2010 Bridge Management A System Approach for Decision Making 67

Bridge Management a System Approach for Decision

Making

Reginald W. Stratt (PhD)

Postdoctoral student at the Department of Engineering and Technology in the
School of Doctoral Studies of the EU
Square de Meeus 37-4th Floor 1000 Brussels, Belgium
email: strattwsphd@yahoo.com

Abstract
Bridges in the United Kingdom (UK) are subject to various physical processes that result in their deterioration over time.
They can be constructed using different materials such as concrete or steel, or in combination with other materials, with
varying number of spans and are used to carry different levels of loads. Inevitably, bridges deteriorate over time at different
rates and they are vulnerable to varying forms of deterioration (e.g. rusting, corrosion). Timely maintenance activities that
are well-planned and carried out with minimal disruption to road users can present substantial savings in terms of both time
and money for both bridge owners and road users. As a result, the likelihood of disruptive emergency maintenance will
be reduced and subsequent maintenance costs over the service life of bridges will not build up significantly due to neglect.
To tackle the complicated issues regarding bridge management, research activities in the UK as well as other countries
in continental Europe concentrate largely on the bridge management process, with attention given to improving the use
of limited finances so as to maximize the returns from the maintenance and repair of the bridge stock as well as reduce
additional costs due to traffic delays and lane closures for these activities.
This project aims to first appreciate current bridge management systems (BMS), understand the practices used for
determining bridge conditions and aid the decision-making process by using a systems approach. The study includes a critical
review of other BMS’s used worldwide, development of models to predict bridge condition over time, analysis of the various
road user costs and using different optimizing techniques to best allocate finances and optimize bridge performance.
Key words: Bridge Management, Decision Making.

1. INTRODUCTION in turn accelerates the deterioration process. Insufficient

waterproofing, or failure of these systems can also allow
Bridges are subject to various forms of deterioration salt concentrations to exceed the threshold required for
in the environment, with different rates of deterioration corrosion. Spray from road traffic can cause deterioration
depending on the degree of exposure, location, type of of adjacent bridge elements that are both above and
bridge and extent of use of the bridge. The main causes below the road surface.
of deterioration are: corrosion of steel; alkali-silica In addition, bridge movement joints are highly
reaction; concrete carbonation; frost damage; sulphate susceptible to corrosion due to flowing of saline water
attack; and structural damage from vehicle impact. over the sub-structure and subsequent pooling. In
Corrosion of reinforcing steel in concrete bridges also addition to the direct cost of engineering work carried
causes cracking, staining and spalling of concrete, that out to maintain the bridge stock, indirect costs due to

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

68 School of Doctoral Studies (European Union) Journal 2010

traffic congestion can be significant. Therefore, these delays during maintenance, impacts on the environment
factors need to be considered in order to achieve good and changes in bridge condition over time are often
bridge management (Znidaric A and Moses F, 1997). not taken into account. Professional judgment is often
Formed in 1994, the Highways Agency (HA) was relied upon to decide which alternatives to select for
formed in the United Kingdom (UK) with the purpose implementation (Znidaric A and Moses F, 1997).
of managing, maintaining and improving the network After gaining an understanding of current practices
(Baker, D, 1999). It acts as an executive agency to the and approaches used in bridge management for the UK
Department of Transport (DoT). The HA is responsible and other countries, this project will consider some of
for the maintenance of approximately 9500 kilometers the latest research techniques undertaken by the HA
of motorways and trunk roads. Maintenance agents and their maintenance agents to model the conditions of
are contracted by the HA to inspect, review, design and bridges at the individual level as well as the bridge stock
carry out repair work in order for bridges to attain an level, to assist in local decisions as well as nationwide
acceptable level of performance and safety. In the year strategic decision-making. This project will have models
2001-2002 alone, for example, about £500 million was developed to estimate costs borne by road users due to
spent, averaging about £56 000 per kilometer. Over traffic congestion and traffic accident costs, subject to
the last eight to ten years, approximately £750 million different scenarios. The impacts on the environment
has been spent annually. In 1994, a study by Transport due to increased emissions will be also addressed and
Research Laboratory for the DoT found that personal estimated accordingly. Finally, Markov chain approaches
injury accident rate was a hundred and thirty per cent will be considered to provide a means to achieve optimal
higher on sites with roadworks (Wallbank E J, 1986). budget allocation to better manage the bridge stock and
Because of this study, reducing the risk of accidents optimize the performance levels for individual bridges.
and their consequential costs to other road users has
become one of the HA’s top priorities with regards to 2. OBJECTIVE
carrying out maintenance or repair work. The HA’s
Structures Management Information System (SMIS) is a The main objective of this project is to understand
bridge management system that is used to help engineers the current bridge management practices and use them
manage structures and it adheres to established codes of to estimate the current state of bridges as well as the
practice (i.e. British Standards, Design Manual for Roads bridge stock and model their condition over time. It
and Bridges (DMRB)). Its processes include planning will relate bridge scoring results with different types of
and inspection procedures, identification of needs and costs and develop a framework that can incorporate this
forming projects after prioritizing and evaluating various information with optimizing strategies to aid engineers
maintenance options. The SMIS can store new input and bridge managers in the decision-making processes
data over the duration of maintenance activities, hence concerning bridge management.
SMIS users are able to review their progress. The following tasks need to be carried out:
In the UK, several bridge condition rating and bridge
scoring systems are currently practiced. Much research 1. Appreciate the current bridge conditions of the
is currently underway to create a version that can be bridge stock in the UK.
standardized and implemented nationwide. The rating or
scoring systems are generally based on individual bridge 2. Understand the typical allocations of budget for
element condition, appearance, importance of the bridge, various bridge maintenance activities.
severity and extent of deterioration. Recommendations
work for repair, replacement or maintenance can then 3. Learn about the different BMS types, including their
be translated into costs, which the engineer can use to structure and information used.
form and manage a list of tasks required. However, other
consequences of their decisions such as additional costs 4. Understand bridge inspection techniques, how key
imposed on road users due to lane closures, effects of elements are assigned rating scores and assessing
road accidents as a result of inadequate maintenance, deterioration.

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 69

5. Understand the scoring system for bridges and Figure 3 2 Distribution of bridge age in the UK in 1997
bridge stock. (Woodward R J et al, 1999)

6. Develop models to estimate road user costs due to Bridge age distribution (1997)
maintenance and accidents .
5%
7. Develop an emission model to investigate 30%
environmental impacts due to congestion. <20 years
20 to 40 years

8. Develop simple programmes to optimize budge >40 years

allocation. 65%

9. Develop models to optimize bridge performance.

10. Use sensitivity analysis to deal with uncertainties. Figure 3 3 Distribution of bridge length in the UK in 1998
(Woodward R J et al, 1999)

3. BRIDGE CONDITIONS IN THE UK Bridge length distribution (1998)

3.1 Introduction 7%

The Bridge Management In Europe programme 25%

<50m
(BRIME) was a project started by the Forum of European 50 to 100m

National Highway Research Laboratories (FENHRL) in >100m

68%
January 1998 that seeks to improve bridge management
in Europe by improving the structure of managing
bridges in the European road network. In a recent survey
reported in the BRIME report, the composition of UK’s
bridges is as follows: Figure 3 4 Distribution of bridge types in the UK over 50
years (Woodward R J et al, 1999)
Figure 3 1 Distribution of bridge types in the UK in 1997
(Woodward R J et al, 1999)

Bridge type distribution (1997)

Steel & Iron

9% 4%
13% Composite Steel

3% Stone
28%
Concrete & Reinforced
Concrete
Prestressed Concrete
43%
Other

As seen from above, a majority of the bridge stock in

the UK comprises of concrete and reinforced concrete
bridges (43%), with 65% of the bridge stock having
the age of 20 to 40 years and of length less than 50m
(Woodward R J et al, 1999). This has implications of
more attention being needed to manage the problems

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

70 School of Doctoral Studies (European Union) Journal 2010

that concrete and reinforced concrete bridges face. Age- is on the increase with approximately 7% in 1994 (almost
related maintenance problems need to be addressed as doubling since 1988’s 4%).
well, such as bridge strengthening due to increased vehicle
weight and increased usage. A significant proportion of Figure 3 6 Average severity ratings for key bridge elements
bridges are between 50 to 100m in length, which could from 1998 to 1994
mean more severe traffic congestion consequences
should bridges of these lengths need emergency repair
due to unexpected failure, or need to be closed after a
serious accident. Modelling of traffic delays and costs
can help the bridge manager deal with these issues and
these are addressed later on.

Figure 3 5 Average extent ratings for key bridge elements

from 1998 to 1994

Similarly, the average severity ratings from 1988 to

1994 has gradually decreased over time in general, with
about 2% of the bridge stock described as having severe
defects needing urgent attention.

3.2 Maintenance & Upgrading

Routine maintenance is carried out with the aim of

prolonging the asset life making up the road network.
Most of these works are planned (about 90%) (Bourn J,
2003). Appropriate action at the right time can avoid
From 1998 to 1994, it can be observed that the future damage and in turn save more money, such as
proportion of bridges in the bridge stock having no repairing cracks on the road surfaces to reduce frost
significant deterioration has been declining gradually, penetration over the winter months, or resurfacing
with just over 50% of them in 1994. This is also the same of badly rutted roads to increase the safety levels by
for those having less than 5% of the element affected. providing a good road surface. Other maintenance
The proportion of bridges suffering from extensive activities include routine work such as inspections,
deterioration (i.e. more than 20% of the element affected) cleaning, painting, drain clearing and lubrication of
bridge bearings.

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 71

3.3 Expenditure In 1987, the DoT started a fifteen-year programme

aiming to improve the condition of the motorway, trunk
Figure 3 7 Distribution of the HA’s expenditure from 1988 to road bridges and other structures to a ‘good’ condition.
1995 (Bourn J, 1995) The total cost of the programme was estimated to be £2.2
billion. In 1995, a review of the expenditure was made
HA expediture (April 1988-March 1995) and it can be seen that a majority of the expenditure by
the HA from 1998 to 1995 was spent on steady state
maintenance. Approximately £500 million had already
19%
been spent within the first seven years (Bourn J, 1995).
Steady state maintenance
Bridge upgrading deals with specific bridge elements
10% Upgrading
that need improvement in terms of safety and durability,
Assessment & such as waterproofing unprotected and vulnerable areas,
strengthening
71%
replacing unsafe parapets and reinforcing collision
areas to absorb the impact of vehicles. A more detailed
breakdown in expenditure can be seen as follows:

Figure 3 8 Distribution of bridge expenditure in 1989 (Bourn J, 1989)

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

72 School of Doctoral Studies (European Union) Journal 2010

Figure 3 9 Comparison of expenditure as a percentage of cost problems for roads and bridges are almost the same,
as these usually involve experts (e.g. highway and
Comparision as % of cost (1989) road maintenance agencies) that are knowledgeable in
both fields. This element of expenditure is generally
60
consistent (approximately 20% of expenditure) and
50 therefore money spent for roads or bridges do not differ
largely compared to the proportional total expenditure.
40
Cost,millions

Roads The breakdown of expenditure by the HA from 1994

30 Bridges to 2002 is as follows:
Total
20
Figure 3 10 Distribution of the HA’s expenditure for various
10
works from 1994 to 2002
0
Design Specifications Supervision Workmanship
Element

From Figure 3-8, Figure 3-9 has been created. This

is a summary of comparisons of design, specification,
supervision and workmanship costs made as percentages
of the total expenditure on roads, bridges and as a total
respectively. For example, the amount of money spent
on design as a percentage of the money spent on roads is
approximately 45%. For the amount of money spent on
bridges, design costs take up approximately 55% and for
the total expenditure; design costs make up about 53%
of the total.
It should be noted from Figure 3-8 that should problems
arise due to more than one factor, repair costs are divided
equally among the various elements of expenditure,
hence the total will exceed the actual amount spent. For
roads, a large proportion of expenditure is spent on design Although the total amount of money spend by the
and specifications, whereas for bridges the expenditure is HA has decreased since 1994-1995 to 2001-2002, this
largely attributed to design followed by workmanship. is attributed largely to the decrease in number of new
The proportional expenditure to meet specifications for road construction. This can be explained by the fact
roads is significantly higher compared to bridges and that the construction of the required new roads have
the total expenditure for both roads and bridges. This been completed and has subsequently reduced to simply
could be explained by the many different specifications replacing those that are decommissioned annually. It
set out for different types of roads that need to be met, should be noted that the amount of money spent on
depending on location, ground conditions, environment, road and bridges maintenance has generally been steady
community and so on. and more money is spent to pay Design Build Finance
Compared with bridges, specifications are more Operate (DBFO) operators (Bourn J, 2003). These two
clearly set out using British Standards and other codes observations are attributed to an increasing number of
of practice and would include standard requirements ageing bridges needing more work done on them with
of meeting load bearing capacity, sufficient moment less work required for the new brides built several years
resistance and so on. Supervision regarding remediation earlier.

Stratt R.School
S. - Bridge ManagementA
of Doctoral Studies System Approach
(European Union)for Decision
Journal Making
- 2010
2010 Bridge Management A System Approach for Decision Making 73

Figure 3 11 Breakdown of the HA’s expenditure components The cost of capital maintenance for 2001-2002 is
for 2001 to 2002 about £37.50 per square meter of road, adjusted for
inflation. This is translates to about a 10% increase from
previous years, which are about £32.50 per square meter
(Bourn J, 2001).

3.4 Department of Transport (DoT)

Programme

The DoT spearheaded a fifteen-year programme

in 1987 to improve the road network that included
motorways, bridges and other structures. With the HA
In the period 2001-2002, £502 million was spent acting as their executive agency, their responsibilities
by the HA, largely on capital maintenance and routine include improving and implementing standards regarding
maintenance. As the number of new road construction network improvement.
was low, a portion of the £32 million was for this cause. Three types of key sub-programmes are carried out:
Since DBFO type roads make up 7% of the network, a
reasonable proportion of the HA’s expenditure has been 1. Assessing and strengthening of structures to carry
channelled towards paying managing agents and other vehicles up to thirty-eight tonnes and forty tonnes
maintenance related fees (Bourn J, 2003). from 1999 onwards

Figure 3 12 Cost of capital maintenance per square meter 2. Carrying out maintenance work, with a developed
(1997 to 2002) programme for such activities

3. Upgrading structures such that they meet the required

design standards set out for them

With standards being improved and made more

rigorous over time, a larger number of elements inspected
received a lower condition rating. It was noted that new
structures built twenty to twenty-five years ago, required
their full maintenance, hence resulting in a large amount
of spending later on 1994 (about £1 billion) (Bourn J,
1989).
In May 1983, the weight limit for lorries in the UK
was increased from thirty-two and a half tonnes to thirty-
eight tonnes (Bourn J, 1988), but just over a year later
in December 1984, the European Council of Minister
approved a Directive to increase the gross vehicle weight
limit to forty tonnes, significantly exceeding the UK’s
thirty-eight tonnes limit established a year before. As
a result of negotiations, the UK delayed increasing the
thirty-eight tonne limit to forty tonnes till 1999. This
was a 23% increase from thirty-two and a half tonnes
by comparison, which pose serious implications for
old structures designed to carry lighter vehicle loads.
Between 1984 and 1986, a review of short span bridges
spanning less than 50m in length by the DoT suggested

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

74 School of Doctoral Studies (European Union) Journal 2010

that up to 22% this bridge type would not pass new 4. WHAT IS A BRIDGE MANAGEMENT
assessment codes used then. About a thousand old short SYSTEM (BMS)?
span motorways and trunk road bridges were estimated
to require strengthening work or total replacement. The 4.1 Introduction
initial load bearing assessment of structures was done by
mathematical modelling, failing which would subject the A bridge management system (BMS) is an organized
bridge to more tests and investigations. approach to handle the various needs and requirements
By comparing the HA’s Structures Database with that regarding good stewardship of bridges. This usually
of the Organization for Economic Development (OECD), involves a software or other database to deal with the
it can be seen that there is scope for improving integration large amount of data involved and display them on
of information using modern computational tools. There both element and strategic level for the user to aid the
is a current general lack of data and access, with little decision-making process (Clausen P, 1992). The main
information concerning bridge management work. functions of a typical BMS are:
Obtaining and managing information can improve the
planning and supervision of inspection or maintenance 1. To provide a full inventory of bridges, with record
activities, with better usage of time to minimize delays and predictions of both the past and future condition
and consequential costs incurred (OECD Report, of bridge elements and components. This also
1992). Integrating traffic and accident information to applies to the bridge’s load carrying capacity.
the database can develop better management of traffic
congestion and allow actions to be taken early so as 2. To assess the deterioration rate, select the optimal
to avoid delays or reduce its severity. There is a need maintenance strategy based on cost effectiveness
to optimize and manage budgets and other costs by and evaluate the cost of other alternatives.
appropriate a systems approach to aid decision-making.
3. To determine the traffic delay costs in order to
Figure 3 13 Comparison of the HA’s NSP Database with improve traffic management.
OECD model BMS
4. Calculate the discounted costs to obtain the life cycle
cost.

5. Consider the safety and congestion issues when

maintenance activities are underway.

6. Plan for a maintenance programme that is optimized

and prioritized.

7. Aid the planning of subsequent budgets.

Bridges can be grouped together based on different

criteria, such as their location (i.e. region), connection
with routes, age, condition (e.g. substandard), number of
spans and construction material (e.g. concrete, steel).

4.2 Department of Transport Structure

The Department of Transport (DoT) is responsible for

the design and implementation of the national roads policy
for the UK and also involves determining the budgets
and targets that need to be attained. It provides the funds

Stratt R.School
S. - Bridge ManagementA
of Doctoral Studies System Approach
(European Union)for Decision
Journal Making
- 2010
2010 Bridge Management A System Approach for Decision Making 75

for the HA, its executive agency for management of 4.3 Maintenance Agents
motorways, trunk road bridges and other structures and
also aids local highway authorities financially to manage The HA bears the responsibility for the bridge
their local road assets. Of the six main Directorates programmes set out by the DoT, however the other works
that make up the HA, the Network Management and are usually handled by maintenance agents who have
Maintenance, Civil Engineering and Environmental agreements with the HA regarding such work. Once the
Policy, Finance and the Roads Programme Directorate required work has been finished, a claim is submitted to
oversee the bridge management elements of programme the HA by the local authority. The remaining amount
management, allocation of funding to regions and of money, after advance payment has been deducted, is
maintenance agents, coordination and planning work as obtained from the next advance payment. Engineering
well as managing them (Bourn J, 1989). consultants, on the other hand, submit invoices to the
There are two main types of maintenance agents, who HA and are then paid by arrears when work has been
are local authorities (e.g. City of London) and consultant completed. Depending on the complexity of bridges
engineering firms (e.g. Halcrow). The duration of and scope of maintenance work, the costs incurred will
contracts are typically about five years. The agents undoubtedly vary from agent to agent.
are responsible for carrying out both inspections and Considering the various costs incurred for six
assessments on a regular basis. Although local authorities maintenance agents in the figure below, it can be seen
tend to have long term arrangements with the HA and that the costs of maintenance per square meter varies
consultant engineering firms who previously obtained (e.g. significant costs for agent A). In addition to simply
bridge and road maintenance work through bidding are awarding the maintenance contract to the lowest bidder,
now beginning to develop long term agreements with the it is important for the HA to match each agent’s specialist
HA as well (e.g. Mouchel Parkman). A diagrammatic skills and technical knowledge to the type of bridge
representation of the structure of the DoT is as follows: maintenance work, so as to minimize waste incurred
from inexperience and unnecessary delays. However,
Figure 4 1 Structure of the Department of Transport (1995) there are other factors not easily identified or quantified,
such as the relationships that the HA has developed over
time with the agents through past experiences.

Figure 4 2 Expenditure per square meter of bridge work

(1988 to 1994)

4.4. BMS in the UK & Other Countries

4.4.1 Introduction

From the Bridge Management In Europe programme

(BRIME), a review of the bridge management systems
(BMS) used in the UK, Europe and other countries

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

76 School of Doctoral Studies (European Union) Journal 2010

can be made. In order to better understand the types 4.4.4 Condition Prediction
of BMS’s and their approaches, comparisons between
them can highlight advantages and disadvantages of Most countries do not use a deterioration model to
each BMS. The countries included in this comparison predict future bridge conditions. However, Finland that
are France, Germany, the UK, Norway, Slovenia, Spain, uses a probabilistic Markovian model for the network and
Denmark, Finland, the USA and Canada (Calgaro J A, a deterministic model for individual projects (Söderqvist
1997, Woodward R J et al, 1999). M K and Veijola M, 1998). USA (New York) and Canada
Eight of the ten countries listed above use a uses past condition data in their deterioration models
computerized form of BMS, except for Germany and (Yanev B S, 1998), whereas France and Slovenia uses
Spain (Der Bundesminister fur Verkehr, 1982). France past ratings instead (Znidaric J et al, 1995).
uses a partially computerized BMS (Calgaro J A, 1997)
and the BMS’s vary between two to twenty years in 4.4.5 Cost Models
age. Countries that have a computerized BMS generally
use the commercially available ORACLE software as Most BMS’s are used to record the costs for
their database, with the UK and Slovenia as exceptions maintenance, repair and inspections, except for Germany,
(Znidaric J et al, 1995). There are no clear guidelines are Norway and Slovenia (Børre S, 1997). They generally
drawn out for bridge management in Slovenia, compared do not consider that costs of disruptions to road users
to other countries having documents that give instructions as a result of maintenance or repair work. In the UK,
concerning this. The UK uses a maintenance manual traffic delay costs are either computed by referring to
and France uses a type of management instruction. tables or using QUADRO, a computer programme for
Norway, Denmark, Finland and USA (New York) have this purpose.
user manuals provided (Andersen N H, 1990, Børre S,
1997). Germany and Spain both do not have any special 4.4.6 Decision for Maintenance & Repair
guidelines for their BMS’s (Vorshrift 169/89, 1989).
Most countries do not base their decisions concerning
4.4.2 Bridge Condition maintenance and repair on the BMS’s. Denmark uses a
prioritization method (Andersen N H, 1990) and Finland
Generally, there are three to four levels of inspection, uses a repair index (Söderqvist M K and Veijola M,
ranging from routine to special. The results are stored 1998). Canada uses a software (PONTIS) to obtain cost
in the database used, however both the UK and Spain do optimal strategies. The UK uses cost benefit analysis
not store the condition results of bridges. Germany only and France and Germany rely on engineering judgment.
stores the overall condition result for the whole bridge, Decisions for maintenance work and the selection of
without the condition results of individual elements options generally hinge on inspections and engineering
(Naumann J, 1998). The condition scale is usually a judgment. Choices made in the UK are dependent on
three to five level rating scale. the alternatives available and the cost of traffic delay.
For larger, more important projects, a whole life costing
4.4.3 Other Information in BMS’s approach is used for evaluation (DETR, 1998).

For each of the ten countries, the date, type, cost 4.4.7 Prioritization
and location of maintenance work is stored in the BMS.
Only the BMS of Denmark and Canada record the bridge The UK’s BMS does not create any optimizing
condition immediately before and after maintenance has strategy for attaining the minimum acceptable level of
been carried out (Andersen N H, 1990, Road Directorate performance of bridges for the minimum cost. However,
& Denmark Ministry of Transport, 1995). it takes into account of policies when strategies are being

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 77

formed. Finland, Canada, USA (New York), Spain and standards. Visual inspection uses a three to five scale
Denmark uses optimization techniques that can handle rating system and is stored in the database. The date,
budget constraints and Denmark and USA (New York) type and cost of maintenance work carried out over each
calculate the consequences of not selecting the optimal bridge are recorded and the pre-maintained and post-
strategy (Lauridsen J and Lassen B, 1998). maintained bridge condition is stored. For predicting the
future state of bridges, BRIDGIT employs a Markovian
Figure 4 3 Information stored on various BMS’s (Woodward, deterioration model that uses a transition probability
R J et al, 1999) matrix approach. The overall condition of a bridge is
based on the worst condition derived from these analyses.
In addition, BRIDGIT can record costs borne by road
users due to traffic delays, accidents, increased travel
time and distances as a result of maintenance activities
(Hawk H and Small E P, 1998).
It should be noted that life cycle costs are calculated
based on a twenty-year horizon and the expected
benefits are quantified on the basis of user cost savings.
Prioritization and planning of projects use incremental
cost benefit analysis methods, which can work out the
optimal cost (minimum) given a budget constraint.
4.5 BMS in the USA
4.5.2 PONTIS System
In the USA, three types of BMS systems are used,
namely PONTIS, BRIDGIT and State Specific Systems PONTIS uses a “health index” to measure the
(i.e. States that develop their own BMS). PONTIS is performance of bridges. It is possible to calculate the
the most common system, used by thirty-nine states. future health index of a bridge based on the size of budget
The BRIDGIT programme is used in the state of Maine, available. PONTIS uses an infinite planning horizon
Washington and Louisiana as a project level system. when determining the optimizing the expected life cycle
Alabama, Indiana, New York, North Carolina and costs and the benefits are based on potential savings by
Pennsylvania have developed their own BMS’s. delaying maintenance action by a year. Since PONTIS’s
objectives are to reduce the costs incurred for bridge
4.5.1 BRIDGIT System elements and overall expenditure, the prioritization
takes on a minimal cost approach that is dictated by the
From a set of alternatives that can meet the required acceptable level of performance. Bridge engineers can
level of performance, BRIDGIT provides a means to use PONTIS to create reports and graphical presentations
select the best optimal solution. This is achieved by an for the wider purposes of long-term planning, budgeting
extensive model that can review the condition states of and forecasting (Thompson P D et al, 1998).
bridges, determine policies that allocate funds optimally
and recommend appropriate maintenance activities. 4.6 SMIS System
BRIDGIT is used as a tool to manage transport
investments, plan projects and oversee programmes. The The Structures Management Information System
type of database used is called FoxPro or Visual FoxPro. (SMIS) is a new database developed by the Highways
Both individual bridges as well as bridge stock can be Agency (HA) to help achieve better maintenance and
managed using the information in the database (Hawk H operation of the road network in the UK. This system
and Small E P, 1998). aims to provide a better integration of information
Inspection information is reviewed every bi-annually which is simple to maintain, update and use, both from
and special inspections are carried out periodically on the engineer as well as the client’s point of view. SMIS
structures classified as high risk of failing to meet design is networked with all sixteen of the HA’s managing

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

78 School of Doctoral Studies (European Union) Journal 2010

agents by the agency’s extranet and information traffic The effects of safety, functionality, sustainability and the
is handled by a dedicated HA server (HA, 2003). It also environment are reviewed and given scores. By their
allows access via the Internet. combinations, a prioritization exercise aids the decision-
The four key processes of the SMIS is as follows: making process. Engineering judgment is required for
decision-making and the SMIS does not undertake any
1. Planning works. of these responsibilities as it is meant to act as a tool to
aid judgment for good structures management.
2. Input inspection data and sign-off (i.e. technical and
budge approval). 4.6.5 Project Creation

3. Defining maintenance actions and forming projects. SMIS is able to create new projects using a
combination of possible maintenance activities. The
4. Link to HA Management Information System various alternatives generated can then be chosen based
(HAMIS) for bidding & works on the minimal project cost, minimal whole life cost,
or full specification criteria (HA, 2003). For a specific
SMIS is capable of many sub-level activities and activity, a report can be generated by SMIS to provide
these are discussed in more detail below. current knowledge for the user regarding it (e.g. a
report of all the structures that need new sound barriers
4.6.1 Inventory installed). This fast and easy access to specific activities
can help engineers understand the implications of their
A database of information regarding bridges, culverts, actions, such as traffic delay due to traffic diversion, a
tunnels, retaining walls and other related data such as lane closure or a full bridge closure. The flexible use of
photographs and drawings SMIS in this way provides a means to visit all possible
alternatives and obtain estimates of expected costs.
4.6.2 Inspection & Assessment
4.6.6 Whole Life Assessment & Costing
Structures undergo a general inspection every
two years and a principal inspection every six years. The net present value (NPV) of maintenance activities
Inspection results are recorded in SMIS and the different can be calculated based on the types of decisions taken
types of recommendations proposed are recorded into to rehabilitate a faulty element of a bridge structure. The
the database. The severity of the defect detected as well cost of traffic delays is also considered in this module of
as the extent is stored at the element level. whole life assessment.

4.6.3 National Structures Programmes (NSPs) 4.6.7 Activities Schedule

NSPs aim to meet policies regarding the rehabilitation A calendar of event records the various activities that
of structures on a national level and as such interact with are being planned, with the ability of calling up reports
the maintenance needs proposed after inspections. For a for activities that are overdue or delayed. Inspections
bridge strengthening programme set out by NSPs, SMIS that have not taken place for more than two years and the
can be used also to monitor progress and track costs. number of high risk activities that are taken can also be
called up for review (HA, 2003).
4.6.4 Prioritization
4.6.8 Data Accuracy
Risk-based prioritization can be used to make
decisions concerning maintenance activities. Taking It is important that data be continually updated and
into account the consequential probability of an event accurate to maintain data integrity. When a project or
occurring if remedial actions are delayed and its individual activity has been completed, SMIS is able
consequences, a risk scoring method can be employed. to prompt for portions of the database that needs to be

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 79

updated. This allows the database to be more interactive detailed procedures such as testing of samples, using
with the user. proof loads, are usually directed towards specific studies
(Thoft-Christensen P, 1996). From these inspections, the
4.6.9 Design Specifications assessment of the bridge or bridge stock can be made
and appropriate maintenance activities can be planned
There is a requirement for approval in the SMIS for after allocating the needed funds.
by the relevant technical parties, so that the designs of Assessments of the bridges structural capacity are
structures adhere to standard codes of practice and will carried out to evaluate its load bearing capacity, which
not be in danger of any breaches. could be affected if the maximum unladen weight of
vehicles is increased. There are predominantly 5 levels of
4.6.10 Access assessments, increasing in order of detail and complexity
(Woodward R J et al, 1999). They are as follows:
SMIS operates with a password requirement for all
users and varying levels of access are given depending 1. Assessment carried out by consulting code prescribed
on the type of user operating it. methods. This process is basic and follows a similar
manner to that of the design of a bridge.
4.6.11 Integration with External Systems
2. A more detailed analysis is carried by means of
SMIS is able to interface with other external systems structural models. The same type of partial safety
of the HA, enabling better links to manage finances, data, factors as the first method is used.
projects on a wider scale. Linking up with the Project
Finance Information System, SMIS helps the collating 3. This level of assessment uses more detailed load and
process of costs in the NSPs (HA, 2003). Data from the resistance values of the bridge. Material strengths
project finance system also gives feedback to users of the can be analyzed and code prescribed reductions in
SMIS, facilitating better coordination. SMIS provides partial safety factors can be implemented.
the Pavement Management Information System with
data regarding structures, allowing the possibility of 4. A study using reliability approach can be carried out,
carrying out resurfacing or repair works concurrently. with more modifications to partial safety factors,
Interface with the HA’s Graphical Information System depending on the load and resistance requirements.
(GIS), the SMIS provides data on structures so that the
information can be mapped out and its distribution be 5. Full reliability analysis with probability outputs.
shown on a national level as well as overlaid onto the
road network. Any changes from standard practice 4.7.1 Bridge Inspection Types
can be updated in the SMIS, making certain that such
changes will be highlighted to the user. Finally, the SMIS There are four kinds of inspections for bridges in
is linked with the Accounting System and by giving the UK and can be used in combination depending on
updates quarterly, it facilitates continual reviews of the inspection needs (HA, 1994). The types of inspections
HA’s financial commitments (HA, 2003). Examples of are as follows:
how SMIS records defects, bridge elements and generate
reports are shown in the Appendix. 1. Superficial inspection of bridges by maintenance
agents of the HA. Outstanding defects that pose
4.7 Bridge Inspection & Assessment a risk to safety are highlighted and action taken
immediately to remedy them.
Bridge inspections are typically carried out by
maintenance agents that are appointed by the HA. 2. General visual inspection of bridge elements that
There are various types of inspections, as well as extent are easily accessible. These need to be carried out
of detail covered. Simple visual inspections are used at least every two years after a general or principal
to assess the general condition of bridges, but more inspection.

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

80 School of Doctoral Studies (European Union) Journal 2010

3. A principal inspection of all bridge elements, Figure 4 4 Examples of inspected bridge types and primary
including those that are difficult to gain access. These deck elements (ATKINS, 2002)
may sometimes require specialist access machinery
or tools for inspection. Such an inspection should
be carried out at least every six years and about one 01 - solid spandrel arch 02 - open spandrel arch 02 - braced spandrel arch

month before a new bridge is commissioned for use.

03 - tied arch (bowstring) 04 – Prestressed beams with deck slab
4. Special inspections involve detailed investigation of
a particular bridge component. This can suffice when
they have been excessively loaded at a point in time,
or have recently undergone a serious traffic accident.
Bridges that have been strengthened using plates 04 – Prestressed beams and deck slab
with pseudo voided slab 04 – Prestressed beams and deck
bonded to them also require special inspections. slab with solid concrete infill

Key bridge elements that are included in bridge

inspections are: foundations, piers, columns, retaining 05 – Prestressed box beams
05 – Prestressed box beam
walls, embankments, bearings, slabs, waterproofing with deck slab

material, expansion joints and so on.

06 - half through girder
Figure 4 4 Key bridge elements for inspection 07 – Filler beam with deck slab(steel beams
with concrete infill)

10 - through truss(cross
08 - underslung truss 09 - half through truss
members above vehicles)
Figure ¡Error! No hay texto con el estilo especificado en el documento.-1 Examples of
inspected bridge types and primary deck elements (ATKINS, 2002)

4.8 Defects

Defects detected from inspections are classified using

a code prescribed procedure set out by the HA. Severity
codes are used in a scale of one to five, describing the
degree of deterioration, from minor (structurally sound)
to a collapsed state (non-functional). They cover a host
of types of structures as well as aspects of the bridge’s
condition. Not all need to be included in the assessment
of a bridge, as some do not apply.
Codes are available to record defects for the following
bridge properties (HA, 1994):

1. Metalwork

2. Reinforced Concrete, Prestressed Concrete & Filler

Joist

3. Masonry, Brickwork & Mass Concrete

4. Painting & Protective Coats

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 81

5. Vegetation 3. Assign an Element Importance Classification &

Factor (EIF)
6. Foundations
4. Assign an Element Condition Factor (ECF)
7. River bed
5. Compute the Element Condition Index
8. Drainage
6. Repeat 1 to 5 for next element, otherwise proceed
9. Bridge Surfacing
7. Compute the Bridge Condition Score (BCS)
10. Expansion Joints
8. Compute the overall Bridge Condition Index (BCI)
11. Embankments
9. Repeat 1 to 8 for next bridge, otherwise proceed
12. Bridge Bearings
10. Compute the Bridge Stock Condition Index (BSCI)
13. Impact Damage
5.2 Definitions
14. Waterproofing
The definitions of terms used in the scoring system:
Pictorial examples of defects for different bridge
types, in varying degrees of deterioration are shown in Table 5 1 Definition of terms used in scoring system
the Appendix. (ATKINS, 2002)

5. BRIDGE SCORING 1 Element Condition Score (ECS)

Scale of 1 to 5 (1 for best, 5 for worst) for severity
& extent
Accounts for the value of the element, defined as
2 Element Importance Factor (EIF)
“Very High”, “High”, “Medium”, or “Low”
5.1 Introduction Accounts for an element’s contribution to overall
3 Element Condition Factor (ECF) bridge condition. Range of values depend on
element importance (see Table 5-6)
4 Element Condition Index (ECI) Element condition using ECS & ECF
In a recent publication (ATKINS, 2002) by the County 5 Bridge Condition Score (BCS) Scale of 1 to 5 for the bridge condition
6 Bridge Condition Index (BCI) Scale of 0 to 100 after conversion from BCS
Surveyors’ Society (CSS) and WS Atkins Consultants Bridge Stock Condition Index Condition of bridge stock, considers the deck area
7
(two organizations that research on strategic planning, (BSCI) of each bridge

transportation, environment and waste management and

economic issues for the HA), a proposed bridge scoring Table 5 2 Element Condition Score (ECS) (ATKINS, 2002)
system takes into account the conditions of bridge
elements, the degree of their contribution to bridge
Extent Severity
integrity and the contribution of bridges to the overall
1 2 3 4 5
bridge stock condition.
Scoring methods may be used to classify bridge A 1.0
elements according to their importance and condition. It B 1.0 2.0 3.0 4.0
is then input into an algorithm that can derive the overall C 1.1 2.1 3.1 4.1
5.0
bridge condition. By building up a standardized system, D 1.3 2.3 3.3 4.3
the condition of the bridge stock can be assessed and
E 1.7 2.7 3.7 4.7
reviewed by the bridge manager. The iterative process
*Shaded boxes represent non-permissible Severity/Extent
is as follows: combinations

1. Select a bridge, select an element

2. Assign an Element Condition Score (ECS)

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

82 School of Doctoral Studies (European Union) Journal 2010

In table above, the severity is defined as the degree Table 5 6 Expressions for Element Condition Factor (ECF)
to which the defect or damage affects the bridge element (ATKINS, 2002)
on the bridge. The extent is defined as the length, area,
or number of defects or damages of the bridge element.
Element Element Condition Factor
The shaded boxes in the table represent scenarios that are
Importance (ECF)
not permitted as there cannot be severity greater than 1.0
with extent A (“no significant defect”). The extent codes Very High ECF = 0.0
and severity descriptions are defined as follows: High ECF = 0.3 − [(ECS − 1) × 0.3 / 4]
Medium ECF = 0.6 − [(ECS − 1) × 0.6 / 4]
Table 5 3 Extent Descriptions for ECS (ATKINS, 2002) Low ECF = 1.2 − [(ECS − 1) × 1.2 / 4]

Code Description
A No significant defect
Table 5 7 Element Condition Index (ECI) (ATKINS, 2002)
B Slight, not more than 5% of surface area/length/number
C Moderate, 5% - 20% of surface area/length/number
D Wide: 20% - 50% of surface area/length/number
E Extensive, more than 50% of surface area/length/number
ECI = ECS – ECF

Table 5 4 Severity Descriptions for ECS (ATKINS, 2002) The BCS is computed based on taking into account
the contribution of each bridge element, weighted
Code Description accordingly depending on their importance. Hence,
1 As new condition or defect has no significant effect on the element
(visually or functionally). key bridge elements contribute to a larger degree to the
2 Early signs of deterioration, minor defect/damage, no reduction in
functionality of element. computed BCS using the definition below, as a low ECI
3
4
Moderate defect/damage, some loss of functionality could be expected
Severe defect/damage, significant loss of functionality and/or element is
for an important element with reduce the BCS greater
close to failure/collapse compared to using an element with lower EIF with the
5 The element is non-functional/failed
same ECI.
There are two types of BCS’s, namely BCS (average)
The various definitions for EIF, ECF and ECI are as and BCS (critical). The first is involved with all the
follows: bridge elements that contribute to the overall score,
while the latter considers the bridge elements that are of
Table 5 5 Element Importance Factor (EIF) (ATKINS, 2002) very important to the bridge’s ability to provide safety
and durability. The two types of BCS’s are computed as
follows, with ‘critical’ or ‘average’ used in the subscripts
Element EIF accordingly (ATKINS, 2002):
Importance N

∑ (ECI i × EIFi )
Very High 2.0 BCS Average = i =1
N

∑ EIF
i =1
i

High 1.5
Medium 1.2 Where N is the total number of bridge elements used.
Low 1.0 ECI for primary deck elements 
ECI for secondary deck elements 
 
ECI for half joints 
 
BCS Critical = max ECI for tie beam/rod 
ECI for parapet beam or cantilever
 
ECI for pier/column 
ECI for cross - head/capping beam 
 

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 83

It should be noted that BCS (average) could be 5.5 Bridge Stock Condition Index (BSCI)
misleading to the actual condition of a bridge, as a bridge
can be at a high risk of failure after experiencing severe In addition to considering the condition of structures
damage to a specific component during a traffic accident. on a bridge level, a broader perspective needs to be
For example, a bridge can have an important element achieved for effective bridge management. The Bridge
damaged (i.e. high EIF and low ECI) after a severe Stock Condition Index (BSCI) suitably fulfils this need.
accident. However, with other bridge elements in good To correctly reflect the contribution of each bridge, the
condition, the computed BCS (average) still indicates size of each bridge should be considered. Hence the
a “good” condition. Thus, the BCS (critical) considers deck areas are used to weight each bridge’s contribution.
the maximum ECI for bridge elements by using the The BSCI (average) and BSCI (critical) are computed
‘weakest link’ principle to identify such scenarios. The on a scale of 0 (worst) to 100 (best) and its method is as
BCS (critical) does not fully reflect how deterioration is follows (ATKINS, 2002):
spread over the bridge, hence both types of BCS’s should
M
be considered. ∑ (BCI × DeckArea )i
BCSI = i =1
M
5.4 Bridge Condition Index (BCI) ∑ DeckArea i
i =1

In order to better represent the BCS, the Bridge

Condition Index (BCI) is used, defined on a scale of 0 Where M is the total number of bridges in the bridge
(worst) to 100 (best). The determined BCS is converted stock and the deck area is computed as:
into a BCI. Similarly to the average and critical BCS
values, the BCI (average) and BCI (critical) can be 1. Area = Overall width * centerline to centerline
obtained. The relationship between BCI and BCS is distance of end supports, or,
as follows with the appropriate BCI (average) or BCI
(critical) values substituted respectively (ATKINS, 2. Area = Overall width * distance between each face of
2002): the end supports + 0.6m

BCI = 100 − 2 (BCS { ) + (6.5 × BCS ) − 7.5}

2
For the same reasons for BCS, both average and
critical values need to be considered, so as to capture
Figure 5 1 Plot of Bridge Condition Index (BCI) & Bridge impacts of a critical bridge influencing the bridge stock
Condition Score (BCS) and also weighting the BCI by the size of the bridge
using the deck area.
BCI & BCS
5.5.1 Multi Span Bridges
100

80
For a multi span bridge, inspections can be made for
60
each span. The BCI is then as follows (ATKINS, 2002):
BCI

40
S

∑ (BCI × Span Deck Area )i

20

0
1 2 3 4 5
BCI = i =1

Whole Bridge Deck Area

BCS

Where S is the total number of spans of the bridge.

By plotting out the relationship between BCI and Spans in multi span bridges are typically inspected
BCS as shown above, it can be observed that the BCI and rated individually, hence the BCI is weighted using
decreases at an increasing rate as the BCS increases. This the span deck area and computed in this way to consider
indicates that their relationship is not perfectly linear. the contributions of individual spans to the BCI. An
important span taking up a larger proportion of the whole

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

84 School of Doctoral Studies (European Union) Journal 2010

bridge deck area will invariably have a greater influence The ten selected bridge elements used in this exercise
to the computed BCI. are:

5.6 Bridge Scoring Example 1. Steel beams

Mr Paul Williams, Route Manager for the HA, 2. Columns

had been contacted for real sample bridge scoring
data. However, he had declined to assist owing to 3. Foundations
security requirements, and his email can be found in
the Appendix. Nevertheless, using the recommend 4. Bearings
approaches to determine the BSCI (average) and BSCI
(critical), original examples have been created to study 5. Waterproofing
the interactions between the various indicators and
scores on the individual bridge level as well as the bridge 6. Painting
stock. The computations have been carried out in Excel
and can be found in the Appendix. The virtual bridge 7. Road surface
stock consists of seventy-three bridges that are made up
of twenty dual span bridges and thirty-three single span 8. Retaining walls
bridges.
The following steps have been taken to create a 9. Embankments
virtual bridge stock containing seventy-three bridges for
investigation in this project: 10. Footway surface

1. Several ‘bridges’ need to be created for a bridge 5.6.1 Interpreting BCS’s

stock. Firstly, ten typical bridge elements are selected
and assumed scores are assigned to them (ECS). The BSCI gives guidance to the general condition of
Different deck areas for each span are assumed (e.g. the bridge stock and works as a tool for developing bridge
1000m2) and to make this exercise more realistic, management strategies. By comparing BSCI values with
these ‘typical’ bridges are analyzed as being made up other local authorities or at a regional level, progress can
of both single and dual span bridges. be charted accordingly and target performance can be
planned for.
2. The bridges are put into three assumed classes with For the purposes of managing individual bridges, a
each class having a generalized set of input data histogram can plot the distribution of BCI’s, or BCS’s
mentioned earlier. The three classes reflect the by bridge types, spans, or any other suitable category to
overall condition of bridges, as ‘good’ condition, highlight decisions that need to be taken. The choice to
‘medium’ and ‘bad’. The ECS values selected reflect use the BCI or BCS depends on the user’s preference,
these classes accordingly (i.e. lower scores for the as engineers who have a better understanding of BCS
‘good’ bridges than the ‘bad’ bridges’). may use that instead and help others understand better by
using the BCI. The BCI ranges from 0 (not functional) to
3. The bridge stock is then generated by means of 100 (best) and the BCS ranges from 1 (best) to 5 (worst).
multiplying each class by an assumed number to It should be noted that the BCI operates on a linear scale,
create more bridges. whereas the BCS measures deterioration on a non-linear
scale, hence more difficult to understand entirely. The
4. Using the BSCI computation, the BSCI (average) following remarks for BCS (average and critical) are as
and BSCI (critical) can be computed which takes follows:
into account the contribution of large and smaller
bridges, as well as the number of such structures.

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 85

Table 5 8 Interpretation of BCS’s (ATKINS, 2002) values, bridge type 1 comes under the ‘very good’ and
‘good’ interpretation as their BCS (average) are between
BCS (Average) BCS (critical) 1.0 and 1.5. For bridge type 2 in the bridge stock, their
(All Bridge Elements)
No significant defects in any elements;
(Worst Critical Element) BCS (average) values range between 1.5 and 2.0, which
1.0 → 1.3 Bridge is in a "Very Good" condition
Insignificant defects/damage;
puts them in the ‘good’ and ‘fair’ category for the 1st and
Capacity unaffected.
overall.
Mostly minor defects/damage; Superficial defects/damage;
2nd span respectively. For the last bridge type, the BCS
1.31 → 1.8
Bridge is in a "Good" condition overall. Capacity unaffected. (average) values are between 3.0 and 3.5, putting both
Minor-to-Moderate defects/damage;
span conditions in the ‘poor’ category.
Bridge is in a “Fair” condition overall; Minor defects/damage;
1.81 → 2.7
Capacity may be slightly affected.
For the bridge type that receives a good condition
One or more functions of the bridge may
be significantly affected. score, the differences between the BCS (average)
Moderate-to-Severe defects/damage;

Bridge is in a "Poor" condition overall;

Moderate defects/damage; and BCS (critical) are larger, compared to a smaller
2.71 → 3.7
One or more functions of the bridge may
Capacity may be significantly
affected.
difference when using the bridge types that score poorly.
be severely affected.
Possibly element failure;
This is due to the fact that the poorly scored bridges are
Severe defects/damage on a number of
elements; Severe defects/damage; also near their critical score, meaning that their lower
3.71 → 4.7 One or more elements may have failed; Capacity may be severely limit is about to be exceeded due to a critical element in
affected;
Bridge is in a "Very Poor" condition risk of failing, thus work is required urgently. Such an
overall. Bridge may need to be weight
restricted or closed to traffic observation is expected, since bridges in a poor condition
Majority of bridge elements have failed; Failure of critical element;
4.71 → 5.0
Bridge is unserviceable. Bridge should be closed.
will be invariably closer to a critical state than those in a
good condition.

The outputs for the generate bridges in the spreadsheet 5.6.3 Histograms for Bridge Stock
exercise have been classified according to the three
different bridge types defined. Bridge type 1, 2 and 3 Since the Bridge Stock Condition Index (BSCI)
represent bridges in good, medium and poor conditions provides only a single output value, much information
respectively. may be lost and it could be misrepresentative of the state
of bridges. Using histograms to plot the distribution of
5.6.2 BCS Results bridge condition for the bridge stock can better achieve
clear representation of the current state of the stock.
The bridge condition score (BCS) values have been In order to generate more ‘bridges’ in the bridge stock,
computed and plotted together, shown below: another spreadsheet that considers single span bridges is
used to create them. The spans for dual span bridges are
Figure 5 2 Plot of BCS’s for each bridge span for each bridge considered as separate ‘bridges’ since they have separate
type in the bridge stock computed scores and are inspection separately. The BCS
values when combined with that of those for the dual span
BCS's for each span for each bridge type bridges, generate the BCS (average) histogram below for
a bridge stock of seventy-three ‘bridges’ (twenty dual
3.5

3.0

2.5
span bridges and thirty-three single span bridges):
BCS 1st span(avg)
BCS Value

2.0 BCS 1st span(crit)

1.5 BCS 2nd span(avg)
Figure 5 3 Histogram plot of BCS (average) values
BCS 2nd span(crit)
1.0

0.5
BCS (average) Histogram
0.0
30
No. of bridge spans

Bridge Type 1 Bridge Type 2 Bridge Type 3 24

Bridge Stock 25
20
15
15 13
10
10 6
It can be seen that for the bridges that are classified 5
5
0 0
to be in the ‘bad’ state, the BCS (average) values for 0
-5
.5

.0

.5

.0

.5

.0

.5

both the 1st and 2nd spans are close to their respective
-1

-2

-2

-3

-3

-4

-4

5
4.
0

0
1.

1.

2.

2.

3.

3.

4.

Range
BCS (critical) values. Considering the BCS (average)

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

86 School of Doctoral Studies (European Union) Journal 2010

From the histogram above, it can be observed that the this potential is far from linear, as the costs to maintain
majority of bridge spans are classified to be well within a good bridge are lower than compared to improving a
the lower half of the BCS range, indicative of a fairly severely deteriorated bridge (ATKINS, 2002)
healthy bridge stock. However, since some or these are
in the ‘poor’ classification, remedial work can then be 5.6.5 BCI Results
planned and carried out to create a phase shift towards
the lower BCS range. Statistical approximations can be The BCI’s for each span for each bridge type is shown:
fitted to model the distribution of BCS’s, such as a gamma
distribution. A normal distribution is not expected to be Figure 5 5 Plot of BCI’s for each bridge span for each bridge
a good fit, due to imposed lower bound value of 1.0 and type in the bridge stock
upper bound value of 5.0. Similarly, the BCS (critical)
histogram is obtained: BCI's for each span for each bridge type
100

Figure 5 4 Histogram plot of BCS (critical) values 80

BCI Value
60

BCS (critical) Histogram

40 BCI 1st span(avg)
40 37
BCI 1st span(crit)
No. of bridge spans

20 BCI 2nd span(avg)

30
BCI 2nd span(crit)
21
0
20 15 Bridge Type 1 Bridge Type 2 Bridge Type 3
Bridges Stock
10
0
0
From the results, bridge type 1 gives a high percentage
.0

.0

.0

.0
-2

-3

-4

-5

of ‘service potential’ and on the other extreme bridge type

0

Range
1.

2.

3.

4.

3 has lost almost 50% of its service potential. Following

a similar pattern to the BCS results obtained, the BCI
Approximately twenty-five per cent of the bridge (average) values for bridge type 1 differ greater with the
spans are between the ranges of 3.0 – 4.0, indicating that corresponding BCI (critical) values for the same bridge
there is a significant proportion experience reduced load type, than compared for bridge type 3. This is owing
carrying capacity due possible element failure or moderate to the situation where bridge type 3 spans are close to
damage. The BCS (critical) indicates the contribution of reaching their threshold value and that maintenance
key elements to the overall condition score and hence for work should be carried out as soon as possible in order
these bridge spans, imposing load limits may be required, not to risk the loss of important bridge elements.
pending further structural analysis of the bridges. Using
histograms to display information gives more scope of 5.6.6 Interpreting BSCI’s
detailed statistical analysis if necessary, as well as serves
as a useful tool in budgeting and sharing information to The BSCI values (average and critical) serve as an
other non-technical parties. aid for high-level strategic decisions that involve budget
allocation for boroughs, councils and local authorities,
5.6.4 Interpreting BCI’s which can be expected to be at the regional or national
level. Funding to increase BSCI values are not linear,
As explained earlier, BCI operates on a linear scale since it is more costly to improve conditions for a low
from 0 (not functional) to 100 (best). Since it is also BSCI score than a high BSCI score. Since the single BSCI
similar to the percentage scale, it can be used as a form output value computed does not give full information
of performance percentage indicator of its ‘service of the bridge stock, the BCI values that used a range of
potential’. Hence a “0” condition index translates to 0 to 100 can be used in conjunction with BSCI’s. The
no potential for service and for “100”, full potential is guidance for interpreting BSCI values (ATKINS, 2002)
available. It should be noted that the costs to improve is as follows:

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 87

Table 5 9 Interpretation of BSCI’s (ATKINS, 2002) 6. PREDICTING BRIDGE CONDITION

WITH TIME
BSCI BCS Bridge Stock Condition Bridge Stock Condition
Range Range based on BSCI Av erag e based on BSCI Critical

100 → 95
Bridge stock is in a very
good condition. Very few
Very few critical load-
bearing elements may be in
6.1 Introduction
Very 1.0 → 1.3 bridges may be in a a moderate to severe
Good moderate to severe condition. Represents very
condition. low risk to public safety.
A few critical load-bearing
In addition to being able to inspect bridges and assess
Bridge stock is in a good
94 → 85
1.31 → 1.8
condition. A few bridges
elements may be in a
severe condition.
their current state, good bridge management includes
Good may be in a severe
condition.
Represents a low risk to
public safety.
the ability to predict the bridge condition in the future.
Bridge stock is in a fair
Wide variability of
conditions for critical load
This will enable bridge engineers to deal with the
condition. Some bridges
may be in a severe
bearing elements, some
may be in a sever
uncertainties regarding the overall structural integrity of
condition.
84 → 65
Fair
1.81 → 2.7 Potential for rapid decrease
condition.
bridges, as well as be capable of coping with the changing
Some bridges may
in condition if sufficient
maintenance funding is not
represent a moderate risk conditions of bridges within the bridge stock. Being able
to public safety unless
provided. Moderate backlog
of maintenance work.
mitigation measures are in to predict future bridge performance enables decisions
place.
Bridge stock is in a poor A significant number of to be made early regarding setting target performance
condition. A significant critical load bearing
number of bridges may be elements may be in a levels, planning and scheduling of maintenance as well
in a severe condition. severe condition.
64 → 40
Poor
2.71 → 3.7 Maintenance work Some bridges may
as budgeting for future expenditure. Preparedness is an
historically under funded
and there is a significant
represent a significant risk
to public safety unless
important element of good bridge management and can
backlog of maintenance
work.
mitigation measures are in
place.
be increased by using suitable prediction models.
Bridge stock is in a very Many critical load-bearing
poor condition. Many elements may be

39 → 0
bridges may be
3.71 → 5.0 unserviceable or close to it.
unserviceable and are in a
dangerous condition. 6.2 Markov Chain Approach
Very Poor
Maintenance work is under Some bridges may
funded with a huge backlog represent a high risk to
of work. public safety. The Markov Chain approach is a system applicable to
bridge management due to its simplicity and properties.
Figure 5 6 Histogram plot of BCI (average) values It can be employed to model the changes in a bridge
condition over time. More importantly, it can be extended
BCI (average) Histogram
28
to model the make-up of a bridge stock over time. This
30 27
approach is applicable when a discrete stochastic system
No. of bridge spans

25

20 possesses Markovian properties (RIMES, 1999).

15
15
These are:
10

5 3
1. The state space (i.e. possible values) are discrete;
0
40-65 (poor) 65-85 (fair) 85-95 (good) 95-100 (very good)
Range
2. The probability of X at t + 1 = k given the previous X0
= k0; X1 = k1, …Xt = j is equal to the probability of X at
The histogram above shows more detailed information t + 1 = k given Xt = j.
than using the BSCI and hence the use of BCI (average)
values is very useful. Each class range from ‘poor’ to This approach assumes ‘no memory’ and the
‘very good’ are not the same, as it is meant to reflect the probability of at the next time step is only dependent on
different conditions of bridges. It can be seen that for a the immediate previous time step. In bridge management,
bridge stock of seventy-three bridges, fifteen are in the bridges are often assessed on a twenty-year horizon. By
‘poor’ category, which is almost 20% of the bridge stock means of allocating the number of states the bridges
by proportion that needs careful attention. The majority can be in, the probability of bridges remaining in that
of the bridge stock is either ‘fair’ or ‘good’ which is a state for each year can be computed. The initial state
positive sign. Three bridge spans have been given the value is 1.0. The subsequent state values for each stage
‘very good’ classification. These may relatively new (twenty stages) are computed, by means of summing up
bridges that have been built, or bridges that are still at an the products of the stage transition probabilities with the
early stage of their service life. stage number (weighting), explained below:

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

88 School of Doctoral Studies (European Union) Journal 2010

N
a linear scale of 0 (worst) to 100 (best). The degree of
StageValue = ∑ P ( state)i
i =1 severity of bridges is linearly distributed over this range
(i.e. BCI of 50 to 51 is the same as 90 to 91), except that
where for the current stage, costs are expected not to have a linear distribution. This
is a useful approach as the BCI (average) is interpreted as
P(state)i = (1 - P(state)i-1 )previous stage * P(state)i + ‘service potential’ and is used as a performance indicator.
P(state)i * (P(state)i)previous stage Using the example for multi span bridges earlier on,
the transition probabilities for a three-state Markov
and N is the number of states used (i.e. three states) chain model with limiting stage value of 3 is proposed.
The probabilities are in accordance to the BCI values
6.2.1 Example Calculation for the ‘good’ bridge arranged in order of descending
magnitude (i.e. P(1) = 0.9845 and P(2) = 0.9246). For
P(1) 0.99 Q(1) 0.01 the purposes of comparison, the other two bridges
P(2) 0.95 Q(2) 0.05
P(3) 1 Q(3) 0 (‘medium’ and ‘bad’) are also modelled and the three are
Stage 0 Stage 1 Stage 2 plotted together.
State 1 1 0.99 0.98
State 2 0 0.01 0.02 A summary of inputs is as follows:
State 3 0 0 0
Stage value = 1.02
Table 6 1 Input transition probabilities for bridge condition
example
An example is shown above to illustrate the
computation for stage 2 of a three-state model. In the Bridge Condition P(1) P(2) P(3)
Good 0.9845 0.9245 1.0
above table, the probability of transition from state 1 to Medium 0.8859 0.8708 1.0
Bad 0.5734 0.5369 1.0
2 is 0.99 (i.e. Q(1) = 0.01). The probability of transition
from state 2 to 3 is 0.95 and Q(2) = 0.05. The computation
for the shaded cell is: Figure 6 1 Bridge condition with time

0.01*0.99 + 0.95*0.01 = 0.02 Bridge condition with time

And the stage value is the sum of the product of the

3
stage probabilities and state values: good
State

medium
2 bad
Limit
Stage value = 1*0.98 + 2*0.02 + 3*0 = 1.02 1

The limiting state is absorbing, such that bridges 0

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

that have deteriorated in that state remain in that state Years

(i.e. transition probability is 1) until maintenance is

carried out. By plotting out the bridge condition with
time over twenty years, the time which exceedence of 6.3.1 Bridge Condition Results
the limiting state value occurs can be located. This aids
bridge engineers by providing a means of probabilities Observing how the stage changes with time over
of ‘failure’ as well as enabling the information to be the twenty-year horizon, it is plain that for the bridge
translated to respective costs and knowing when to act in ‘bad’ condition, the limiting stage is reached quickly
accordingly. after ten years. For both the ‘medium’ and ‘good’ bridges
the maximum not exceeded and the ‘medium’ bridge is
6.3 Bridge Condition Example approaching the limit first. To take this example further,
it is possible that the limiting stage value be defined as
It is proposed that the transition probabilities to be used the BCS, thereby increasing the limiting stage value to 5
are the Bridge Condition Index (BCI), which operates on instead of 3. However, it should be noted that the BCS

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 89

does not reflect the extent of severity linearly as BCI Figure 6 2 Bridge condition with time using aggregation
does and the additional inputs that need to be generated
(i.e. new P(3), P(4)) either by more extensive bridge type Bridge condition with time

modelling or judgment. 4

3
6.4 Bridge Aggregation Example very good
good

State
2 fair
poor

Due to the need for considering the changes in bridge 1

Limit

condition over time for the entire bridge stock, it may be

computationally exhaustive to carry out analyses using 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

each bridge’s BCI value. A single output related to the Years

BCI classifications can be used more easily for high-

level strategic decision-making.
As such, after an evaluation of the distribution of 6.4.1 Bridge Aggregation Results
bridge conditions in the bridge stock, bridges can be
suitably aggregated into the 4 possible categories, It can be seen that the bridges in ‘poor’ condition
depending on their BCI values (previously plotted using deteriorate very quickly and plateau off to reach the limit
histograms). These are poor (40 to 65), fair (65 to 85), after about ten years. The bridge in ‘very good’ condition
good (85 to 95) and very good (95 to 100). The transition deteriorates very slowly and the one in ‘fair’ condition
probabilities may be assigned as the lower and upper just about reaches the limit after twenty years. For the
limits of each category to generate the four-stage model. one in ‘good’ condition, the condition is not too severe,
100% for the ‘very good’ BCI classification cannot be however, maintenance would be necessary to prevent
used, as bridges will definitely not remain perfectly in it from reaching the severe state. The deterioration of
the same condition in the next year, 99% is proposed bridges is clearly non-linear; increasing at a decreasing
as an alternative. Taking a step further, it is possible to rate and it is explained by the greater ease of bridges
aggregate bridges in to several different categories using to deteriorate from a very good condition (e.g. brand
other definitions of ranges to work with, depending on new) to a poorer state. This has implications for the
the bridge management needs. scheduling of maintenance and costs that are involved
The input probabilities are as follows: for the bridges in the bridge stock that come under any
of the four categories. Taking this example further, the
Table 6 2 Input transition probabilities for bridge limiting stage value can be increased also to 5 instead of
aggregation example 3, similar to the BCS scale.

Bridge Condition
Very good
P(1)
0.99
P(2)
0.95
P(3)
1.0
6.5 Bridge Stock Example
Good 0.95 0.85 1.0
Fair 0.85 0.65 1.0
Poor 0.65 0.40 1.0 To further examine how the conditions of bridges
in the bridge stock change with time, the proportion
for each bridge condition in the bridge stock can be
Using the classification similar to those for interpreting modelled. This can be achieved by using a deterioration
BCI values, the output is as follows: model of the form:

State 0: p(1) q(1) 0

State 1: 0 p(2) q(2)
State 2: 0 0 1

In this three-state model, state 0 represents the best

condition and 2 the worst condition, which is also the
absorbing state. It is possible for states to move from

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

90 School of Doctoral Studies (European Union) Journal 2010

0 to 1, or 1 to 2, but not backwards. Bridges in state 2 Figure 6 4 Distribution of bridge stock condition with time
remain in state 2, unless maintenance is carried out to (fair conditions)
improve the condition. p(1) is the probability that given
a section is in state 0 now, it will remain in the same state Fair Conditions
the subsequent year. q(1) is the probability that given 1

probability distribution
a section is in state 0 now, it will transit to state 2 the 0.8
subsequent year. state 2
0.6
Due to the constraint of total probability, state 1
0.4 state 0

p(1) + q(1) = 1, 0.2

0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
and p(2) + q(2)= 1 Years

Only p(1) and p(2) need to be entered into the model

to define the transition probability matrix. Figure 6 5 Distribution of bridge stock condition with time
(bad conditions)
Applying to a bridge stock model, p(1) represents the
proportion of bridges in the bridge stock being in state 0 Poor Conditions

will remain in state 0 the following year. This is similarly 1

applied for p(2). To analyze the different behaviour of probability distribution
0.8
how the bridge stock’s proportion of bridges are made state 2
0.6
up of over time, it is proposed that three types of bridge state 1
0.4
stock behaviour be used, for simplification, to be ‘good’, state 0

‘fair’ and ‘poor’. The different scenarios with different 0.2

probabilities are as follows: 0

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

years
Table 6 3 Input transition probabilities for bridge stock
example
6.5.1 Bridge Stock Results
Bridge Stock Condition P(1) P(2)
Good 0.95 0.90
Fair 0.90 0.80
Bridge stocks subject to good conditions can represent
Poor 0.80 0.70 maintenance being carried on time, to schedule, with
sufficient allocation of funds. This is the opposite for
The various bridge stock distribution plots showing bridge stocks subject to poor conditions. However,
the bridge stock profiles are as follows: several other factors may come into play such as the
location of bridges as a certain bridge stock may have a
Figure 6 3 Distribution of bridge stock condition with time large proportion of its bridges being sited near the coast
(good conditions) and be subject to severe corrosion by sea spray.
By observing the three types of condition plots, the
Good Conditions bridge stock that is subject to good conditions have a
1
higher proportion of bridges that still remain in the state
probability distribution

0.8 0 when compared to the bridge stock that is subject to

0.6 poor conditions. In the good condition scenario, the
0.4 proportion of bridges in the bridge stock that are in state
0.2
state 2 1 increases from zero over time. For both fair and poor
state 1
state 0
conditions, the number of bridges in state 1increase and
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 then begin to decrease when the years approach twenty.
Years
This is due to the bridges in the bridge stock deteriorating

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 91

towards state 2. By the end of twenty years, less than half Vehicles (HGV), with information on alternative
of the bridge stock is in state 2 for the good condition routes, additional time for travel and distance to
scenario and for fair conditions, it is slightly under 80%. travel using these alternatives.
For the poor conditions, a very high percentage (about 3. The history of imposed restrictions on load, height
97%) of the bridge stock has deteriorated to the state 2 and width is also recorded.
condition.

7. TRAFFIC COSTS
The different types of bridge maintenance related
7.1 Introduction costs and how they are quantified are mentioned below:

In the UK, QUADRO (Queues And Delays at 1. Costs due to traffic delays, quantified in terms of
Roadworks) is a computer program used to determine average delay time and value of time of the road user.
the traffic delay costs due to maintenance activities and 2. Costs due to traffic having to detour, using
as such taken on board by the HA as a tool to aid the information provided concerning length of detour,
planning of roadworks. With respect to the management value of time and additional travel time.
of bridges, examples of such works are resurfacing, 3. Accident costs and congestion costs, caused by
general maintenance, waterproofing of bridge decks. reduction in traffic flow speed, also quantified using
QUADRO takes into account the costs borne by the user, value of time of the road user and typical costs of
in terms of value of time, cost of operating vehicles and accidents.
costs due to accidents. Over time, several versions of 4. Environmental impact and social costs are relating to
QUADRO have been improved upon and integrated emissions of additional toxins and greenhouse gases
with cost benefit analysis packages which can extend caused by detouring traffic or congestion. These
computations for all elements of a project that need to be can be estimated using vehicle speeds and detour
reviewed before obtaining approval (DETR, 1998) information.
QUADRO can be used to help engineers decide what
optimal strategies are available to carry out roadworks 7.2 Traffic Count Example
in terms of scheduling of works and when they should
be carried out. This is particularly useful for a series of Several traffic counts were carried out in ten-minute
works extending over varying locations. The program intervals on 19th May 2004 on two bridges in the Central
can also be used to evaluate proposed schemes and London area, namely Blackfriars bridge and Southwark
review the potential impacts of implementing them. bridge. Six counts were carried out in total for Blackfriars
Although QUADRO has been designed for use in rural bridge and four for Southwark bridge. These bridges are
areas, it can also be implemented in urban areas, with located in a congestable area with different number of
some limitations due to the complexity of junctions, lanes carrying road traffic. The traffic counts included
configurations of diversions and invariably unpredictable all road users, which were classified as cars, bicycles,
elements of road user behaviour (DETR, 1998). Different HGV, motorcycles and buses. A vehicle was classified
types of information need to be available and input to the as a HGV if it was larger than a car in size and not a bus.
QUADRO program. They are: This does not include small delivery vans as such.
The time of the traffic counts were carried out during
1. Record of maintenance works carried out previously, the morning peak hours, although some modification
with details on date, type and location of such are required for weekend and off peak hour behaviour.
activities. The cost of maintenance is also used. The The costs borne by pedestrians also need to be assessed
duration and severity of traffic delay are taken into but time did not allow for a full pedestrian count to be
consideration. carried out. The pedestrian count is therefore estimated
2. The traffic history of areas are used, defined in to be 2000 per hour for Blackfriars bridge and 1500 per
standardized means such as the Annual Average hour for Southwark bridge during peak hours.
Daily Traffic (AADT), proportion of Heavy Goods

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

92 School of Doctoral Studies (European Union) Journal 2010

7.2.1 Traffic Count Results It is observed that for Blackfriars bridge, there is a
larger proportion of cars travelling in the southbound
Using both northbound and southbound counts, an direction than the other way. Both car proportions for
averaged traffic count and traffic composition are as northbound and southbound are increasing with time up
follows: to about 10am in the morning. In the same manner, the
proportion of buses noted were decreasing. The larger
Table 7 1 Flows from traffic count on Blackfriars & proportion of HGVs travelling earlier in the morning is
Southwark bridges observed and this is attributed to the fact that congestion
builds up at morning hours and hence the need for
Flow Blackfriars Southwark earlier travel. Goods need to be unpacked, distributed
Total 1504 477
Cars per hr 1005 306 and get ready for the next process of shipping or selling
Bicycles per hr 218 72
HGV per hr 81 36
and hence extra time is required, compared to people
MC per hr 165 51 travelling to work who need only to travel to the office
Bus per hr 35 12
from the carpark (less transit time).
The number of bicycles in the southbound direction
The make up of traffic for Blackfriars bridge for both is close to constant and for the northbound direction it
directions are plotted as follows: seems to decrease as time approaches 10am. This is due
to the cyclists getting to work earlier so that they have
Figure 7 1 Traffic breakdown for Blackfriars bridge adequate time to change clothes, or freshen up before
(northbound) starting work. It is possible also that since congestion
is peaking, their considerations for greater safety with
Blackfriars (Northbound) less traffic affects their decision to travel earlier in the
100% morning. It is believed that more employment avenues
80% are located at the south of the river and hence a larger
Bus
proportion of road users are made up of cars.
Proportion

60% Motorcycles
HGV The following observations are made in the absence
40% Bicycles of a complete statistical survey of the bridges involved
Cars
20% which is expected to take a disproportionate amount of
0%
time when compared to other analysis and discussions
0840-0855 0935-0945 0955-1005 needed in this project. Hence, it has been accepted
Time
that although some slight inadequacies will invariably
remain in this exercise, however, they are not severely
7 2 Traffic breakdown for Blackfriars bridge (southbound) misrepresented. The make up of traffic for Southwark
bridge for both directions are plotted as follows:

Blackfriars (Southbound) Figure 7 3 Traffic breakdown for Southwark bridge

100% (northbound)
80%
Bus
Proportion

Motorcycles
Southwark (Northbound)
60%
100%
HGV
40% Bicycles
80%
Cars Bus
Proportion

20%
60% Motorcycles
0% HGV
40%
0820-0830 0925-0935 0945-0955 Bicycles
Time 20% Cars

0%
0855-0900 1040-1045
Time

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 93

Figure 7 4 Traffic breakdown for Southwark bridge 2. A reduction percentage is used to adjust for decreased
(southbound) occupancy over time, due to increased income levels
which allow people to purchase more cars. These
Southwark (Southbound) values are 0.05% per year for working vehicles and
100% 0.22% for non-working vehicles.
80% 3. To adjust for increase in Gross Domestic Product
Bus
(GDP), an indicator of wealth and hence related to
Proportion

60% Motorcycles
HGV value of time and annual rate of 3.28% has been
40%
Bicycles applied up to 2002 then 2.19% up to 2004.
Cars
20% 4. The split proportion between working and non-
0% working vehicles needs to be entered. Since the
0900-0905 1035-1040
traffic counts were carried out during a typical peak
Time
hour period, a larger percentage is used for working
vehicles (e.g. 75% or 80% working vehicles, with
For Southwark bridge, there is a similar pattern the remaining as non-working).
emerging for the northbound route, with the proportion 5. Fuel costs are then computed, using appropriate cost
of cars increasing as time goes by. The proportion of coefficients.
bicycles is large for the earlier hours of the morning. The
proportion of HGVs seems to be unchanged. For the The cost of fuel is estimated using the following
southbound direction, the number of cars decline for the function (DoT, 2003):
second period of the traffic count. This can be attributed
to the people commuting to work have reached their place C = a + bV + cV2
of work since it is mid-morning. The HGV proportions
are larger than the first period, which is explained by Where
less congestion after the morning peak hour congestion
allowing them to move more easily and hence, this time C = cost in pence per km of travel,
of travel is selected.
V = average speed in kilometers per hour,
7.3 Traffic Delay Cost Examples
a, b and c are parameters defined for each vehicle
Using the basic information above, costs of traffic category.
delays have been computed using Excel to investigated
the interaction between the various components of traffic The non-fuel resource cost are combined in a formula
and their contribution to traffic delay costs (e.g. traffic of the form (DoT, 2003):
composition, length of detour, duration of delay).
The following steps have been carried out, using the C = a1 + b1/V
traffic survey data collected and guidance notes (tables)
that are in the Appendix: 6. Estimates for the link speeds are used. It is estimated
1. Input the annual averaged value of time per that cars travel at 40km/h, for HGV it is 30km/h, for
vehicle for each vehicle type for 1998 scenario. bicycles it is 15km/h, for motorcycles it is 40km/h,
This includes the different types of cars, such as for buses it is 35km/h and for pedestrians it is 8km/h.
‘working’ car, or ‘non-working’ car, which relates These speeds are for free flowing traffic.
to the purpose of travel as working or only to and 7. Correction percentages for increased fuel cost are
from work respectively. When HGV information then used as well as fuel consumption corrections
is unavailable, they are delegated values given for to account for increase in fuel and the growing
LGV instead. The working and non-working costs efficiency of engines in prices over time. The fuel
are allocated to all other vehicles except buses which cost per kilometer for each vehicle type and purpose
are standardized. can then be obtained. It should be noted that fuel

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

94 School of Doctoral Studies (European Union) Journal 2010

is not needed for bicycles, but to consider the costs Table 7 3 Cost of detour due to delay from example
of the cyclist expanding energy (a form of cost to
user), costs to bicycles are a proportion of that Cost of detour per day, £
Blackfriars
400 701.43
Southwark
209 813.07
used for vehicles, but without adjustments made to Cost of detour per day (half closure), £ 354 147.06 184 295.78

increase in ‘fuel consumption’ and ‘costs’. The costs

allocated to pedestrians have been selected to be a The proportional costs depending on each situation is
proportion of the costs borne by cyclists. shown below:
8. The detour distance (i.e. additional distance)
experienced by road users is estimated to be 3.0km Figure 7 5 Breakdown of costs (per hour) for each bridge
and additional time taken to travel this additional
distance varies with each mode of travel. For full Costs for each bridge per hour

bridge closure and half bridge closure, the costs 40000

Travel

incurred will not be directly proportional since the 30000 Detour cost

Costs, £
responses to road user behaviour is expected to be 20000

highly non-linear, as well as the complex traffic 10000

interactions with alternative routes. The estimate
0
reductions in speeds by a proportion of free flow Blackfriars (full Blackfriars (1/2 Southwark (full Southwark (1/2

speed is selected and then computed separately. closure) closure) closure) closure)

Situation
The bridge length is estimated to be of the order of
0.35km and hence the two types of additional cost
(full bridge closure and half bridge closure) to road The cost of detour per day for Southwark bridge is
users due to detour and delay can be computed. lower than that for Blackfriars bridge, due to the lower
9. To adjust for off peak travel behaviour and respective flows of traffic. The costs seem to be conservative in
costs incurred, it is estimated that there daily, there nature as both Blackfriars and Southwark bridge are very
are 4 peak hours, 8 off peak hours and 12 quiet hours important bridges in the Central London area. It is likely
(i.e. at night). The costs each road user experiences in that some uncaptured elements of traffic delay costs were
the off peak and quiet hour period are expressed as a not accounted for, such as a different set of costs borne by
percentage of the peak hour, for simplification, taken travellers if information concerning the delay was made
as 50% and 30% of the peak hour costs respectively. available and their selected mode of travel changed.
10. The costs of detour per day for both a full bridge No attempt has been made to quantify the costs borne
closure and half bridge closure are then computed. by other travellers in the vicinity (e.g. Underground,
This accounts for congestion due to reduced capacity river boat service) due to the exponential increase of
of the bridge when 1 lane is not in service, as well complexity in both data collection and computation. It
as the congestion that builds up in alternative routes is unknown how road users would change their mode of
which road users choose to take. travel or how the available configurations of alternative
routes will be utilised as well as the degree of utilisation.
7.3.1 Delay Costs Results (1st Example) It is recognised that a high degree of uncertainty
inherently lies with the computation carried out. The
A summary of delay costs is as follows: classification of proportion of working and non-working
vehicles, average link speed on the bridges, detour
Table 7 2 Summary of delay costs from example lengths and additional travel time, proportion of costs
allocated to off peak and peak hours and quiet hours,
Cost per mode of travel, £
Cars
2.20
HGV
4.05
Bicycles
2.05
Motorcycles
5.39
Bus
37.70
Pedestrians
0.68
all contributed to the uncertainties of the computation.
Additional delay & detour
(full closure), £
4.70 11.58 11.71 11.55 92.32 7.32 However, with refinement, more traffic counts and better
Additional delay & detour
4.03 9.92 10.03 9.90 79.13 6.27 understanding of how the complex interactions of road
(1/2 closure), £
user behaviour and congestion function, an improved
estimate can be produced.

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 95

To investigate a potential upper value of traffic delay 2. The proportions of accidents are then allocated,
costs, higher values have been used for the inputs. to three categories, namely ‘fatal’, ‘serious’ and
An increased percentage of working vehicles (85%), ‘slight’, in descending order of severity (1%, 14%,
increased number of off peak hours (14 hours) and 85% respectively) using the prescribed guidance by
reduction of silent hours (6 hours) have been used. the Department of Transport (2003). The number of
casualties per accident is then assumed.
7.3.2 Delay Costs Results (2nd Example) 3. Using a percentage correction factor (2%) to adjust
for the changes in cost over time, the average cost
A summary of delay costs is as follows: per casualty and per accident is used for the final
computation of the total cost of accidents per year.
Table 7 4 Summary of delay costs from example (2nd example)
7.4.1 Accident Costs Results
Cars HGV Bicycles Motorcycles Bus Pedestrians
Cost per mode of travel, £ 2.20 4.05 2.05 5.39 37.70 0.68
Additional delay & detour
(full closure), £
4.70 11.58 11.71 11.55 92.32 7.32 A summary of accident costs is as follows:
Additional delay & detour
4.03 9.92 10.03 9.90 79.13 6.27
(1/2 closure), £
Blackfriars Southwark
Fatal Serious Slight Fatal Serious Slight
Cost of accidents, £ 39 504.69 62 510.82 30 751.00 12 538.51 19 836.66 9976.99
Total accident
132 766.50 42 152.16
Table 7 5 Cost of detour due to delay (2nd example) cost per year, £

Cost of detour per day, £

Blackfriars
442 137.86
Southwark
231 513.17
Due to the simplified inputs of casualties per accident
Cost of detour per day (half closure), £ 390 769.03 203 356.61 and proportion of fatal, serious and slight accidents for
both Blackfriars and Southwark bridge, the proportional
The costs have now increased by 10.3% as a result. contributions of cost for each type of accidents are the
More detailed statistical surveys of traffic flows, as well same. In reality, this is expected to be different. Using
as the modelling of traffic behaviour under different the costs of accidents by type (adjusted for percentage
delay circumstances can yield more accurate results, but changes each year), fatal accidents make up about 30%
these generated for this example satisfies the key design of the total costs, with serious injuries accounting for
procedures and considerations. 47% and slight injuries 23%. This is different from the
proportion of accidents allocated which are 1%, 14% and
7.4 Accident Cost Example 85% respectively. This is difference is observed due to
the high cost of fatal accidents which do not occur very
In addition to the costs incurred due to traffic delays, often. Therefore in terms of costs, their contributions
traffic accidents that occur on bridges can severely are larger overall, similarly for serious injuries that make
increase the costs incurred by road users. These are up 14% of injuries, but accounts for 47% of the accident
considered separately and can be used in combination costs.
to traffic delays. How these two complex mechanisms
interact remains unknown and it is expected to require Figure 7 6 Breakdown of cost of accidents on Blackfriars
extensive computational analyses. For the purposes of bridge
understanding accident costs, a basic model has been Adjusted cost of accident
developed. (Blackfriars)
The accident costs borne by road users are estimated Slight
Fatal
23%
using the following procedures: 30%

1. The accident rate per hundred million vehicle Fatal

kilometers is used, scaled to the volume of traffic Serious

from the traffic survey data; therefore the number Slight

of vehicle kilometers travelled per year is computed

first. Serious
47%

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

96 School of Doctoral Studies (European Union) Journal 2010

By means of another iteration using an increased 1. 100% free flow (no congestion)
percentage of working vehicles (85%) and increased 2. 80% free flow (little congestion)
number of off peak hours (14 hours) with a a reduction 3. 60% free flow (moderate congestion)
in silent hours (6 hours), a summary of accident results 4. 40% free flow (severe congestion)
is as follows: 5. 20% free flow (very severe congestion)

Table 7 6 Cost of accidents for Blackfriars & Southwark The following procedure is carried out according to
bridges DMRB (1998):

Fatal
Blackfriars
Serious Slight Fatal
Southwark
Serious Slight
1. Input of vehicle speed, proportion of HGV, receptor
Cost of accidents,
£
43 589.95 68 975.77 33 928.47 13 834.17 21 887.05 10 784.74 distance.
Total accident cost
146 494.20 46 505.96
2. Compute the number of cars and HGV. For
per year, £
simplification, buses and HGVs come under the same
heading and bicycles are removed from the analysis.
There has been a 10.3% increase in costs of accidents Motorcycles are included in the cars category for
for both bridges, with the types of accidents taking simplification since this is the less conservative than
the same proportions as the earlier scenario due to the if they were to be classified as HGVs.
simplified inputs. The classification of peak hours and 3. Speed correction factors for both cars and HGV are
silent hours can by defined differently, as they can be first applied, followed by year correction factors, to
defined accordingly to road users, residents in the vicinity, account for changes in emissions with speed and
or to the local authority urban planning guidelines. The more efficient engines being developed over time
developed model can thus be used in to investigate further respectively.
the impacts of these variables. It is expected that there 4. Correction factors for flows to take into account the
are upper and lower bound values to these variables, number of vehicles; distance correction factors are
such as the percentage of working vehicles not reaching then applied. The final outputs (per hour) are then
a full 100% and the number of peak hours to be greater obtained.
than one and less than twelve (i.e. half a day).
8.2.1 Emissions Results
8. ENVIRONMENTAL IMPACT
A summary of emissions is as follows:
8.1 Introduction
Table 8 1 Summary of emissions calculations
When traffic congestion builds up due to an accident
delays or maintenance works, the amount of emissions Units
% of free flow
parts per million (ppm)
CO
parts per billion (ppb)
NOx
parts per billion (ppb)
HC
from vehicles increase because of reduced vehicle 100
80
0.239
0.314
103.046
118.211
64.119
79.705
speeds. These gases comprise of greenhouse gas carbon 60 0.366 129.880 91.021
40 0.523 171.760 132.167
dioxide, particulates, lead compounds, hydrocarbons, 20 0.682 216.370 181.595

compounds of nitrogen oxides and carbon monoxide.

The UK’s National Air Quality Strategy (NAQS)
8.2 Emissions Example objectives for reduced emissions for 2005 include
reductions of NOx and HC emissions to not more than
For Blackfriars bridge, three key emission gases are 150 parts per billion (ppb) and for CO emission it is 1.25
modelled in various scenarios to investigate the impacts parts per million (ppm) (DMRB, 1998). The modelled
on the environment, namely carbon monoxide (CO), results and superimposed NAQS limits are shown in the
nitrogen oxide compounds (NOx) and hydrocarbons following plots:
(HC). The different scenarios investigated are made up
of:

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 97

Figure 8 1 Carbon Monoxide (CO) emissions for different However, the limits for NOx and HC emissions are
flow conditions exceeded when the flows deteriorate to approximately
50% of free flow conditions for NOx and 35% for
CO Emissions HC respectively. In addition, the NOx emissions
1.4 approximately doubles when congestion becomes very
CO
1.2 severe. For HC emissions this approximately triples. In
1.0
order to ensure that the excessive emissions do not occur,

ppm
UK NAQS 0.8
Objective
0.6
it is essential that good maintenance procedures be
0.4 followed and all reasonable steps are taken to disseminate
0.2 information (e.g. early warnings of intentions to carry
0.0 out maintenance work, use of appropriate diversion signs
100 80 60 40 20 0
in the event of accidents) and manage traffic (e.g. good
Flow conditions (% of free flow)
signage, well planned maintenance work).

Figure 8 2 Nitrogen Oxides (NOx) & Hydrocarbon emissions 9. DECISION-MAKING &

for different conditions PRIORITIZATION

NOx & HC Emissions 9.1 Decision-making

250

200 9.1.1 Introduction

150
ppb

Upon the successful collection of bridge scoring data,

100
NOx
modelling bridge conditions with time, estimating traffic
HC
UK NAQS Objective
50
delay and accident costs and environmental impacts, it
0 is important to develop methods to aid decision-making
100 80 60 40 20 0
to maximize the returns for investments into the various
Flow condition (% of free flow)
elements of bridge management. As such, understanding
the breakdown of maintenance expenditure is necessary.
The amount of CO emissions approximately triples A brief summary of typical costs of maintenance
when congestion becomes very severe, but does not work carried out on bridges is as follows:
indicate a likelihood of exceeding the NAQS limits.

Figure 9 1 Breakdown of maintenance costs for various countries (RIMES, 1999)

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

98 School of Doctoral Studies (European Union) Journal 2010

For the UK, the costs per bridge for the various of not implementing the optimal strategy in terms of
maintenance activities are as follows: economic cost. The criteria for prioritization are as
follows:
Table 9 1 Breakdown of costs for various maintenance
activities (RIMES, 1999) Figure 9 2 Criteria for prioritization for various countries
(Woodward R J et al, 1999)
Activity Cost, £
Inspection 1579
Routine maintenance 1579
Specialized maintenance 10 263
Repair & strengthening 9473
Mean maintenance cost per bridge 23 700

It is obvious that not all maintenance actions are

required for all bridges, since older bridges need more
repairs and strengthening and new bridges do not require
them as much. However, the mean cost of maintenance
per bridge is useful as a reference for allocating costs
and alternatives. Comparing how decisions are made
by other BMS, the BMS used in Denmark involves a
prioritization programme (EC BRITE and EURAM,
1993), while Finland uses a repair index. Canada, which
uses PONTIS, uses long term cost optimal strategies.
France and Germany both rely on the judgment of
engineers (Mahut B, 1998, Der Bundesminister fur
Verkehr, 1992). Spain’s BMS evaluates the cost of
repair as a proportion of replacement costs (Das P C,
1996). For the UK, cost benefit analysis is carried out 9.2.2 Dynamic Programming
for decision-making.
Dynamic programming can be used in bridge
9.2 Prioritization management to prioritize how resources are allocated
to maintenance activities when given a set of possible
9.2.1 Introduction alternatives in order to maximize the returns. This is a
useful tool in determining long term strategies for bridge
In addition to decision-making, prioritization is an or element replacement and can take on board imposed
important element in bridge management. Denmark, restrictions in order to fulfill the requirements of bridge
Spain, Finland, USA (New York) and Canada uses management.
optimal strategies within their BMS that imposes a The definitions of terms used are as follows:
minimum level of performance that is required. The
influence of both cost and transport related policies are 1. Programming progresses in stages, which can be
taken into account for the UK. Canada uses the lowest taken to mean the same as ‘steps’.
costs in the long term that avoids bridge failure. However, 2. For each stage, the system is defined by the state it
France, Germany, Norway, Slovenia and the UK do not is in.
use a prioritization technique when there is insufficient 3. At each stage, a decision is made (e.g. amount of
budget (Der Bundesminister fur Verkehr, 1982). This investment to be put in).
is practiced in Spain, Denmark, USA (New York) and 4. The state return function gives the results
Canada (Lauridsen J and Lassen B, 1998). Denmark and (consequential value) of the decision taken.
USA’s(New York) BMS’s both evaluate the consequence 5. The optimal value of the system is then computed

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 99

sequentially by stages, using a forward or backward Decision: xk

pass. This then gives the optimal allocation of
resources. f k * ( sk ) = Max{g k ( sk , xk ) + f k +1 * ( sk +1 )}
xk
For a backward pass:
sk+1 = sk – xk
Backward pass

Where,
Stage k Stage k+1

fk*(sk) is optimal for current stage k and fk+1*{sk+1) is the

optimal for next stage k+1.
Let the stage be called k
In the table below, four maintenance activities
Then sk represents the amount of investment available at are assumed. They represent varying degrees of
that time maintenance, from less intensive (e.g. routine inspection)
costing £25 000, to very intensive (£100 000). They can
The notation for decision to be taken is xk be either a single type of maintenance activity, such as
road resurfacing, or a combination of activities, such as
The function is painting and bridge bearings replacement. It is proposed
that four types of bridges in the bridge stock be used,
similar to the four types of conditions defined by the BCI
fk*(sk) = Max {gk(sk,xk) + fk+1*{1)} values. The benefits reaped from maintenance activities
are assumed to increase with increasing investment levels
Where, and differ between the four different bridge conditions.
This is due to the fact that bridges in the ‘very good’
fk*(sk) is the optimal for the current stage k (denoted condition will not yield a significant amount of benefits
using the symbol ‘*’) due to their present favourable state.
Comparing with a ‘poor’ or ‘fair’ bridge receiving
and fk+1*{sk+1) is the optimal for the ‘previous’ stage k+1 substantial maintenance, the potential for greater benefits
to be reaped is higher. Not all maintenance activities
Also, have the same effect for bridges and it is estimated in
proportion to the investment levels used. It should be
sk+1 = sk – xk noted that the benefits are larger than each investment
alternative as it is an assumed imposed condition that
9.3 Budget Allocation Approach the benefits should outweigh the cost, otherwise the
activity would not be beneficial at all and should not be
In a bridge stock, the benefits derived for different considered altogether. For the purposes of this exercise,
bridges for the same maintenance activity vary, due to the sum of £100 000 budget is estimated to be available
the current condition that bridges are in, as well as other for investing and the tables of cost and benefits is as
benefits to road users and the community as a whole. follows:
For a simple model, a dynamic linear programme can
be used to optimize the returns when different levels of Table 9 2 Costs & benefits from budget allocation example
investment are put in.
To define, for a backward pass method:
Maintenance Activity 1 2 3 4
Investment, in £ 000 25 50 75 100
Stage: k Poor
Fair
70
45
85
75
90
85
110
130
Good 40 65 85 120
Very good 40 70 80 100
State: sk, represents the amount of investment available
for the current and next bridge

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

100 School of Doctoral Studies (European Union) Journal 2010

Table 9 3 Budge allocation computation for example The results obtained are as follows:

Stage State Decision Benefit

Table 9 5 Budget allocation computation for 2nd example
4 25 25 40 *
50 50 70 *
75 75 80 * Stage State Decision Benefit
100 100 100 * 4 25 25 30 *

3 25 0 40 * 50 50 70 *

25 40 * 75 75 80 *

50 0 70 100 100 110 *

25 40+40= 80 * 3 25 0 30

50 65 25 35 *

75 0 80 50 0 70

25 40+70= 110 * 25 35+30= 65

50 65+40= 105 50 75 *

75 85 75 0 80

100 0 100 25 35+70= 105 *

25 40+80= 120 50 75+30= 105 *

50 65+70= 135 * 75 85

75 85+40= 125 100 0 110

100 120 25 35+80= 115

2 25 0 40 50 75+70= 145 *

25 45 * 75 85+30= 115

50 0 80 100 120

25 45+40= 85 * 2 25 0 35 *
25 35 *
50 75
50 0 75 *
75 0 110
25 35+35= 70
25 45+80= 125 *
50 75 *
50 75+40= 115
75 0 105
75 85
25 35+75= 110 *
100 0 135
50 75+35= 110 *
25 45+110= 155 *
75 90
50 75+80= 155 *
100 0 145
75 85+40= 125
25 35+105= 140
100 130
50 75+75= 150 *
1 100 0 155
75 90+35= 125
25 70+125= 195 *
100 130
50 85+85= 170
1 100 0 150
75 90+45= 135
25 50+110= 160 *
100 110
50 85+75= 160 *

Note: The ‘*’ represents optimal value for the stage and state. 75 100+35= 135
100 150

9.3.1 Budget Allocation Results (1st Example) 9.3.2 Budget Allocation Results( 2nd Example)

The optimal allocation is £25 000, £25 000, £25 The optimal allocations are: £50 000, 0, £50 000, 0;
000, £25 000 for each bridge. This is essentially equal or £50 000, £50 000, £0, £0; or £25 000, £25 000, £50
distribution as the benefits gained by the bridges in 000, 0; or £25 000, £50 000, £25 000, £0. There is a
‘poor’ condition are not significant enough to warrant larger number of potential combinations for this model
more investment. and more allocations are given to the bridges in the
Using a different assumed case, benefits gained from ‘poor’ condition because of the higher benefits reap from
bridges in the ‘poor’ condition are higher: investment when compared to investing on bridges that
are in ‘good’ condition.
Table 9 4 Costs & benefits for 2nd example
A third assumed case uses even higher benefits gained
Maintenance Activity
Investment, in £ 000
1
25
2
50
3
75
4
100
from investing in bridges that are in poor condition:
Poor 50 85 100 150
Fair 35 75 90 130
Good 35 75 85 120
Very good 30 70 80 110

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 101

Table 9 6 Costs & benefits for 3rd example 10. IMPROVEMENTS TO BUDGET
ALLOCATION APPROACH
Maintenance Activity 1 2 3 4
Investment, in £ 000 25 50 75 100
Poor 50 90 120 160 10.1 Introduction to BCI optimization
Fair 35 65 85 120
Good 35 60 85 115 approach
Very good 30 55 80 110

Although budgets can be allocated optimally, as

Table 9 7 Budge allocation computation for 3rd example demonstrated earlier, it does not necessarily translate
to optimal bridge conditions since the benefits derived
Stage State Decision Benefit for the different bridge conditions have been aggregated
4 25 25 30 *
50 50 55 * into a single value for optimization. The benefits derived
75 75 80 *
100 100 110 * can be made up of traffic time savings, aesthetics and
3 25 0 30
25 35 * environment benefits, which may not take into account
50 0 55
25 35+30= 65 * the condition of the bridge itself. As such the previous
50 60
75 0 80 approach involves budgets allocated to obtain the
25
50
35+55=
60+30=
90
90
*
* optimal aggregated benefits. In this suggested approach,
100
75
0
85
110 a specific optimization target has been selected, namely
25
50
35+80=
60+55=
115
115
*
*
the Bridge Condition Index (BCI). This approach seeks
75
100
85+30= 115
115
*
to allocate budgets to obtain the highest possible BCI.
2 25 0
25
35
35
*
*
The dynamic programme is defined as follows:
50 0 65
25 35+35= 70 *
50 65 3
75 0
25
50
35+65=
65+35=
90
100
100
*
*
Maximize: z = ∏p
i =1
i
75 85
100 0 115
25
50
75
35+90=
65+65=
85+35=
125
130
120
*
Subject to ∑c i < Budget
100 120
1 100 0 130
25 50+100= 150
50 90+70= 160 * The stage is the maintenance activity number; the
75 120+35= 155
100 160 * state is the funds available to carry out the maintenance
actions; the decision is the type of maintenance actions
carried out. For a backward pass, the recurrence equation
9.3.3 Budget Allocation Results ( 3rd Example) is:

This results in obtaining the optimal investment

allocation of £50 000, £25 000, £25 000, £0; or £100 f i * ( si ) = Max{ pi (n) × f i +1 * ( si +1 )}
n =1, 2 , 3
000, £0, £0, £0. For this third case, more investment is
allocated for the bridges in ‘poor’ condition. Although
allocating resources to bridges in the ‘poor’ condition Where
optimizes the benefits to be derived, the developed
model to predict future bridge conditions can be used to fi *(si) is optimal for current stage i and fi+1*{si+1) is the
evaluate the consequence of not allocating resources to optimal for next stage i+1 And the transition equation is
the other bridges.
si +1 = si − g i (n)

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

102 School of Doctoral Studies (European Union) Journal 2010

Where Table 10 3 Computation for service potential example

gi is the cost of the maintenance actions and pi is the Stage State Decision Total cost Available service potential

available service potential (BCI).

3 S3>50 3 50 0.85 *

30<S3<50 2 30 0.75 *

25<S3<30 1 25 0.6 *

2 S3>90 3 40 + 50 = 90 0.9*0.85 = 0.765 *

10.2 Service Potential (BCI) Examples 85 3 40 + 30 = 70 0.9*0.75 = 0.675

2 35 + 50 = 85 0.8*0.85 = 0.68 *

1 30 + 50 = 80 0.7*0.85 = 0.595

A simple model is set up to investigate how budget 80 3 40 + 30 = 70 0.9*0.75 = 0.675 *

constraints influence the available service potential as

2 35 + 30 = 65 0.8*0.75 = 0.6

1 30 + 50 = 80 0.7*0.85 = 0.595

well as the choice of maintenance actions to be taken. 70 3 40 + 30 = 70 0.9*0.75 = 0.675 *

2 35 + 30 = 65 0.8*0.75 = 0.6
Three bridges are assumed in this model with service 1 30 + 30 = 60 0.7*0.75 = 0.525

potential increasing with the number of maintenance 65 3

2
40 + 25 = 65

35 + 30 = 65
0.9*0.6 =

0.8*0.75 =
0.54

0.6 *

actions taken. For simplification, three types of actions 1 30 + 30 = 60 0.7*0.75 = 0.525

can be carried out and each bridge must undergo any one
60 2 35 + 25 = 60 0.8*0.6 = 0.48

1 30 + 30 = 60 0.7*0.75 = 0.525 *

of them. The costs are assumed for each bridge based on 55 1 30 + 25 = 55 0.7*0.6 = 0.42 *

1 120 3 50 + 70 = 120 0.95*0.675 = 0.64125 *

their conditions and the three bridges used are defined to 2 45 + 70 = 115 0.9*0.675 = 0.6075

be declining in performance (i.e. bridge 3 is the oldest 1 40 + 70 = 110 0.85*0.675 = 0.57375

and bridge 1 to be the newest). Hence higher values

are attributed to the newer bridge. For simplification, 10.2.1 BCI Optimization Results (1st Example)
the BCI are divided by a hundred to convert them to
probability first. They are then converted back to BCI. The optimal BCI of the system is 64.125, with the
decisions of 3, 3, 2. All £120 000 is spent. It is expected
Table 10 1 Inputs for service potential example that budget has a strong influence on decisions made for
maintenance actions, therefore further steps are taken to
Maintenance Actions
1
Bridge 1
0.85
Bridge 2
0.7
Bridge 3
0.6
investigate this.
2
3
0.9
0.95
0.8
0.9
0.75
0.85
Two additional scenarios are assumed, one for having
an increased budget by 10% to £132 000 and another
having a reduced budget by 10% to £108 000. Their
The costs are also put in tabular form: respective results in BCI values and decisions are
computed below:
Table 10 2 Costs for service potential example
Table 10 4 Computation for service potential example
Costs, £ in thousands for each
maintenance action
Bridge 1 Bridge 2 Bridge 3 (increased budget)
1 40 30 25 Stage State Decision Total cost Available service potential
2 45 35 30 3 S3>50 3 50 0.85 *
3 50 40 50
30<S3<50 2 30 0.75 *

25<S3<30 1 25 0.6 *

A simple linear programming model is created,

2 S3>90 3 40 + 50 = 90 0.9*0.85 = 0.765 *

85 3 40 + 30 = 70 0.9*0.75 = 0.675

working with an assumed budget of £120 000. 2

1
35 + 50 = 85

30 + 50 = 80
0.8*0.85 =

0.7*0.85 =
0.68

0.595
*

The programme is as follows: 80 3 40 + 30 = 70 0.9*0.75 = 0.675 *

2 35 + 30 = 65 0.8*0.75 = 0.6

1 30 + 50 = 80 0.7*0.85 = 0.595

70 3 40 + 30 = 70 0.9*0.75 = 0.675 *

2 35 + 30 = 65 0.8*0.75 = 0.6

1 30 + 30 = 60 0.7*0.75 = 0.525

65 3 40 + 25 = 65 0.9*0.6 = 0.54

2 35 + 30 = 65 0.8*0.75 = 0.6 *

1 30 + 30 = 60 0.7*0.75 = 0.525

60 2 35 + 25 = 60 0.8*0.6 = 0.48

1 30 + 30 = 60 0.7*0.75 = 0.525 *

55 1 30 + 25 = 55 0.7*0.6 = 0.42 *

1 132 3 50 + 70 = 120 0.95*0.675 = 0.64125

2 45 + 85 = 130 0.9*0.68 = 0.612

1 40 + 90 = 130 0.85*0.765 = 0.65025 *

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 103

10.2.2 BCI Optimization Results (2nd Example) By comparing the outcomes in terms of BCI for the
three different budgets, it is observed that when the
For a budget of £132 000, the optimal decision is 1, budget is increased or decreased by about 10.0% (£132
3, 3 with optimal BCI of 65.025. There is a remainder of 000 or £108 000), the changes in BCI by proportion
£2 000 left over. of the original are +1.40% and –20.47% respectively.
This non-proportional behaviour reflects the increased
Table 10 5 Computation for service potential example amount of budget required to increase the BCI and a
(decreased budget) sharp decline in BCI if budgets were to be reduced.

Stage State Decision Total cost Available service potential 10.3 Maintenance Costs Examples
3 S3>50 3 50 0.85 *

30<S3<50 2 30 0.75 *

2
25<S3<30

S3>90
1

3
25

40 + 50 = 90 0.9*0.85 =
0.6

0.765
*

*
Applying weightings to each bridge by means of
85 3 40 + 30 = 70 0.9*0.75 = 0.675 using assumed maintenance costs of £50 per square
meter for the deck areas for the three bridges used in the
2 35 + 50 = 85 0.8*0.85 = 0.68 *

1 30 + 50 = 80 0.7*0.85 = 0.595

80 3 40 + 30 = 70 0.9*0.75 = 0.675 *
earlier model for traffic costs calculations, the cost of full
2 35 + 30 = 65 0.8*0.75 = 0.6

1 30 + 50 = 80 0.7*0.85 = 0.595 maintenance (option 3) can be obtained. They are then

70 3

2
40 + 30 = 70

35 + 30 = 65
0.9*0.75 =

0.8*0.75 =
0.675

0.6
*
scaled down to account for the two other maintenance
1 30 + 30 = 60 0.7*0.75 = 0.525 options. The inputs are as follows:
65 3 40 + 25 = 65 0.9*0.6 = 0.54

2 35 + 30 = 65 0.8*0.75 = 0.6 *

1 30 + 30 = 60 0.7*0.75 = 0.525
Table 10 7 Costs & bridge deck areas for new service
60 2 35 + 25 = 60 0.8*0.6 = 0.48

1 30 + 30 = 60 0.7*0.75 = 0.525 * potential example

55 1 30 + 25 = 55 0.7*0.6 = 0.42 *

1 108 3 50 + 55 = 105 0.95*0.42 0.399

2 45 + 60 = 105 0.9*0.48 0.432 Deck area, m2 Cost, £ in thousands

1 40 + 65 = 105 0.85*0.6 0.51 *
Bridge 1 1200 60
Bridge 2 700 35
Bridge 3 1000 50

10.2.3 BCI Optimization Results (3rd Example)

Table 10 8 Inputs for new service potential example
Using a reduced budget of £108 000, the optimal
decision is 1, 2, 2 with a BCI of 51. There is a remainder Maintenance Actions
1
Bridge 1
0.92
Bridge 2
0.87
Bridge 3
0.53
of £3 000 left over. 2
3
0.95
0.98
0.88
0.89
0.55
0.57
A summary of results is as follows:

Table 10 6 Results for different budget allocations The costs are also put in tabular form:

Case
1
Budget
£120 000
% change in budget
-
BCI
64.1
% change of original BCI
-
Table 10 9 Costs for new service potential example
2 £132 000 + 10.0 65.0 + 1.40
3 £108 000 - 10.0 51.0 - 20.47
Costs, £ in thousands for each
Bridge 1 Bridge 2 Bridge 3
maintenance action
Figure 10 1 Changes in BCI with investment 1 40 30 30
2 50 32 40
3 60 35 50

Changes in BCI with investment

70
A simple linear programming model is created,
65 working with an assumed budget of £120 000. The
upper and lower limits of the available BCI that has been
BCI

60
converted to probabilities are obtained from the BCI
55
(average) model and the upper limit of maintenance costs
50 have been obtained above. The mid-values are proposed
100 105 110 115
Investment, £000
120 125 130 135
to be approximately the median value for the probabilities
and for the costs the other values are estimated. Both

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

104 School of Doctoral Studies (European Union) Journal 2010

costs and probabilities decrease in bridge maintenance Investigating a decrease in budget available, a
actions, with 1 being the most basic maintenance activity 10% decrease (i.e. £108 000) was introduced with the
and 3 being the most extensive activity. following results:
The outputs are obtained:
Table 10 12 Computation for 6th service potential
Table 10 10 Computation for 4 service potential example
th
example (decreased budget)

Stage State Decision Total cost Available service potential

3 S3>50 3 50 0.53 *
Stage State Decision Total cost Available service potential
40<S3<50 2 40 0.55 *
3 S3>50 3 50 0.53 *
30<S3<40 1 30 0.57 *
40<S3<50 2 40 0.55 *
2 S3>85 3 35+50=85 0.89*0.57= 0.5073 *
30<S3<40 1 30 0.57 *
75 3 35+40=75 0.89*0.55= 0.4895 *
2 S3>85 3 35+50=85 0.89*0.57= 0.5073
2 32+40=72 0.88*0.55= 0.484
75 3 35+40=75 0.89*0.55= 0.4895 *
1 30+40=70 0.87*0.55= 0.4785
2 32+40=72 0.88*0.55= 0.484
65 3 35+30=65 0.89*0.53= 0.4717 *
1 30+40=70 0.87*0.55= 0.4785
2 32+30=62 0.4664
65 3 35+30=65 0.89*0.53= 0.4717 *
1 30+30=60 0.4611
2 32+30=62 0.4664
1 30+30=60 0.4611
62 2 32+30=62 0.4664 *

62 2 32+30=62 0.4664 * 1 30+30=60 0.4611

1 30+30=60 0.4611 60 1 60 0.4611 *

60 1 60 0.4611 * 1 108 3 Not possible

1 120 3 60+60=120 0.452 * 2 Not possible

2 50+65=115 0.448 1 40+65=105 0.92*0.4717= 0.434 *

1 40+75=115 0.45

10.3.1 BCI Optimization Results (4th Example) 10.3.3 BCI Optimization Results (6th Example)

The optimal decision is 3, 1, 3 with costs of £120 000 The optimal decision is 1, 3, 1 with costs of £105 000.
(no remainders left over). The optimal BCI is 45.2. The optimal BCI is 43.4. There is a remainder of £3 000.
Increasing the budget by 10% to £132 000, the A summary of results is as follows:
following results were obtained:

Table 10 11 Computation for 5th service potential example Table 10 13 Results for different budget allocations
(increased budget) (Examples 4 to 6)

Stage State Decision Total cost Available service potential Case Budget % change in budget BCI % change of original BCI
1 £120 000 - 45.2 -
3 S3>50 3 50 0.53 *
2 £132 000 + 10.0 46.7 + 3.3
40<S3<50 2 40 0.55 * 3 £108 000 - 10.0 43.4 - 4.0
30<S3<40 1 30 0.57 *
2 S3>85 3 35+50=85 0.89*0.57= 0.5073 *
75 3 35+40=75 0.89*0.55= 0.4895 *
2 32+40=72 0.88*0.55= 0.484 Figure 10 2 Changes in BCI with investment (using
1 30+40=70 0.87*0.55= 0.4785
maintenance costs)
65 3 35+30=65 0.89*0.53= 0.4717 *
2 32+30=62 0.4664
1 30+30=60 0.4611 Changes in BCI with investment
62 2 32+30=62 0.4664 * 47
1 30+30=60 0.4611
60 1 60 0.4611 *
46
1 132 3 60+65=125 0.98*0.4717= 0.462
2 50+75=125 0.95*0.4895= 0.465
BCI

45
1 40+85=125 0.92*0.5073= 0.467 *

44
10.3.2 BCI Optimization Results (5th Example)
43

The optimal decision is 1, 3, 3 with costs of £125 000. 100 105 110 115 120 125 130
Investment, £000
The optimal BCI is 46.7. There is a remainder of £7 000.

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 105

For the reduced budget scenario, constraints to the After scoring systems have been developed to assess
expenditure have been imposed due to lack of resources. the condition of bridges and extending them to the bridge
It is observed that the percentage change in BCI available stock, Markov chain models have been used to model
is greater when funds are reduced than if they were to be changes in condition with time at both the individual
increased. The first two cases have optimal strategies bridge level as well as bridge stock level. Using different
that involve utilizing maintenance option 3 for bridge 3 traffic conditions, delay cost models have been developed
which represents sound judgment as investment should to derive estimates of both delay costs as well as accident
be put into the most severe BCI contributor. However, costs. An emission model has also been created to
with a reduced budget, only maintenance activity 1 can estimate the increased impacts on the environment due
be selected. to reductions in speed.
In real terms, only an additional £5 000 is necessary Finally, this project utilizes dynamic programming
to increase the BCI by 3.3% since there has been a saving to accomplish optimal budget allocation and optimizing
of £7 000 for the £132 000 budget case. For a reduced bridge conditions given a specific budget allowance.
budget, £105 000 has been spent (remainder of £3 000). It These two approaches have different aims and as such,
is possible that the £3 000 or £7 000, whichever scenario their procedures yield different results. The use of such
applies, be channelled towards the bridge that has the approaches is expected be proven as useful tools to both
low BCI value to further improve the overall system. the engineer as well as transport policy makers.
Comparing percentage change for actual money It has been demonstrated that the limiting
spent with the base value of £120 000 scenario, for £5 bridge conditions may be reached prematurely due
000 more (4.2%), the BCI increases by 3.3%. For an severe deterioration of a key bridge element. This
expenditure of £105 000 (12.5% less), the BCI decreases has implications for emergency unplanned bridge
by 4%. In this manner, the costs to increase the BCI in maintenance, which could potentially disrupt traffic and
terms of value are higher (4.2% budget increase for 3.3% consequentially increase road user costs significantly.
BCI increase) and for a significant reduction of budget, Much attention is required for the input values for
the BCI falls by a smaller proportion. This indicates that modelling bridge conditions over time and diligent
a reduced budget does not severely affect the BCI value, monitoring and adhering to inspection schedules that
but a low value of 43.4 can be considered as ‘poor’. have been set out can achieve this.
The poor ratings of the 3rd bridge have dominated the The costs borne by road users are dependent many
final output value. However, as mentioned before, the variables and more extensive studies can be made to
‘savings’ can be re-channelled back to areas that need establish upper and lower bounds of the derived estimates.
investment and the remainder of £3 000 is a third of Assigning ‘costs’ to the increased emissions was
the reduced expenditure (£120 000 reduced to £105 anticipated to be both difficult and subjective owing to the
000), which is significant. This translates to a potential assumptions required for their assignment. Therefore it
increase of the eventual optimal BCI. has not been attempted, however this could be a potential
area for more extensive research. The examples of
11. CONCLUSION optimizing budget allocation and bridge condition can
be extended over to a bridge stock, or several bridge
This project seeks to provide the reader an stocks and this requires the specific decisions taken by
understanding into the key bridge management concepts the bridge engineer. From the examples in the project,
and principles practiced in the UK. For a broader it has been shown that the available service potential of
perspective, bridge management systems elsewhere in bridges can potentially fall disproportionately due to a
the world, such as countries in continental Europe and decrease in budget.
the USA have been discussed as well. Learning the latest With the use of the developed models to estimate
methods practiced in the UK for inspecting and assessing various road user costs, optimize bridge conditions,
bridges, reviewing maintenance needs and prioritization, optimize budget allocation, the systems approach
several different models have been developed to assist in to bridge management discussed in this project is
good bridge management. expected to benefit both technical and non-technical

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

106 School of Doctoral Studies (European Union) Journal 2010

users. The tools can aid higher-level strategic REFERENCES

decision-making individuals who seek to understand
the different interacting mechanisms of traffic flows, Andersen, N H (1990) DANBRO – A Bridge
costs and different policies, as well as potentially use Management System for many levels. In: Bridge
the modelling techniques to champion their proposals Evaluation, Repair and Rehabilitation, edited by
for improving the current local environment or region. Andrzej Nowak, Kluwer Acadamic
Simple computational techniques can aid the engineer
to establish a sensitivity to the environment impacts of ATKINS (2002) CSS Bridge Condition Indicators
decisions made in bridge management, as well as be Volume 3: Evaluation of Condition Indicators
more capable of tackling social and political issues that
can involve planning policies and regeneration projects. ATKINS (2002) CSS Bridge Condition Indicators
Volume 2: Bridge Inspection Reporting
12. FUTURE RESEARCH
Baker, D (1999) The maintenace of highway
To take this project further, there are several possible structures. In: Management of highway
research topics that can be considered. Structural structures, ICE, Thomas Telford
assessments and reliability models can be developed
using established methods and these can be integrated Bourn, John (2003) Maintaining England’s
into the prioritization techniques or used together with Motorways and Trunk Roads. National Audit
developed models to better improve bridge condition Office
modelling with time. The assumptions on the available
budget for computation examples have been based on a Bourn, John (2001) The National Road Maintenance
limited number of factors. As such, more detailed study Condition Survey. National Audit Office
into typical budget allocations for each county, council or
local authority could be undertaken in order to investigate Bourn, John (1996) Highways Agency: The Bridge
the different results derived from ‘local’, optimized Programme. National Audit Office
solutions and nationwide (i.e. ‘global’), optimized
solutions. Additional prioritization techniques can be Bourn, John (1995) Expenditure on Motorways and
considered, such as using event tree analysis, probability Trunk roads. National Audit Office
and statistical theories to deal with other uncertainties
as well as regression approaches. The composition of Bourn, John (1991) Management of Road
the bridge stocks can also be modelled using appropriate Maintenance. National Audit Office
distributions and sufficiently sized sample sizes.
With regards to the environmental impacts of bridge Bourn, John (1989) Backlog of Maintenance of
maintenance, noise level modelling can be carried out to Motorways and Trunk Roads. National Audit
supplement the emissions models that have already been Office
developed. Further detailed study on the road user choice
behaviour can be taken and integrated with government Bourn, John (1988) Regulation of Heavy Lorries.
policies encouraging the use of public transport (e.g. National Audit Office
trains, buses). It should be noted that good bridge
management not only includes ensuring the bridge Branco, F A and Brito, J (1996) Bridge management
stock is well managed in terms of structural integrity from design to maintenance. In: Recent Advances
and budget management, but the environmental, social in Bridge Engineering, CIMNE Barcelona 1996,
and political issues need to be addressed as they play edited by J. R. Casas, F.W. Klaiber and A. R.
important roles in each engineering process. Marí, 1996

School of Doctoral Studies (European Union) Journal - 2010

2010 Bridge Management A System Approach for Decision Making 107

Calgaro, J A (1997) Maintenance et Reparation des Hawk, H and Small E P (1998) The BRIDGIT
Ponts. Presse de l’ENPC Znidaric A, Moses F Bridge Management System. In: Structural
(1997) Structural safety of existing road bridges. Engineering International, Volume 8, Number 4,
In: 7th ICOSSAR conference, Kyoto, 1997 November 1998

Clausen, P (1992) Introduction to Bridge Lauridsen, Jørn and Lassen, Bjørn (1998) The
Management Systems. In: 14th Congress New Danish Bridge Management System (DANBRO).
Delhi 1992, IABSE In: The Management of Highway Structures,
The Institution of Civil Engineers and Highway
Das, P C (1998) New Developments in Bridge Agency, London, 22-23 June 1998
Management Methodology. In: Structural
Engineering International, Volume 8, Number 4, Mahut, Brigitte (1998) Maintenance of the Bridges
November 1998 on the French national network; Classification of
the condition of structures. Bridge Management
Das, P C (1996) Bridge Management Methodologies. in Europe (BRIME) Workpackage 1, March 1998.
In: Recent Advances in Bridge Engineering, Transport Research Laboratory, Crowthorne
CIMNE Barcelona 1996, edited by J.R. Casas, F.
W. Klaiber and A. R. Marí, 1996 Naumann, J (1998) Development of a Bridge
Management System in Germany. In: Operation
Department of Transport (2003) Economic and Maintenance of Large Infrastructure Projects,
Assessment of Road Schemes. The COBA Balkema, Rotterdam 1998, edited by Vincentsen
Manual Volume 13 Section 1 & Jensen

Der Bundesminister fur Verkehr (1992) Richtlinie OECD Report (1992) Bridge Management
zur Tragfahigkeitseinstufung bestehender
Strassenbrucken der neuen Bundeslander in Road Directorate, Denmark Ministry of Transport
Lastklassen nach DIN 1972. Ausg. Dez. 1985 (1995) Bridge Maintenance on the Danish main
road network
Der Bundesminister fur Verkehr (1982), Shaden
an Brucken and anderer Ingenierurbauwerken, Road Infrastructure Maintenance Evaluation Study
Dokumentation 1982 (1999) Network level Management model
(Pavement and Structure management system)
EC BRITE and EURAM (1993) Assessment for Markov chains. RIMES, Workpackage 3
of Performance and Optimal Strategies for 1999
Inspection and Maintenance of Concrete
Structures Using Reliability Based Expert Söderqvist, M K and Veijola, M (1998) The Finnish
Systems. In: Report P3091, Aalborg, Denmark, Bridge Management System. In: Structural
1993 Engineering International, Volume 8, Number 4,
November 1998
HA (2003) SMIS, A guide for structure inspectors
Stensvold, Børre (1997) Computer Aided Bridge
HA (2003) SMIS, A guide for structure engineers Management in Norway. In: International
Conference Repair of Concrete Structures from
HA (1998) DMRB Volume 11 Section3 Theory to Practice in a MarineEnvironment,
545-553, Svolvær, Norway, May 1997

Stratt R. S. - Bridge ManagementA System Approach for Decision Making

108 School of Doctoral Studies (European Union) Journal 2010

The Department of the Environment, Transport and Woodward R J, Bevc L, Perus I, Mahut B and
the Regions (1998) Getting the best roads for our Grefstad K (1999) Review of existing procedures
money, An introduction to QUADRO for optimisation. Bridge Management in Europe
(BRIME) Deliverable D3. Transport Research
The Highways Agency, BA 63/94 (1994) Inspection Laboratory, Crowthorne
of Highway Structures, Part 5
Woodward, R J, Godart B and Vassie P R (1999)
Thoft-Christensen, P (1996) Bridge Management Review of existing BMS and definition of inputs
Systems. present and future. In: Recent Advances for the proposed BMS. Bridge Management
in Bridge Engineering, CIMNE Barcelona 1996, in Europe (BRIME) Deliverable D4. Transport
edited by J. R. Casas, F. W. Klaiber and A. R. Research Laboratory, Crowthorne
Marí, 1996
Woodward R J, Asudillo Pastor R, Arreieta Torrealba
Thompson, P D, Small, E P, Johnson M and Marshall, J M, Velando Cabanas C and Lozano Bruna C
A R (1998) The Pontis Bridge Management (1999) Decision on repair and replacement.
System. In: Structural Engineering International, Bridge Management in Europe (BRIME)
Volume 8, Number 4, November 1998 Deliverable D7. Transport Research Laboratory,
Crowthorne
Vorshrift 169/89 (1989) Brucken im Verkehrsbau;
Nachrechnung von Strassenbrucken. Beton und Yanev, B S (1998) Bridge Management for New York
Mauerwerk City. In: Structural Engineering International,
Volume 8, Number 3, August 1998.
Woodward, R J, Kaschner R, Cremona C and
Cullington D (1999) Review of current procedures Znidaric J and Moses F (1995) Assessment of safety
for assessing load carrying capacity. Bridge and remaining service life of existing bridges.
Management in Europe (BRIME) Deliverable ZAG Ljubljana
D1. Transport Research Laboratory, Crowthorne

School of Doctoral Studies (European Union) Journal - 2010