IAHR AIRH Newsletter 2 Volume 20 / 2003 (Supplement to JHR - Vol 41 - No 2)

Investigation of extreMe flood International Processes And unCerTainty – Journal of IMPACT River Basin IMPACT is a major three- year European research Management project looking at the failure of flood defence embankments and JRBM embankment dams co- ordinated by IAHR Corporate Member HR Wallingford. More information on page 29

Breaching a 6m high embankment in Norway (Oct. 2002)

IMPORTANT Roman aqueducts NOTE: The NOVATECH 2004 Dr. P.-L. Viollet and Dr. abstracts submission H. Chanson review the deadline (see leaflet history of aqueducts in within this mailing) the section ‘Chronicles has been extended on the history of to April 28th, 2003 hydraulics’ on page 26 for IAHR Members, instead of March Arches of the Anio Novus 28th, 2003 as , in its upstream A new Journal, JRBM - the International indicated in the course, between Tivoli and Subiaco (photo: P.L. Journal of River Basin Management - will Viollet) leaflet. be launched at the Third World Water Forum in 2003 with a specific role to provide a forum for the scientific community to adopt a more cross-sectoral approach to research and practice in river basin management. The new Journal will be published by IAHR in collaboration with other leading international associations. See page 20 for further information on this new Journal.

Keep up-to-date with the latest Congress news. Continues on page 21

17 IAHR IAHR AIRH Newsletter AIRH Newsletter

and climbing back up the slope to the Fes (Morocco) is known to have had HODGE, A.T., Roman aqueducts and water downstream reservoir. Some of these conduits delivering pure water to all supply, Duckworth, 1995 reservoirs can still be seen, including some houses of the al-Qarawwiyyin town area, VIOLLET, P.L., L’ Hydraulique dans les of the holes from where the lead pipes as reported by the medieval geographer civilisations anciennes, Presses des Ponts departed (see picture). al-Idrissi. et Chaussées, 2000 When the collapsed in the West, the aqueducts slowly Bibliography. deteriorated, through lack of maintenance ARGYROPOULOS P. A., “Ancient Greek (see the table in the following article by hydraulic engineering experience and Are YOU interested in the history of hydraulics? Is there any study or report in this area that YOU would like to publish in our new Chanson); only in Rome, the Popes used technique”, IAHR Congress, Cagliari, Sept newsletter section ‘Chronicles on the history of hydraulics’? Please don’t hesitate to contact Dr. Pierre-Louis Viollet at: to maintain some of the aqueducts for the 1979 [email protected] needs of waterwheels. In the Western CHANSON, H., “Hydraulics of Roman middle ages, people in cities returned to aqueducts : steep chutes, cascades and the rivers for their domestic needs, and dropshafts”, American Journal of Aqueducts : a short history many hygiene problems followed. In the Archeology, 104, p 47-72, 2000 By Pierre-Louis Viollet, Arabic world, the Roman idea of water in GARBRECHT, G. H., “Die Entwicklung der EDF R&D the city remained: for instance Samarkand Wasserwirtschaft Pergamons bis zur The upstream reservoir of a siphon of the Gier [email protected] kept its aqueduct systems operating until Romischen Kaiserzeit”, Journées d’Etude aqueduct: of the ten holes from where the lead pipes departed, four can be seen on this picture. the city was attacked by the Mongols in sur les Aqueducs Romains, Les Belles The white arrows show the direction of the flow What is an aqueduct? It can be defined as flow only, following the ground level with a masonry channels with more or less 1219. Furthermore, in the XIIth century, Lettres, 1983 (photo P.L. Viollet). “an artificial channel or conduit aimed at gentle slope. By the IIIrd century BC, the rectangular shapes allowing larger cross- delivering water to human settlements for effect of pressure in fluids became better sections and thus higher flow rates than domestic or industrial purposes”. This known, thanks to the works of a number of older aqueducts. The first aqueduct of the definition excludes irrigation and philosophers and mathematicians orbiting city of Rome, , was built by navigation canals, and implies that water is around the famous Alexandria Library and 312 BC, and was mostly a subterranean Roman aqueducts: delivered with a sufficient head for feeding Museum -the most illustrious one being simple channel. Between 312 BC and 226 a distribution network or producing energy. Archimedes (287 – 212 BC) who lived in AC, 11 aqueducts were built for the city of hydrology Vs. hydraulics !? In the birthplaces of the oldest Syracusa but had visited Alexandria and Rome, the longest ones being about 90 civilisations, ancient Egypt and the fertile used to correspond with the km long: (144 BC), the first By H. Chanson, catchment hydrology. Four aqueduct flowrate varied on average within 35% of the plain of lower Mesopotamia, the building mathematician Erathostenes of Cyrene, one to use a siphon allowing water to pass Reader, Environmental Fluid Mechanics, systems are discussed and the results mean at Gorze, but these variations were of aqueducts was not possible because no director of the Alexandria Library. By that from Caelius to Aventin hills inside Rome, Department of Civil Engineering, The demonstrate severe hydrological limitations greater during dry periods: e.g. between 40% water resources were available in that time, pressure conduits made of stone- and Aqua Anio Novus (52 AC), issuing University of Queensland, Brisbane QLD during dry periods, implying needs for some and 200% of the average daily flowrate at altitude. The oldest aqueducts are located carved elements began to be used, from a dam on the river Anio. The amount 4072, Australia form of dynamic regulation. Gorze, and the spring outflow was zero more in the eastern Mediterranean area where allowing aqueducts to descend a valley of water supplied to Roman cities reached E-mail: [email protected] - than once at Mons. While the flowrates springs or streams are found in mountains and rise a hill using the principle of the so- a peak: 1 cubic meter per day and per Website: http://www.uq.edu.au/~e2hchans/ Hydrology and operation of some Roman during Roman times are unknown, it is or hills above the cities. By 2000 BC The called “siphon”. This technique was not person for the city of Rome. aqueducts plausible that hydrological variations were palace of Cnossos in Crete had a water developed in Greece but in Hellenistic The Romans built aqueducts all around Roman aqueducts supplied water to cities for The hydrology of some catchment areas similar to present trends. This suggests that distribution system, and there were also Anatolia (modern Turkey). In Pergamon, their empire. The longest ones are the public baths and toilets in addition to public supplying Roman aqueducts were recently the aqueducts conveyed relatively low flows terracotta or wood aqueducts in the astounding Mandradag aqueduct was aqueducts of Apamea (Syria, 116 AC), 150 fountains (HODGE 1992, FABRE et al. 2000). studied (CHANSON 2002), the “source de during dry periods. In turn, the operation of Mycenaean Greece (1600 – 1200 BC), for probably built under king Eumenes II (197 km long, and of Carthage (Tunisia, 162 They were long subterranean conduits, l’Eure” at Uzés supplying the Nîmes the aqueduct and the water distribution in the instance in the palace of Pylos (see – 159 BC). It consisted of a 40 km AC), 118 km long in its longest branch, following contour lines (Fig. 1 & 2). Numerous aqueduct; the “source de Gorze” feeding the Roman city had to be adjusted, possibly with Chronicle n°2 in Newsletter 6, 2002). In upstream section with 3 parallel terracotta and with wonderful arches in the valley of aqueducts were used for centuries and some Gorze aqueduct (Metz); the “source du Thou” a dynamic regulation. classical Greece (VIth to IVth centuries pipes, followed by a siphon made of a Oued Milliane. In Europe, the longest are still in use (e.g. Carthage, Mons). Their and “ruisseau d’Arches” supplying the Mont BC), terracotta aqueducts used to follow single lead pipe. The siphon started from a aqueducts outside Rome are found in construction was a huge task, often d’Or aqueduct (Lyon); and the “sources de la Aqueduct flow regulation? the level of the ground, in which there cistern facing the Pergamon acropolis (fed Koln, 98 km long, and Lyon with the performed by the army under the guidance Siagnole” feeding the Mons aqueduct Several aqueducts were equipped with were buried, in order to protect them by the aqueduct’s upstream section), went aqueducts of Gier and Brevenne which are of military hydraulic engineers. Their cost was (Fréjus), which are all in use today (Table 1). regulation basins installed along the canal. against accidental damages and to hide down in a valley to an altitude 200 m lower more than 70 km long. The most extra-ordinary considering the real flow rate The comparison between the Gorze and For example at Ars-sur-Moselle (Metz); at the them from the eyes of enemies: they were than the reservoir, and then up 175 m to impressive roman bridges are in Nîmes (less than 400 l/s) : i.e., about 1 to 3 million Nîmes aqueducts is particularly relevant, Vallée de l’Eure, upstream of Pont-du-Gard, made with elements 60 cm to 1 m long, the Pergamon acropolis. The outside (see the photo by Chanson in the following sesterces per kilometre on average (FEVRIER considering that they were among the largest at Lafoux along the Nîmes aqueduct; at and 11 to 22 cm inside diameter. By 530 diameter of this pipe was 30 cm, (probably article), Segovia and Tarragona. The 1979, LEVEAU 1991). Today this would aqueducts in Roman Germany and Gaul Segovia upstream of the aqueduct bridge. BC on the island of Samos, a 1200 m long 20 cm inside), and it was about 3 km long. highest and most sophisticated siphons represent about 20 to 60 million US$ per km. respectively, and that they had similar Most regulation basins were equipped with a tunnel had been drilled in order to allow It was around that time that the are probably those of the 4 roman For comparison, the construction of the characteristics. Both were supplied by a series of gates and an overflow system. such a terracotta aqueduct to pass Romans came along -the most famous aqueducts of Lyon: the Gier aqueducts Tarong water pipeline (Australia, 70 km long, natural spring with a catchment area Basic hydraulic considerations imply that through a mountain and to reach the city; aqueduct builders of all times- and in the had four siphons which were between 575 Q = 0.9 m3/s) costed about 100,000 US$ per between 45 and 60 km2, and the aqueducts undershoot gates were used to regulate the this tunnel is said to have been drilled by a article on page 27, Dr. H. Chanson m and 2660 m long, and between 30 m km in 1994. were equipped with a massive aqueduct aqueduct flow while overshoot gates were man from Megara called Eupalinos, and discusses some aspects of their and 114 m high. Each siphon consisted of Despite superb ruins, little is known of bridge (Fig. 1). used for the overflow discharge (CHANSON was considered by the Greek historian hydrology. an upstream and a downstream reservoir the hydraulic engineering of the Roman Overall, recent hydrological data show 2002). Hydraulic calculations were conducted Herodotes as one of the three most The Romans used tunnels, bridges, at each side of the valley, and about 10 aqueducts. What was the flow rate? How did large variations in streamflow (Table 1). for two large regulation basins on the Gorze impressive civil engineering works of cascades, arches, and siphons -all parallel lead pipes (outside diameter they operate? How were they designed? In During dry periods, the daily flow was on and Nîmes aqueducts. The results Greece. techniques allowing to cross hills and between 20-25 cm) descending down the this note, it is argued that the hydraulics of average less than 10% of the maximum demonstrated that the undershoot gate All these old aqueducts used gravity valleys. Rather than terracotta, they used valley, crossing the river on a low bridge, the aqueducts was limited by their discharge. In a month, the daily spring openings had to be small: i.e., between 2

26 27 IAHR IAHR AIRH Newsletter AIRH Newsletter

and 10 cm at Gorze, and between 3 and 12 References HODGE, A.T. (1992). “Roman Aqueducts & Water Supply.” Duckworth, London, UK, 504 pages. BURDY, J. (2002). “Les Aqueducs Romains de Lyon.” Presses Making an IMPACT cm at Nîmes. This type of operation implied LEFEBVRE, C. (1996). “L’aqueduc Antique de Gorze à Metz.” Universitaires de Lyon, Lyon, France, 204 pages. (‘The Antic Aqueduct of Gorze at Metz.’) Itinéraire du fine gate opening adjustment systems to CHANSON, H. (1999). “The Hydraulics of Open Channel Flows : Patrimoine No. 119, Direction Régionale des Affaires enable precise flow regulation. An Introduction.” Butterworth-Heinemann, Oxford, UK, 512 If you visited UK IAHR Institution Member numerical breach models, as well as in the – sometimes by tens of metres - due to Culturelles de la Lorraine, Nancy (in French). pages. (already translated into Spanish and Chinese) What type of flow regulation was used? LEVEAU, P. (1991). “Research on Roman Aqueducts in the past HR Wallingford in September 2002 you development of new models. their effect on bed level. Within IMPACT, CHANSON, H. (2002). “Certains Aspects de la Conception Ten Years.” Future Currents in Aqueduct Studies, Leeds, UK, T. Water supply operation can be based upon hydrauliques des Aqueducs Romains.” (‘Some Aspect on the might have seen engineers building what Statkraft Grøner and HR Wallingford researchers from the Université Catholique HODGE ed., pp. 149-162. two different techniques; i.e. On/Off (100% or Hydraulic Design of Roman Aqueducts.’) Jl La Houille Blanche, looked like giant sandcastles in a test will also undertake research to look at de Louvain, Universitá de Trento, Statkraft VALENTI, V. (1995a). “Aqueduc Romain de Mons à Fréjus. 1. No. 6/7 (in French). 0%), or a dynamic flow regulation. In the Etude Descriptive et Technique. Son Tracé, son Profil, son flume. In fact these were carefully scaled factors contributing to breach location. Grøner and Instituto Superior Technico will FABRE, G., FICHES, J.L., and PAILLET, J.L. (1991). Assise, sa Source ...” Research Report, Fréjus, France, 97 former case, the gates were open constantly, “Interdisciplinary Research on the Aqueduct of Nîmes and the and instrumented model dams. The This has particular applications in flood use physical and numerical modelling to pages. and the waters flowed to the cities without .” Jl of Roman Archaeology, Vol. 4, pp. 63-88. exercise was part of IMPACT defence management. address issues of near- and far-field VALENTI, V. (1995b). “Aqueduc Romain de Mons à Fréjus. 2. FABRE, G., FICHES, J.L., LEVEAU, P., and PAILLET, J.L. (1992). further regulation than the force balance Etude Hydraulique. Son Débit ... de sa Mise en Service à son (Investigation of extreMe flood Processes sediment flow during dambreak and “The Pont du Gard. Water and the Roman Town.” Presses du Déclin.” Research Report, Fréjus, France, 123 pages. between gravity and flow resistance (e.g. CNRS, Caisse Nationale des Monuments Historiques et des And unCerTainty) a major three-year Flood Propagation extreme flood flows. HENDERSON 1966, CHANSON 1999). The Sites, Collection Patrimoine au Présent, Paris, France, 127 European research project looking at the It is extremely difficult to model complex gates and valves were used to stop the flow pages. failure of flood defence embankments and flood flows in urban areas, yet these are Risk and Uncertainty FABRE, G., FICHES, J.L., and PAILLET, J.L. (2000). “L’Aqueduc for repairs, maintenance and cleaning. de Nîmes et le Pont du Gard. Archéologie, Géosystème, embankment dams. the very areas that can be most affected The research that has already been Dynamic flow regulation is commonly used in Histoire.” CNRS Editions, CRA Monographies Hors Série, Paris, Supported by the European by flooding after embankment failure or outlined focuses on processes. One modern times and it involves a system France, 483 pages & 16 plates. Commission under its Fifth Framework dambreak. IMPACT focuses on how best important aspect of any process that FEVRIER, P.A. (1979). “L’Armée Romaine et la Construction des operation to respond constantly to the users’ Aqueducs.” Dossiers de l’Archéologie, Séries Les Aqueducs Programme, IMPACT is contributing to produce and extend reliable modelling contributes towards an overall risk demand. In Roman times, this type of Romains, Vol. 38, Oct./Nov., pp. 88-93. towards implementation of the Generic methods for the propagation of assessment (i.e. prediction of flood risk) is operation would have required an engineer in HENDERSON, F.M. (1966). “Open Channel Flow.” MacMillan Fig. 1 - Pont du Gard, Nîmes aqueduct in June Activity on ‘Natural and Technological catastrophic flood flows. Partners from an understanding of the uncertainty Company, New York, USA. 1998 - Looking from the right bank charge of the regulation, gangs of workmen Hazards’ within the Energy, Environment & the Universidad de Zaragoza, CESI, the associated with the prediction of that operating the gates and a good Table 1 - Hydrological and hydraulic characteristics of four Roman aqueducts Sustainable Development programme. It Université Catholique de Louvain and particular process. A further goal of communication system along the aqueduct Gorze (Metz, Nîmes Mons (Fréjus, Mont d'Or Remarks is also supported by the UK’s Department CEMAGREF aim to: IMPACT is to identify sources and canal. CHANSON (2002) discussed some Fra.) (France) Fra.) (Lyon, Fra.) for Environment Food and Rural investigate flow behaviour in urban areas magnitude of uncertainty associated with implications. It is plausible that several Hydrology Affairs/Environment Agency and feeds into and identify the most appropriate the prediction of each process, and to aqueduct systems (e.g. Gorze, Nîmes, Mons) Catchment area (km2) : 58 45-50 130 a current R&D project entitled ‘Reducing techniques for simulating these conditions demonstrate these through application to Spring(s) : source des Eure (Uzés) sources de la (1) source du were controlled dynamically: e.g. gates were Bouillons Siagnole (Mons) Thou the Risk of Embankment Failure under identify dambreak flow behaviour in a case study. Implications for the end possibly operated twice per day to store (Gorze) (2) ruisseau Extreme Conditions’. IMPACT involves complex natural valleys and its interaction users of such data – including asset water in the canal at night. d'Arches eleven main collaborators1 and is co- with infrastructure. managers and emergency services – will Study period of the springs : 1/1997 to 7/1967 to 1/1981 to end of 20th 12/1998 5/1968 & 12/1993 century ordinated by HR Wallingford. Work will involve a combination of then be reviewed. Website 1/1976 to Embankment and dam failure can physical modelling, numerical analysis and 12/1978 This note is complemented by the following Spring average discharge 8,050 (*) 29,600 97,200 (1) 400 Average daily data. (*) include cause catastrophic flooding and loss of life. case study simulation. website : http://www.uq.edu.au/~e2hchans (m3/day) : (2) 1,000 overtoppings. The IMPACT work programme investigates If you would like further information about Standard deviation (m3/day) : 2,950 ------Modern data. the IMPACT project, or if you wish to con- /rom_aq.ht three related ‘extreme’ flood process areas: Sediment Movement Maximum daily discharge 10,980 (*) 143,400 1,550,000 (1) 1,500 Modern data based upon daily tribute to the work in any way, please con- (m3/day) : (2) 3,000 data. (*) include overtoppings. breach formation Experience in the USA has shown that tact Mark Morris at HR Wallingford (01491 CHANSON, H. (2002). “Some Hydraulics of Minimum daily discharge 1,100 10,800 0 (1) 100 Modern data. flood propagation significant quantities of sediment and 822283, email: (m3/day) : (2) 150 Roman Aqueducts. Myths, Fables, sediment movement debris can move during extreme floods. [email protected]). Further Realities. A Hydraulician’s Perspective.” Hydraulics A fourth area comprises geophysical These can affect rates of flood wave information is also available on the project Aqueduct length (m) : 22,300 49,800 39,400 26,000 website at www.impact-project.net. Internet resource Total drop in invert elevation (m): 14.19 17 481 372 techniques and data collection whilst a propagation and influence floodwater level Channel (internal) width (m) : 1.1 1.2 0.60 0.5 Main channel. (*) aqueduct- fifth crosscutting theme addresses the 2 * 0.85 (*) bridge. Acknowledgments Estimated maximum discharge 15,000 35,000 52,500 10,000 Estimates (?). issues of uncertainty associated with The writer thanks the numerous people who capacity (m3/day): prediction of each of these processes. provided him with relevant information, in Maximum water depth (m): 0.92 1.0 possible transi- 0.65 Corresponding to the height of tion to pipe flow the waterproof mortar (enduit de particular Mr G. BERGE Jussy (Fra) and in some sections mortier de tuileau). Breach formation Société Mosellane des Eaux. Storage volume in the 21,200 58,800 -- -- Excluding the aqueduct-bridge. Breach research has involved both aqueduct (m3) : laboratory and field tests. Large scale field Aqueduct-bridge work was carried out last autumn near Fig. 2 - Brévenne River : Moselle Gardon -- -- aqueduct (Lyon, Bridge height (m) : 30 48.3 -- -- Pont sur la Moselle and Pont- Lake Røssvatnet in northern Norway, Fra.) at Biternay - Inside view of the du-Gard respectively. where specially built 6m high subterranean Bridge length (m) : 1,300 360 -- -- embankments were failed under controlled conduit, looking Channel invert slope (So=sinq): 3.9 E-3 7 E-5 -- -- upstream Internal width of channel (m) : 2 * 0.85 1.2 -- -- conditions. Three further tests are planned Upstream regulation basin - 18.0 4.0 -- -- Banks full. during 2003. Laboratory investigations Volume (m3) : Downstream dissipation structure 4.24 N/A -- -- Banks full. have also been conducted at Wallingford. - Volume (m3) : Data from both laboratory and field work is being used to test and validate existing Aqueduct usage Start : AD 100/200 AD 40/80 BC 31/AD 70 BC 20 Estimates (?). AD 450/500 AD 350/500 AD 370/470 -- Estimates (?). End : 1 The IMPACT project team comprises Universität Der Bundeswehr München (Germany), Université Catholique de Louvain (Belgium), CEMAGREF (France), Universitàdi Trento (Italy), Universidad de References : FABRE et al. (1991,1992,2000), VALENTI (1995a,b), LEFEBVRE (1996), BURDY (2002), CHAN- Zaragoza (Spain), CESI (Italy), Statkraft Grøner AS SON (2002). (Norway), Instituto Superior Technico (Portugal), The Geo Group (Czech Republic), H-EURaqua Physical model undergoing laboratory testing at Wallingford (Hungary) and HR Wallingford Ltd (UK).

28 29