for proof reading this paper and providing invaluable comments. DEFENC E

163 66 Explosives safety - An oxymoron?

References 1. Nova Scotia Archives & Record Management, “Halifax explosion 1917 online resources”, http://www.gov.ns.ca/nsarm/virtual/explosion 2. Kelly J, “Gunpowder, A history of the explosive that changed the world”, Atlantic Books, 2004 3. Davies N, “Pyrotechnics handbook”, Cranfield University, (2001). 4. Brown GI, “The Big Bang: A History of Explosives”, Sutton Publishing Ltd, (2005) 5. Crocker G, “The Gunpowder Industry”, Shire Publications, (2002) 6. Rolt L, “Isambard Kingdom Brunel”, Longmans, Green & Co., p. 136, (1957). 7. Holmyard E, “Makers of chemistry”, Oxford University Press, p. 189, (1931) 8. Copyright ICI Archives 9. HSE, “Explosives Incidents Database Advisory Service”, http://webcommunities.hse.gov.uk 10. Cooper PW, “Explosives Engineering”, Wiley-VCH Inc., p.53, (1996) 11. Izod D, “Ammunition Technology Volume 1: Introduction”, Cranfield University, (1996) 12. Cooper-Key A, “Circumstances attending an accident which occurred on February 1st 1909, in connection with the removal of a wreck…near Caister”, Home Office Report, (1909) 13. EMTAP, “Manual of Tests”, Defence Ordnance Safety Group, (2005) 14. Gilbert SG, “Assessing hazards from explosives as part of explosives facility safety cases”, 27th DOD 15. HSE Information sheet, “Guidance on risk assessment for offshore installations”, OIS 3/2006, (2006) 16. Explosives Industry Forum, ”Requirements for remote explosives manufacturing facilities”, HSE/CBI Explosives Industry Group, (2005) 17. Merrifield R, “Safe handling requirements during explosive, propellant and pyrotechnic manufacture”, HSE Specialist Inspector Report No. 31, (1991) E C DEFEN 164 Papers 051 - 066 Papers 051 - Technical Journal Technical 04

Technical Journal PAPERS 051 - 066 04 Printed on 9Lives Offset, part of the family Printed on 9Lives Offset, of ’born again’ papers, manufactured with National Association of Paper Certification (NAPM) using Merchants pulp, fibre 100% recycled approved the ultimate in offers 9lives Offset management. The environmental manufacturing mill and printer have also with the internationallybeen accredited ISO 14001. standard recognised © ATKINS Ltd except where stated otherwise. The Ltd except where © ATKINS logo, the “open A” device and the strapline ATKINS Ltd. trademarks of ATKINS “Plan Design Enable” are