THE REAL SERVICE LIFE OF ROAD BRIDGES IN – A CASE STUDY Road bridges in Sweden

G. RACUTANU Department of Structural Engineering, The Royal Institute of Technology, , Sweden

Durability of Building Materials and Components 8. (1999) Edited by M.A. Lacasse and D.J. Vanier. Institute for Research in Construction, Ottawa ON, K1A 0R6, Canada, pp. 56-70. Ó National Research Council Canada 1999

Abstract

To clarify the real condition of a country’s bridge stock and their structural members at a given time, reliable regular and systematic inspections are needed. Accurately documented, these inspections are essential for the road administration authority in the planning of the necessary maintenance and repair works on the existing bridge stock, thus contributing to an overall cost-effective bridge management. The Swedish Road Administration has, at the present, a relatively intact and well- organised bridge archive system. Most of the former bridge inspections performed on the public road network after the nationalisation of the Swedish roads in 1944, are documented and stored in different archives throughout the country. The information from these inspections has been fed into a Microsoft Access database called BEA (Bridge Element Analysis), with the purpose to explore the possibility to determine the damage occurrence, damage growth and the real service life for bridge structural members. In this case study, the database is used to examine, if certain information and factors like geographical location, traffic, and weather data at the time of casting, affect the service life of bridge structural members thus the service life of bridges. Based on the information stored in the database, the service life of bridge structural members in certain service life conditions may be estimated. For example, the real service life of bridge members as a function of the outside airs temperature at the time of the con- crete casting. The information stored in the database may also be used in the devel- opment of a damage growth model based on the assessed condition at the time of in- spection.

Keywords: Bridges, service life, bridge structural member, bridge management, data- base, estimated service life, damage type, damage cause. 1 Introduction

1.1 General This paper is the result of a case study on the real service life of some 270 bridges located on the -network. However, the results from the extended study during 1998, comprising a total of 370 bridges are not presented in this paper. Bridges are an important part of a nations road system. Maintaining and repair- ing existing bridges is a major economic concern for many governments and local authorities around the world. Many bridges are built with outdated technical codes and demands. The service life of a bridge is often defined as the time from construc- tion to the time a critical limit state or property has been reached. It is generally ex- pected that during their service life, bridges can fulfil certain demands such as traffic safety, continuous traffic flow and a designed load bearing capacity. Regular and systematic inspection of the existing bridge stock should be performed in order to verify that such demands are met at all times. With the 1944 nationalisation of the public roads in Sweden, information about most of the regular bridge inspections have been carefully documented and filed by the Swedish National Road Administration in different archives. Damages are rated in condition grades on a scale between zero and three.

1.2 Background The research program ”The durability, management, repair and life cycle cost of concrete structures” conducted at the Royal Institute of Technology was initiated by the Swedish National Road Administration in 1991. As the program title suggests the program consist of a number of research projects. The project ”Actual durability and service life of Swedish concrete bridges” was started in late 1992. The aim of the project was, initially, to establish the research possibilities of the vast amount of in- formation in the archives and data in the bridge management system of the Swedish National Road Administration. The project resulted in the later part of 1997 in a Li- centiate Thesis, “The Real Service Life of Bridges”.

1.3 Aims and scope The second stage of the project was a further investigation in the use of the bridge information available with the aim of developing deterioration models. This has been achieved for bridge structural members in certain service conditions. In the future, bridge designers and inspectors would predict the performance of certain bridge structural members in certain service environments according to type and cause of the damage. Service life models in combination with a relational database containing actual durability information would result into a tool for assessing and predicting the actual service life. If such a model could be achieved, it would give researchers and owners a precious tool in the planning of maintenance works, design of the different struc- tural parts and component and feedback for future codes and regulations. The development in time of the condition grade can be a way of predicting the service life of bridge components, contributing to cost effective bridge management and improved bridge component design in the future.

2 Method

Sweden has many different local climate conditions. In the north of Sweden there are few yearly frost-cycles, mainly because the temperature is always below zero degrees Celsius throughout the winter season while it can oscillate around zero degrees in central and southern Sweden leading to several frost-cycles. This implies better frost-resistance performance of the concrete in the central and southern parts of the country. De-icing salts are used on many roads during the cold season of the year. Be- cause of the various climate conditions in Sweden, bridges in the central and southern part of Sweden are especially exposed to de-icing salts. It became evident in the early part of the study that to minimise the service life variables, bridges should be studied on geographically close road sections. 15 different road and highways sections with 188 bridges were selected in the central part of Sweden. In order to be able to study how different climates affect the bridge structures, the geographical extent should be increased with road sections in different parts of the country. This resulted in the gathering of information on 82 bridges in the northern part of Sweden. The roads in that particular county are de-iced using very little salt or no salt at all which is an in- teresting factor to study.

2.1 The information gathering process

2.1.1 Extent In order to gather the information in an efficient way a strategy to find relevant information was prepared. During the first part of 1993, detailed checklists and rou- tines for the gathering process were produced. This was essential in order to minimise the risk for errors and loss of information. The gathered information was placed in project files for each road section in the investigation. This material is unique, and should be an important asset for many research projects. The limitation of the road sections, in both length and the amount of bridges were adjusted so that traffic flow and climate have similar conditions. This was made to make road sections comparable and thus enabling the study of certain factors im- posed by the surrounding environment that might affect the service life of structural members. In 1997, the investigation included 19 different road sections with a total of 270 bridges. The results and conclusions were presented in the Licentiate thesis “Konstbyggnaders reella livslängd”, the Royal Institute of Technology, Stockholm. The R&D project was continued and during 1998, 100 bridges on four road sections in the western and eastern parts of Sweden were added to the investigation, thus covering two new different parts of Sweden. 2.1.2 Selection of information The gathered information for each bridge can be divided into three different groups:

· General information on the bridge · Inspections and repair works performed on the bridge · Information from the construction stage and the final inspection

All performed inspections and repairs by the Swedish National Road Admini- stration were summarised in the inspection- and maintenance sheet of the bridge and archived. Even these have been copied, when available, for the bridges included in the investigation. Information given is notes the inspectors have taken during inspec- tion or the maintenance work and the date they were performed. The basis of the inspection- and maintenance sheet is ”the full inspection re- port”. That report gives information, in addition to that in the summarised inspection- and maintenance report, on for example suggestions of bridge maintenance measures or activities and who should carry them out. The full inspection reports were not always easy to find. Unfortunately, in many cases only the inspection- and maintenance report had been saved. The documentation from the performed special inspections is often in the form of tables and diagrams and have been found both at the central and regional archives of the Swedish National Road Administration. Values from the chloride profiles have been inputted the database BEA for analysis. Other gathered information, on bridges in the investigation found in the different archives, was copied and placed in the proj- ect files. Concrete castings are documented in concrete casting reports. The report includes facts about mixing conditions, quantity, concrete quality and meteorological conditions at the time of concreting. At many concrete castings, concrete tests are taken for analysis. The concrete samples are related to a testing record in the concrete casting reports. Some tested qualities are compressive strength, water tightness and frost resistance of the con- crete.

Examples A lot of the gathered information is of historical value. One can, for instance, look back in time and see how certain parameters have developed in reality. An ex- ample of this is the development of the real (test results) compressive strength of the concrete in certain structural members. 70

60

50

40

30

20

10

Compressive strength, MPa 0 1930 1940 1950 1960 1970 1980 1990 2000

Fig 2a: The real development of compressive strength (test values) according to documented concrete castings for deck slabs in the investigation

0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995

Fig. 2b: The average development of water cement ratio, wcr, throughout the years according to all documented concrete castings in the investigation

Another example is how the water-cement ratio, (WCR), has decreased throughout the years. All gathered information on the WCR are originate from the concrete casting journals from the investigated bridges that were found in the differ- ent bridge archives of the Swedish National Road Administration

2.1.3 Other sources of information One of the aims of the R&D project was to identify possible sources of infor- mation within the Swedish National Road Administration. A number of available da- tabases were used extensively and successfully in the information gathering process. These are:

· Road Data Bank (Vägdatabanken VDB) · Bridge Data (Brodata) · Central and Regional Bridge Archives · Road Museum in Borlänge Another important source of information was gathered during interviews with experienced personnel working with the Swedish National Road Administration. Bridge inspectors, bridge engineers, archivists and historians helped to complete the missing ends, which occurred during the entire time.

3 Bridge element analysis, BEA –– Relational database

It became clear at an early stage that Brodata, which is the bridge database of the Swedish National Road Administration, could not be used with the aims and goals of this R&D project. Brodata does not contain any information on performed bridge inspections prior to 1987. A new database should be created. As all available and relevant concrete information such as wcr, designed and real compressive strength, entrained air, weather data from the time of casting, maintenance and repair work performed on the structures and more was gathered, the choice fell on using Micro- soft ACCESS relational database. The Windows - ACCESS database, BEA (Bridge Element Analysis), specially developed for the needs and requirements imposed by this research project, deliber- ately uses the same codes, definitions and terms as used by the Swedish National Road Administration. The database is described in detail under chapter four in the Licentiate Thesis “Konstbyggnaders reella livslängd”. The process of creation and development of the Bridge Element Analysis, BEA, database was achieved in five major steps:

1. Step one: Planning; What will the results be used for? How to design the tables and relations for an optimal database structure? Why certain data will be se- lected for the analysis? 2. Step two: Creating tables and relationships; All tables and relations were cre- ated in such a way that data could be combined after needs. 3. Step three: Creating quarries; Quarries can cross-examine data in one or more tables, after the needs of the project. 4. Step four: Creating forms; Data input needs a number of forms, which make the database user friendly. 5. Step five: Creating reports; When all data has been put into tables, quarries are completed with reports. New reports have been created continuously, according to increased search parameters.

The database BEA has great development possibilities. New objects can be cre- ated without difficulties. The performance can be improved if that becomes a priority, for example, through division of the database or better query design. Many queries have developed during the investigation. This is mainly due to the many questions of issues and reports that have occurred throughout the project. Fig. 3: Tables and relations in the relational database BEA. (For principal contents of the database see section 6.1)

4 Damage picture in the investigation

2898 performed bridge inspections on the 270 investigated bridges have brought forward 2165 damage remarks where type- and cause of damage is stated. Even if the result is not significant for the entire country, they are giving a clear indi- cation of the general condition of bridges in the central and northern parts of Sweden. The first step is to present how the 2165 damage remarks are divided on the dif- ferent structural members of the bridges. Later the some types of damage over- represented on the structural member level for all condition classes will be presented. The last step is to present, which causes of damage, for all condition classes, that lay behind the most common damage types.

4.1 The damage picture on structural member level As previously stated, the bridge inspector must record certain damage informa- tion during the inspection. The extent of the data depends on the type of performed inspection. The requirements are established in the bridge inspection manual of the Swedish National Road Administration.

4.1.1 Assessment of condition classes (CC) Two important requirements in the damage documentation process are the measurement and condition assessment of damages. This is done for damaged struc- tural elements in the following two stages: · Stating the physical condition in terms of measurement and measured value · Assessment of the functional condition in terms of condition classes

The physical condition is determined with reference to the development of pre- vious or new damages and certain known deteriorating processes. The different methods of measurement that are to be used for a particular type of damage are de- scribed in publication 1993:35 (in Swedish) or publication 1996:038(E) (in English). The physical condition of a damaged structural element can then be described using the variable defined for each method of measurement. The functional condition of a structural member is described by the bridge in- spector in terms of condition classes. The condition class describes to what extent a certain structural member satisfies the designed functional properties and require- ments at the time of inspection. It can be said that the assessment of condition classes is composed on previous and current measured values (the physical condition) and the inspector’s competence in the propagation of different deterioration processes. The condition class (CC) for a structural member can be registered on a scale of four. The scale implies that, at the time of inspection, the structural member was consid- ered to have:

Table 4: Assessment of condition classes for bridge structural members

Condition class (CC) Assessment

3 Defective function 2 Defective function within 3 years 1 Defective function within 10 years 0 Defective function beyond 10 years (No damage at time of inspection)

4.2 The damage picture in the investigation on structural member level, all condition classes With this search criteria, all damage remarks will be expressed as a percentage, even those damages that have been given the condition class ”0”. Interesting in the context is that the structural members ”edge beam”, ”support” and ”deck slab” together answer for 47% of all remarks, regardless of year of con- struction, type- and cause of damage. The result confirms the hypotheses that these three structural members can be- come the largest maintenance problem for the Swedish National Road Administration in the future. The service life problem on Swedish bridges is directly dependent upon the physical condition development for these three structural members. The structural member’s ”slope and embankment end” and ”parapet” together answer for 26% of all remarks. The restoration of embankment ends is a frequent maintenance problem for the Swedish bridge management. Steep slopes and embankment ends, above all on the older bridge stock, have such inclination that it seems like the former bridge design- ers wished to defy the law of gravity. On most Swedish bridges the parapet is fixed to the edge beam. This is, in many cases, an unfortunate solution. The structural member ”parapet” is more frequently involved in collisions, and this can lead to a domino effect where the edge beam is damaged. Condition class 3 means that, at the time of inspection, the structural member has entirely lost its function. The structural member ”parapet” is over-represented in the group of condition class 3. It answers for 22% of all reported damages; most of them caused by collisions. The structural member ”edge beam”, ”support” and ”deck slab” together answers for 27% of all damages in this particular group. Worth noting is that the structural member ”primary load-bearing element” has managed fairly well in the investigation with only 1% of all reported damages with assessed condition class 3.

WING- AND RETAINING WALL EXPANSION JOINT SUPPORT BEARINGS 6% 1% 12% 2% SLOPE AND OTHER EMBANKMENT END <1% 11% SURFACING 8%

DECK SLAB 10% PARAPET 15% DRAINAGE SYSTEM 4%

FOUNDATION 1% PRIMARY LOADBEARING EDGE BEAM ELEMENT 24% WATERPROOFING 4% 2%

Fig. 4: 2165 damage remarks, all condition classes, and percentage division by the bridges’ structural members

5 Damage types and causes

Bridges consist of a number of structural members, which, in their turn can con- sist of a number of structural elements. Even structural elements can consist of a number of element parts. When a damage is detected, the damage is documented with information about the type of damage and the cause of the damage. 5.1 Types of damage The damage types used in the investigation follows that used by the Swedish Bridge Management System, SAFEBRO. Initially, the investigation of 270 bridges, 2165 reported damages were registered. “Corrosion” as a damage type can even be related to other steel components such as railings.

SCOUR

5%

DEFORMATION

5%

SPALLING CORROSION

6% 20%

CRACKING

7%

MOVEMENT WEATHERING 9% TENSION CRACKS 15%

11%

Fig. 5a: Distribution of the 2165 (>5%) damage reports in the investigation after type of damage

5.1.1 Causes of damage If an inspection report contains damage remarks, the bridge inspector must es- tablish the probable cause of the reported damage. This will be accomplished in three steps. First, the primary cause has to be determined. These can be one of the follow- ing:

· Defective construction · Service condition · Environmental action · Accident · Defective design · Defective maintenance

Each of the primary damage cause is subdivided in their turn into secondary and tertiary causes. Over 40 percent of all the damages in the investigation are related to “Environ- mental action” and remarkably only one percent to “Defective design”. Defective Defective Defective construction maintenance design 5% 3% 1%

Accident 9%

Other 9%

Environmental Service action condition 43% 30%

Fig. 5b: Distribution of the primary causes of damage in the investigation

Table 5: Environmental action as primary cause of damage. The distribution of the secondary and tertiary causes of damage in the investigation

Environmental action (937) Environmental Physical attack Chemical attack Biological attack Action (163) (30) (19) (725) Environmental Physical attack (24) Chloride attack (10) Biological attack (19) Action (183) Frost action (139) Initiated chloride attack (8) Carbonation (6) Water (5) Alkali-silica reaction (ASR) (1) Table 5-2: Service condition as primary damage cause. The distribution of the secondary and tertiary causes of damage in the investigation

Service condition (650) Loading Abrasion Erosion (483) (34) (133) Loading (183) Loading, shrinkage (178) Abrasion, traffic (31) Erosion (133) Loading, traffic (60) Abrasion, movement (2) Loading, temperature (25) Other (1) Loading, overload (22) Loading, creep (8) Loading, movement of support (4) Other (2) Loading, earth pressure (1)

5.2 Service life model A simple method in predicting the development in time of the condition class (CC) for a given type of damage on a certain bridge element is introducing a linear trendline on the development of the average condition class. The condition put is that at the time of final inspection (age = 0) the average condition class (CC) was “0”. This has been integrated in the BEA database and trendlines can be drawn for a num- ber of search criteria.

3

All highways E18 2 E20 Linear (All highways) Linear (E20) 1 Linear (E18)

Estimated condition class (CC) 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Age of the bridge at time of inspection (years)

Fig 5c: Variation in time of the condition class development for the element ”parapet fixing”, all causes of damage for road section E18 Köping- Västerås (38 remarks), E20 Eskilstuna-Arboga (17 remarks) compared to all road sections (143 remarks). 6 Conclusions and further research

6.1 Relational database – A must in a modern bridge management system The results presented so far with the help of the relational database BEA apply to the limited geographical areas. Data for bridges in big parts of the country are missing. At the present, the database is being completed with a further 100 bridges located on four new road sections, bringing the total number of investigated bridges to 370 on 23 different road sections. The relational database BEA ought to be updated with new inspections. Among other things, BEA today (1998-09-10) contains: · 2898 Bridge inspections made between 1944 - 1995 · 2269 Major inspections · 446 General inspections · 105 Investigations · 1044 Records from concrete castings · 2165 Damages with stated type and cause · 387 Chloride profiles A relational database is today a must for an effective bridge management sys- tem. This relational database should be user-friendly and flexible. One of the obvious advantages using a relational database is the possibility to use a number of search criteria. In the Licentiate Thesis, “Konstbyggnaders reella livslängd”, a number of different search criteria were used. It was discovered that five structural members are, together, responsible for most of the inspection remarks.

Table 6a: Damage distribution on certain structural members situated on 19 different road sections in the investigation Road section Edge beam Support Deck slab Slope and Embank- Sum (%) (%) (%) ment end (%) (%) Arboga-Eskilstuna 19 11 10 12 52 Arboga-Örebro 31 8 9 7 55 Dorotea-Vännäs 16 2 0 18 36 Enköping-Bålsta 7 51 0 0 58 Eskilstuna-Arboga 19 17 12 13 61 Eskilstuna-Mariefred 19 18 18 9 64 Köping- 27 0 9 7 43 Köping-Västerås 31 3 15 7 56 -Piteå 41 7 6 13 67 Nyköping-Stockholm 16 20 13 7 56 Ramnäs-Kopparberg 19 12 2 23 56 -gränsen 14 3 6 18 41 Strömsholm-Ramnäs 18 18 2 11 49 Strömsund-Arvidsjaur 24 7 5 20 56 Uppsala-Gävle 32 9 13 3 57 Västerås-Enköping 20 20 7 19 66 Örebro-Grängesberg 39 6 8 13 66 Örebro-Laxå 19 18 1 15 53 Östhammar-Karlholm 28 4 20 10 62 By changing the search criteria in BEA, new facts can be revealed. The percent- age of damage remarks reported on certain structural members for certain construc- tion types of bridges are presented in table 6-2. The following four types of construc- tion were investigated:

· Beam and slab bridges (32 bridges), · Beam and slab frame bridges (24 bridges), · Frame bridges (162 bridges) · Slab bridges (44 bridges)

The structural member ”edge beam” are responsible for 24% of all inspection remarks in the investigation. The structural members ”edge beam”, ”support”, ”deck slab” and ”slope and embankment end” together answer for almost 50% of all in- spection remarks, regardless of search criterion. Changes in the design, above all in details such as parapet fixing and support placing. More careful inspection of slopes, embankment ends and fillings at the final inspection should be taken into account.

Table 6b: The percentage contribution, from four structural members, to the inspection remarks on four different types of construction.

Type of Edge beam Support Deck slab Slope and Em- Sum construction (%) (%) (%) bankment end (%) (%) Beam and slab 25 13 4 6 48 bridge Beam and slab 24 6 5 19 54 frame bridge Frame bridge 25 12 11 11 48 Slab bridge 24 12 10 11 57

6.2 Bridge inspections The cause of damage “environmental action” is highly overrated as a primary cause. The cause of damage “defective construction” has not been used, as a primary cause in the correct extent. The cause of damage “defective design” has not been used, as a primary cause, in the correct extent. The extent of the final inspection as well as the physical measuring points should be integrated in valid codes and regulations. This should be even considered for the guarantee inspections.

6.3 Service life models When assessing the service life for a structural member in a certain service life condition, a lot of search criteria is needed. It is also important to note that the service life of bridges often depends on the condition of its different structural members. The deterioration of one can inevitably start a deterioration process on one or more other structural members – The domino effect. 6.4 Future research The R&D project continued through 1998 by increasing the number of bridges in the investigation to 370. Even the relational database BEA has been further devel- oped for new search criteria. Several service life models have been taken into consideration. The author will present the results in “The real service life of Swedish Road Bridges” during 1999.

7 References

Racutanu G. (1997) Royal Institute of Technology, Stockholm, Sweden. Konstbygg- naders reella livslängd. Licentiate Thesis. Racutanu, G, and Troive, S. (1998) Use of Empirical Database for Evaluation of Service Life of Bridge Structural Members. Beständighet och livslängd för ma- terial och byggnader, Royal Institute of Technology, Gävle.