1227

Managing bridge infrastructure under budget constraints: a decision support methodology

Saleh Abu Dabous and Sabah Alkass

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by ETH ZUERICH on 06/11/12 For personal use only.

Abstract: Managing the deteriorating bridge infrastructure has become a major social and economic concern in North America. The methodology developed in the present research consists of three methods: (1) ranking and prioritizing method to evaluate and rank all bridge projects in a network, (2) ranking method to evaluate the available rehabilitation strategies for each project, and (3) recommended work program development method. The recommended work program specifies a list of projects to allocate the available budget to and what type of action to be implemented for each project. The developed methodology allows engineers to use multiple and conflicting criteria and to incorporate their experience and judgment in the decision making process. The methodology can assist bridge engineers and decision makers in selecting the most suitable work program that can be performed with a limited budget. An example application is presented to demonstrate the use and capabilities of the developed methodology. Key words: bridge, decision, budget, rehabilitation, multi-criteria, utility, AHP. Rsum : La gestion de la dtrioration de linfrastructure des ponts est devenue un enjeu conomique et social majeur en Amrique du Nord. La mthodologie dveloppe dans la recherche prsente dans cet article comporte trois mthodes: (1) une mthode de classement et de dtermination de lordre des priorits afin dvaluer et de classer tous les projets de ponts dans un rseau, (2) une mthode de classement pour valuer les stratgies de remise en tat disponibles pour chaque projet et (3) une mthode de dveloppement dun programme pour le travail recommand. Ce programme de travail recommand spcifie une liste de projets afin dallouer le budget disponible ainsi que les actions prendre pour chaque projet. La mthodologie dveloppe permet aux ingnieurs dutiliser plusieurs critres incompatibles et dincorporer leur exprience et leur jugement dans le processus de prise de dcisions. Cette mthodologie peut aider les ingnieurs de ponts et les dcideurs choisir le meilleur programme de travail qui peut tre ralis lintrieur dun budget limit. Un exemple dmontre lutilisation et les capacits de la mthodologie dveloppe. Motscls : pont, dcision, budget, remise en tat, multicritres, utilit, MHM. [Traduit par la Rdaction]

Introduction
Reports on the status of civil infrastructure in North America demonstrate that the deteriorating bridge infrastructure requires immediate attention. In the 2009 report card for Americas infrastructure, the American Society of Civil Engineers reported that more than 26% of bridges in the United States are either structurally deficient or functionally obsolete. The society estimated that a $17 billion annual investment is needed to substantially improve current bridge conditions. The current yearly spending on the construction and maintenance of bridges in the United States is $10.5 billion (ASCE 2009). In 2008, AASHTO estimated that it would cost about $48 billion to repair structurally deficient bridges and $91 billion to improve functionally obsolete bridges that are no longer adequate to serve traffic (AASHTO 2008). The Federation of Canadian Municipalities reported that 83% of Canadian bridges need some sort of repair (Mirza

and Haider 2003). Vanier (2000) concluded that managers and decision makers in the United States and Canada are required to maintain the deteriorated bridge infrastructure under limited budget considerations. Thus, these managers need decision support systems to help them to manage the existing deteriorating bridge infrastructure. Many transportation agencies make bridge management decisions based on a combination of analyzing available quantitative data such as the inspection reports and using subjective judgments of the decision and policy makers (Kulkarni et al. 2004). This process aims toward distributing the available budget among bridges that need intervention and defining what type of action to be taken for each bridge. The subjective nature of this decision making process could raise questions about whether the investment decisions are being made in a fair, equitable and systematic manner or if they more often reflect the intuitive judgments of the decision makers. Alternatively, bridge engineers and policy makers

Received 8 February 2011. Revision accepted 27 July 2011. Published at www.nrcresearchpress.com/cjce on 31 October 2011. S. Abu Dabous. American University of Sharjah P.O. Box 26666, UAE. S. Alkass. Department of Building, Civil and Environmental Engineering, Concordia University, 1455 de Maisonneuve, Montreal, QC H3G 1M8, Canada. Corresponding author: Saleh Abu Dabous (e-mail: sabudabous@aus.edu). Written discussion of this article is welcomed and will be received by the Editor until 31 March 2012.
Can. J. Civ. Eng. 38: 12271237 (2011) doi:10.1139/L11-082 Published by NRC Research Press

1228

Can. J. Civ. Eng. Vol. 38, 2011

need rational decision support tools to help them in managing bridge networks under budget constraints. This research develops a decision support methodology to allocate limited budget to the most deserving projects in a network and to recommend the most effective rehabilitation strategy for each project. The methodology integrates multiple criteria decision support methods within the developed framework to generate a recommended work program. The focus in the current research is directed at bridge decks. Further future research is recommended to study other elements.
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by ETH ZUERICH on 06/11/12 For personal use only.

1988; Madanat et al. 1997). In addition, solution techniques are intractable as the discrete set of condition states become large. Thus, it is very difficult to manage large networks of facilities jointly or to consider complex relationships among relevant variables. Benefit-to-cost ratio analysis The benefit-to-cost ratio analysis evaluates all of the benefits and costs associated with a project, including both direct agency cost and indirect user cost using the dollar as the unit of measure. Priority is given to projects that provide more benefits and incur less cost. The direct agency cost can be estimated from the available cost data. The indirect user costs or benefits are difficult to quantify and are usually estimated using certain parameters or simplifying assumptions. Kulkarni et al. (2004) reported that concerns arise when the benefit concept is applied to evaluate a large number of diverse projects at many different locations, as opposed to a small number of projects. Also, an excessive amount of effort is needed to apply the concept to a large number of projects. Life cycle cost analysis Cost is a major factor in the decision making process, especially within tight budgets considerations. The cost concept has evolved over the years into life cycle cost, which implies that the preferred alternative is an alternative that would cost less in the long run. Life cycle cost (LCC) analysis for bridge engineering is defined by the National Cooperative Highway Research Program as a set of economic principles and computational procedures for comparing initial and future costs to arrive at the most economical strategy for ensuring that a bridge will provide the services for which it was intended (Hawk 2003). Kong and Frangopol (2003) developed reliability-based life cycle maintenance cost optimization of deteriorating bridges. Frangopol and Neves (2003) investigated uncertainty effects on the evaluation of condition and safety indices as well as on the LCC of deteriorating bridges under different maintenance strategies using Monte Carlo simulation. Huang et al. (2004) developed a project-level decision support tool to rank maintenance scenarios for deteriorated concrete bridge decks based on probabilistic LCC analysis. Applying the optimized life cycle cost methodology may create practical difficulties especially when the available budget is larger or lower than the computed minimum life cycle cost. If the available budget is larger than the computed minimum life cycle cost, the bridge performance can be improved to a higher level than the level achieved by the minimum life cycle cost solution. On the other hand, if the available budget is less than the computed minimum life cycle cost, an alternative solution is needed since the minimum life cycle cost solution cannot be implemented (Frangopol and Liu 2007).

Literature review
The main objective of a bridge management decision support system is to select the actions necessary to maintain bridge networks within acceptable limits of safety and serviceability. A review of existing literature on bridge management decision making is completed as a part of the research. The review divides the relevant available literature into three categories: (1) Markov decision process to find optimal policies describing what maintenance actions to take; (2) benefit cost analysis to compare monetary values for benefits and costs associated with individual projects or strategies; and (3) life cycle cost analysis procedures to design alternatives with full recognition of the costs and benefits associated with each alternative life cycle. Markov decision process (MDP) Past research in infrastructure management utilized the Markov decision process for decision making (Gopal and Majidzadeh 1991; Madanat and Ben-Akiva 1994; Abraham and Wirahadikusumah 1999). The MDP is an extension of Markov chains. The MDP utilizes the state dependence assumption in Markov models and adds actions that can lead to improvements under specific certainties. The state dependence means that the probability of improvement is independent of the history. When the process is in condition i and an action A is taken, the process improves into condition j with probability Pij. 1 Pij A PXn1 jjXn i

Optimization of maintenance policies using the Markov decision process can be performed by applying recursively the following equation (Frangopol et al. 2004): 2
N X Va i Ci; A a Pij AVa j j1

where a is the discount factor for one year, estimated by a = (1 + r/100) 1; r is the yearly discount rate; Va is the value function using a; and C(i, A) is the costs that are incurred when the process is in condition i and action A is taken. Then a cost-optimal decision can be found by minimizing the previous cost equation with respect to the action under consideration through using one of the mathematical programming techniques such as linear programming or one of its variations. The main shortcoming of the MDP is mainly due to the state dependence assumption that overlooks the history of deterioration process. Empirical research has confirmed that age is a significant factor in the deterioration process (Jiang et al.

Decision support methodology

Interviews with bridge experts from two private corporations in Montreal and with bridge engineers from the Ministry of Transportation of Ontario concluded that decision support systems are needed to assist in improving performance of the bridge network and in distributing the limited
Published by NRC Research Press

Abu Dabous and Alkass

1229 Fig. 1. Flow chart for the decision support methodology.

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by ETH ZUERICH on 06/11/12 For personal use only.

budget effectively. One of the most critical challenges in the decision making process is the evaluation of data and information especially when both subjective and objective data is incorporated in the process. The proposed decision support methodology in this paper integrates the Analytic Hierarchy Process (AHP) and the Multi-Attribute Utility Theory (MAUT) to extract experts knowledge and judgments while incorporating quantitative and qualitative criteria in the decision making process. The AHP is one of the most versatile decision making tools due to its simplicity and ability to cope with complex decision making problems. The MAUT enables decision makers to include multiple and conflicting criteria. In addition, the theory provides flexibility for the decision makers in expressing their degree of satisfaction with decision attributes and captures the decision makers attitude toward risk. The flow chart in Fig. 1 depicts the decision support methodology. The following sections present in detail the different elements of the developed decision support methodology.

Ranking and prioritizing projects

At the network level, a large number of projects (can be up to 5000 bridges) need to be ranked and prioritized while incorporating experts inputs. The MAUT is utilized to perform this task. The basic principle of the MAUT is based on estimating performance using attributes that are concrete, measurable, and representative of the degree of satisfaction with the various aspects of each alternative. The foundation of MAUT is the use of utility functions, which are utilized to quantify the preference of the decision maker by depicting the degree of satisfaction, as the attribute under consideration takes values between the most and least desirable limits. Figure 2 presents a flow chart for the ranking method based on the MAUT. The method uses a default hierarchy structure and a defined set of criteria. At the same time the method allows the decision maker to modify these elements. The default hierarchy is four levels. The first level of the hierarchy is the overall goal of the ranking exercise that is the effective allocation of available funds. The second level contains the three objectives necessary to achieve the overall goal. The third level of the hierarchy holds the criteria to be used for evaluating the objectives. These criteria are condition rating, live load carrying capacity and seismic risk to measure the objective of maximizing bridge preservation and safety, average daily traffic and supporting road type to measure the objective of maximizing effectiveness of investment, and vertical clearance, approach and drainage system conditions to measure the objective of minimizing bridges deficiencies. The alternatives are added at the bottom level. Each objective or criterion has a specific weight reflecting its importance. The weights of the different criteria and objectives are developed using the Eigenvector approach. Abu Dabous (2008) discusses this hierarchy structure development in detail. Figure 3 presents the utility functions developed to be in4

corporated within the proposed decision support methodology. These functions are developed based on data and judgments extracted during the previously mentioned interviews with bridge engineers and managers. A utility model can be used to aggregate the utility values for the various attributes. Since the elements in each level of the hierarchy structure are considered to be independent, the additive utility model can be used as a simple and practical approach to aggregate utilities. In such a model, the overall expected utility is expressed as follows (Keeney and Raiffa 1993): 3 Ux
n X i1

ki ui x

where ki is the weight for attribute i, ui is utility value for attribute i, and x1 to xn are the available alternatives. The utility scores obtained from the utility functions are aggregated using eq. [3] to estimate the utility associated with each objective. Then, the utilities of the various objectives are aggregated using the same equation to evaluate the overall utility of the bridge. All bridges in the network or sub-network can be ranked based on the overall utility values. For example, the attribute values for bridge number 10 are given in Table 1. The expected utility value for this bridge is as follows:

U10 78:66 0:4 28:89 0:45 28:89 0:15 0:60 33:33 0:55 11:11 0:45 0:20 24:44 0:4 77:77 0:35 62:95 0:25 0:20 44:49

Published by NRC Research Press

1230 Fig. 2. Flow chart for the project ranking method.

Can. J. Civ. Eng. Vol. 38, 2011

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by ETH ZUERICH on 06/11/12 For personal use only.

The overall expected utility for each bridge can be estimated and used to rank all bridges in the network.

Ranking rehabilitation strategies

At the project level, a number of alternative maintenance and rehabilitation strategies are available for each project. Decision makers are required to evaluate these strategies based on multiple criteria. The AHP is a multi-criteria decision process that utilizes both actual measurements and expert judgment. Using the AHP for this task can enable providing preferences between the different maintenance and rehabilitation strategies that are specific to each project. Two fundamental steps are required to use the AHP methodology. First, a complex system is broken down into a hier-

archic structure to represent the problem. Second, pairwise comparisons are performed to measure the relative impact of different elements in the hierarchy and to establish relations within the structure. The pairwise comparisons are performed using a fundamental scale of absolute values that represents the strength of judgments (Saaty 1980). The pairwise comparisons lead to dominance matrices from which ratio scales are derived in the form of principal eigenvectors. To demonstrate the applicability of the AHP, a case study is discussed to rank rehabilitation strategies for a deteriorating bridge deck. The assessment and pairwise comparisons in this case study are performed by an expert from the industry who is involved in bridge management decision making. The available strategies are major rehabilitation, minor rehabilitation or increased routine maintenance. These three alternatives should be evaluated using the following criteria:
Published by NRC Research Press

Abu Dabous and Alkass Fig. 3. Utility functions for the bridge attributes.
Utility function for condition rating
100 90 80 70 60 50 40 30 20 10 0 35 40 45 50 55 60 65 70 75 80 85 90 95 100 100 90 80 70 60 50 40 30 20 10 0 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 2.3 2.5 2.7 2.9

1231

Utility function for load carrying capacity 100 90 80 70 60 50 40 30 20 10 0 0

Utility function for seismic risk

Utility

Deck condition index

Utility

Utility

Live load capacity factor Utility function for average daily traffic 100 90 80 70 60 50 40 30 20 10 0 0 20 40 60 80 100 120 140 160 180 200 Average daily traffic (Thousands of vehicles) Utility function for approach condition 100 90 80 70 60 50 40 30 20 10 0 0 1 2 3 Approach condition rating 4 5 100 90 80 70 60 50 40 30 20 10 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Drainage system index 1 100 80 Utility function for supporting road type 100 90 80 70 60 50 40 30 20 10 0

3 4 5 6 7 Seismic vulnerability index

10

Utility function for vertical clearance

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by ETH ZUERICH on 06/11/12 For personal use only.

Utility

Utility

40 20 0

Highway

National

Regional

Local

Utility

60

0.1

0.2 0.3 0.4 0.5 Vertical clearance index

0.6

0.7

Drainage system

Utility

Utility

Table 1. Attributes and utility values of projects (Bridge 10). Bridge 10 Criterion Condition rating Load carrying capacity Seismic risk Average daily traffic (thousands) Supporting road type Vertical clearance Approach condition Drainage system Expected utility value Attribute value 64 1.6 2.6 40 Local 0.25 Fair 0.60 Utility value 78.66 28.89 28.89 33.33 11.11 24.44 77.77 62.95 44.49

Finally, global priorities of the different rehabilitation strategies are estimated by multiplying the weights of the strategy with respect to each criterion by the criterion weight and finding the overall sum as follows: Major rehabilitation = 0.093 0.121 + 0.089 0.088 + 0.571 0.548 + 0.571 0.121 + 0.669 0.121 = 0.483 Minor rehabilitation = 0.221 0.121 + 0.323 0.088 + 0.286 0.548 + 0.286 0.121 + 0.257 0.121 = 0.278 Routine maintenance = 0.685 0.121 + 0.587 0.088 + 0.143 0.548 + 0.143 0.121 + 0.074 0.121 = 0.239 The analysis prefers bridge deck replacement and gives approximately the same weight for minor rehabilitation and routine maintenance.

Work program development

agency cost, user cost, bridge safety, bridge deck useful life, and environmental impact. Figure 4 presents the decomposition of the problem into a hierarchy. The first level is the overall goal of choosing a rehabilitation strategy. The second level represents the five criteria that contribute to the overall goal, and the third level represents the three candidate rehabilitation strategies. The judgments provided by the expert are provided in Figs. 5 and 6. The judgements are solicited by asking the expert to compare the decision elements in pairs and select linguistic expressions to define the intensity of relative importance as defined above in the second step of the AHP methodology. The matrix in Fig. 5 is obtained by comparing the set of criteria in pairs with respect to the overall goal. The matrices in Fig. 6 are obtained by comparing the alternative rehabilitation strategies in pairs with respect to each criterion. One of the most challenging tasks for bridge managers and decision makers is to select a work program to be performed when the available budget is limited. The purpose of this work program is to recommend a list of projects for improvement and to specify a rehabilitation strategy to be performed for each project. These recommended strategies maximize benefits to the agency and the users and can be implemented within the available budget. The recommended work program is developed by evaluating the various combinations of the different projects and the available rehabilitation actions. The problem under consideration is difficult to analyze manually since a large number of possible combinations can be developed and included for consideration. Simulation can be used to develop and evaluate the various possible work programs. The following is a description of the first three iterations of the simulation process. These iterations are intended to explain how work programs can be developed and evaluated.
Published by NRC Research Press

1232 Fig. 4. Hierarchy structure for choosing bridge rehabilitation strategy.

Can. J. Civ. Eng. Vol. 38, 2011

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by ETH ZUERICH on 06/11/12 For personal use only.

Fig. 5. Comparison of criteria with respect to the overall goal.

Fig. 6. Comparison of alternatives with respect to each criterion.

In the first iteration of the simulation, one project is considered. The selected project is the one with the highest priority. For this single project, the candidate work programs are the available rehabilitation actions: replacement, repair or maintenance of the deck. If the available budget is sufficient to perform any of these three alternatives, the one that has the highest weight is selected as the current best work program. For example, if the weight for replacement is 0.45, the weight for repair is 0.35, and the weight for maintenance is 0.2, the current

recommended work program will be to replace the bridge deck since it has the highest weight. The second iteration of the simulation considers two projects, which have the highest and the second highest utility. One of the available three maintenance, replace, and repair (MR&R) options can be selected for each project as the project is included in a possible work program. In this case, nine candidate work programs can be developed for evaluation. These work programs are: (replace1 and replace2), (replace1 and repair2), (replace1 and mainPublished by NRC Research Press

Abu Dabous and Alkass Fig. 7. Work program development using simulation.
Send a request to develop a recommended work program

1233

Retrieve the project with the highest overall utility

Develop the possible work programs

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by ETH ZUERICH on 06/11/12 For personal use only.

Evaluate the total weights of the rehabilitation strategies associated with each program

Estimate the cost of each program

Select a current best program

Add the next highest utility project

Yes

Cost < Budget

No
The current best with total cost less or equal the available budget is the recommended work program

tain2), (repair1 and replace2), (repair1 and repair2), (repair1 and maintain2), (maintain1 and replace2), (maintain1 and repair2), or (maintain1 and maintain2). The simulation develops these combinations and estimates the total weights and the total cost for the rehabilitation strategies associated with each of the developed programs. The work program with the highest total weight and total cost less than or equal to the available budget will become the current best replacing the one from the previous iteration. The third iteration will include the three highest utility projects and can have 27 candidate work programs. The simulation develops these programs and evaluates the total cost and total weight for each program. The program with the highest total weight for its rehabilitation strategies and total cost less than or equal to the available budget becomes the current best. The process continues adding new projects based on their utilities until the point when adding an additional project will produce work programs with total costs that cannot be implemented with the available budget, such that the cost of each work program exceeds the available budget. At this point the simulation stops and eliminates the programs that exceed the available budget. The program that has the highest total

weight and can be performed with the available budget is flagged as the recommended work program. Figure 7 shows the sequence of steps to develop the different candidate work programs and to select a recommended one. The cost of each MR&R action must be estimated since that will be needed to estimate an overall cost for the developed work program. This is important to ensure that the overall cost of the recommended work program does not exceed the available budget. The following section discusses the development of cost models for the MR&R actions.

Cost estimating models for rehabilitation actions

A number of maintenance and rehabilitation strategies are available for the bridge deck once it reaches the intervention level. These strategies can range from do-nothing to complete replacement. Cost estimating models to evaluate the cost of bridge rehabilitation actions are essential elements of the proposed methodology. One cost model for bridge deck replacement and one for deck major repair are developed here based on data and reports collected during the above mentioned interviews. The unit costs of the different items are extracted from winning bid proposals submitted by specialized contracPublished by NRC Research Press

1234 Table 2. Cost elements for bridge deck replacement. Item A.1 A.2 A.3 A.4 A.5 A.6 B.1 B.2 B.3 B.4 B.5 B.6 B.7 B.8 B.9 B.10 B.11 B.12 B.13 B.14 B.15 Item description Removal of asphalt wearing surface Removal railing Removal of concrete end posts Removal of existing deck including curbs Removal of top of pier Removal of existing approach slab Granular backfill Concrete in new top of existing piers Prestressed members (fabrication and erection) Concrete in barrier wall Concrete in deck (150 mm topping) Concrete in new deck extensions Concrete in approach slabs Stainless steel rebar in barrier wall Coated rebars in deck topping Rebars in deck extensions Rebars in approach slabs Bearings Abutment repairs Bridge deck waterproofing Asphalt Unit m2 LS LS m3 m3 m3 m3 m3 m m3 m3 m3 m3 tonne tonne tonne tonne each m3 m2 tonne Quantity 830 1 1 270 6 51 75 10 505 12 100 12 50 1.6 9 0.6 1 84 2 650 180

Can. J. Civ. Eng. Vol. 38, 2011

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by ETH ZUERICH on 06/11/12 For personal use only.

Unit cost ($) 6 3 000 7 500 525 1 500 150 60 750 900 1 275 675 900 375 15 000 2 250 1 500 1 500 165 6 000 18 90

Note: The cost data is in Canadian dollars, adjusted for inflation and based on the 2008 dollar value. LS, allowance or lot.

tors to Canadian ministries of transportation. The unit cost of each item includes the direct cost, indirect cost and the markup added to cover the contractors profit and contingency. A third model for maintenance cost is developed based on a model from the literature. The following sections discuss the development of cost models for bridge improvement projects. Bridge deck replacement Bridge deck replacement provides a brand new deck with the longest useful life. This option is normally performed by replacing the superstructure of the bridge. In concrete bridges, a bridge deck is integral with the girders which makes it difficult to remove the deck slab while keeping the girders in place. Hence, deck replacement typically includes replacing the girders which provides a new superstructure for the bridge. This is a major improvement with a relatively high cost. The cost of replacing the bridge deck depends on the type of the new superstructure to be constructed and the area of the deck. Saito et al. (1988) reported that the unit superstructure cost can be estimated in terms of dollars per square unit of deck area. The superstructure type is defined according to the slab and girders configuration. The most common arrangement is pre-stressed girders with a composite concrete slab on top. During the interviews with the bridge engineers, the development of a work breakdown structure for this type of arrangement was discussed and cost data for all the elements included in the structure was collected. Table 2 presents the work breakdown structure and the cost elements for bridge deck replacement. The total deck area is 930 m2 and the superstructure arrangement is 150 mm thickness concrete slab on pre-stressed concrete girders. The cost data presented in the table includes both direct and indirect cost elements in addition to the markup for profit and contingency. The cost data is in Canadian dollars, adjusted for inflation and based on the 2008 dollar value.

Analysis of the cost estimate for 17 bridge deck replacement projects showed that the unit cost can be between $837 and $1115 with a mean value equal to $917.50 per square metre. The replacement unit cost for each project is estimated by dividing the total project cost by the deck area. Bridge deck major repair Major repair can improve the deck to an excellent condition state. This option is performed by repairing the deck surface and installing a cathodic protection system. It involves the removal of the delaminated concrete from the deck surface and soffit and patch repair of the removal areas. A titanium mesh anode embedded in a normal concrete overlay will be installed to ensure cathodic protection of the reinforcing steel in the deck. This system is recommended since the overlay will allow the placing of waterproofing to prevent the ingress of water and deicing salts into the concrete. In this research, the total cost is estimated and linked to the bridge deck area to evaluate the cost per square metre for the major repair option. Table 3 presents a work breakdown structure for the major repair and the cost associated with each item. The cost includes both direct and indirect cost elements. As before, the cost data are in Canadian dollars based on the 2008 value. Analysis of the cost estimate for 19 bridge deck major repair projects showed that the unit cost can be between $669 and $792 with a mean value of $701 per square metre. The major repair unit cost for each project is estimated by dividing the total project cost by the deck area. Bridge deck maintenance cost The maintenance option does not involve any improvements to the condition or the structural aspects of the bridge deck. The maintenance activities include patching, sealing cracks or eliminating visible distresses that can accelerate the
Published by NRC Research Press

Abu Dabous and Alkass Table 3. Cost elements for bridge deck major repair. Item C.1 C.2 C.3 C.4 C.5 C.6 C.7 C.8 C.9 C.10 D.1 D.2 D.3 D.4 D.5 D.6 D.7 D.8 D.9 D.10 D.11 D.12 D.13 D.14 D.15 D.16 D.17 D.18 Item description Removal of asphalt wearing surface Removal of railings Removal of concrete end posts Removal of concrete curbs Type A removals from top of deck (delaminations only) Access to work area Type B removals from deck soffit Type C removals from fascia Type C removals from deck ends (tendon anchors) Removal of existing approach slab Granular backfill Scarify deck surface Cathodic protection Abrasive blast cleaning of rebar Abrasive blast cleaning for overlays Concrete overlay (includes padding for new crown) Finish and cure overlay Concrete barrier wall Concrete in new deck extensions Concrete in approach slabs Stainless steel rebar (barrier wall & deck extensions) Coated rebar for overlay padding area Rebars in deck extensions Rebars in approach slabs Abutment repairs Deck soffit repairs Bridge deck waterproofing Asphalt Unit m2 LS LS m3 m2 LS m3 m3 m3 m3 m3 m2 m2 m2 m2 m3 m2 m3 m3 m3 tonne tonne tonne tonne m3 m3 m2 tonne Quantity 830 1 1 20 300 1 2 5 1 51 75 625 625 300 325 100 625 12 12 50 2.2 4 0.6 1 2 4 650 180 Unit cost ($) 6 3 000 7 500 525 285 7 500 6 000 4 500 15 000 150 60 17 300 82 26 638 38 1 275 900 375 15 000 2 250 1 500 1 500 6 000 6 000 18 90

1235

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by ETH ZUERICH on 06/11/12 For personal use only.

Note: The cost data is in Canadian dollars, adjusted for inflation and based on the 2008 dollar value. LS, allowance or lot.

Fig. 8. Bridge deck maintenance cost.

Table 4. Rehabilitation stratey costs for the three top ranked bridges. Bridge Bridge 70 Bridge 40 Bridge 30 Deck area (m2) 1 250 1 100 950 Replacement cost ($) 1 146 875 1 009 250 871 625 Repair cost ($) 876 250 771 100 665 920 Maintenance cost ($) 61 806 52 556 39 583

Table 5. Weighted priorities for the rehabilitation strategies. Bridge Bridge 70 Bridge 40 Bridge 30 Replacement 0.45 0.35 0.60 Repair 0.30 0.35 0.32 Maintenance 0.25 0.30 0.08

Note: The cost data is in Canadian dollars, adjusted for inflation and based on the 2008 dollar value. Published by NRC Research Press

1236 Table 6. Candidate work programs. Program Program 1 Program 2 Program 3 Program 4 Program 5 Program 6 Program 7 Program 8 Program 9 Program 10 Program 11 Program 12 Program 13 Program 14 Program 15 Program 16 Program 17 Program 18 Program 19 Program 20 Program 21 Program 22 Program 23 Program 24 Program 25 Program 26 Program 27 Action 1 Replace 70 Replace 70 Replace 70 Replace 70 Replace 70 Replace 70 Replace 70 Replace 70 Replace 70 Repair 70 Repair 70 Repair 70 Repair 70 Repair 70 Repair 70 Repair 70 Repair 70 Repair 70 Maintain 70 Maintain 70 Maintain 70 Maintain 70 Maintain 70 Maintain 70 Maintain 70 Maintain 70 Maintain 70 Action 2 Repair 40 Repair 40 Repair 40 Replace 40 Replace 40 Replace 40 Maintain 40 Maintain 40 Maintain 40 Repair 40 Repair 40 Repair 40 Replace 40 Replace 40 Replace 40 Maintain 40 Maintain 40 Maintain 40 Repair 40 Repair 40 Repair 40 Replace 40 Replace 40 Replace 40 Maintain 40 Maintain 40 Maintain 40 Action 3 Maintain 30 Repair 30 Replace 30 Maintain 30 Repair 30 Replace 30 Maintain 30 Repair 30 Replace 30 Maintain 30 Repair 30 Replace 30 Maintain 30 Repair 30 Replace 30 Maintain 30 Repair 30 Replace 30 Maintain 30 Repair 30 Replace 30 Maintain 30 Repair 30 Replace 30 Maintain 30 Repair 30 Replace 30 Total cost ($) 1 957 558 2 583 895 2 789 600 2 195 708 2 822 045 3 027 750 1 239 014 1 865 351 2 071 056 1 686 933 2 313 270 2 518 975 1 925 083 2 551 420 2 757 125 968 389 1 594 726 1 800 431 872 489 1 498 826 1 704 531 1 110 639 1 736 976 1 942 681 153 945 780 282 985 987

Can. J. Civ. Eng. Vol. 38, 2011

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by ETH ZUERICH on 06/11/12 For personal use only.

Total weight 0.88 1.12 1.4 0.88 1.12 1.4 0.83 1.07 1.35 0.73 0.97 1.25 0.73 0.97 1.25 0.68 0.92 1.2 0.68 0.92 1.2 0.68 0.92 1.2 0.63 0.87 1.15

Note: The cost data is in Canadian dollars, adjusted for inflation and based on the 2008 dollar value.

corrosion of the deck reinforcement. The maintenance cost depends on the condition of the bridge. The available studies report that annual maintenance costs range from 1% to 2% of the reconstruction cost (Branco and de Brito 2004). De Brito and Branco (1998) developed a graphical representation for maintenance cost in relation to the bridge deck area. They described a linear relationship between the deck area and the maintenance cost. In addition, they specified that the maintenance cost for a 4000 m2 deck is double the maintenance cost for a 400 m2 deck and that the maintenance cost for a 400 m2 deck is double the maintenance cost for a 100 m2 deck. Increased maintenance cost for a bridge deck is estimated to be 5% of the reconstruction cost. Applying the linear relation suggested by Branco and de Brito (2004), the graph for the maintenance cost is developed as shown in Fig. 8. From this representation, the unit cost for a 400 m2 deck is $40 per square metre and for a 4000 m2 deck it is $80 per square metre. Using this graphical representation, the maintenance cost for any bridge deck can be estimated by multiplying the corresponding unit cost for the area by the bridge deck area.

Case example
To illustrate the development of a recommended work program, the following case example is presented. The case example is based on a case study discussed in Abu Dabous (2008). The three bridges with the highest rank from the case study are considered since they have the highest utility and must be prioritized for intervention. The cost of the three rehabilitation strategies for each project is estimated from the

cost models developed earlier. The cost estimates are shown in Table 4. The decision maker provides specific judgments for each bridge to evaluate the different rehabilitation actions and to develop a weight for each option. The weights for the rehabilitation actions are provided in the Table 5. The 27 candidate work programs are developed as all possible combinations of high priority projects and the available rehabilitation strategies as shown in Table 6. The total cost in Table 6 is estimated by finding the sum for the cost of all the rehabilitation actions associated with each program. The cost of each action is provided in Table 4. Similarly, the total weight is estimated by finding the sum for the weights of all the rehabilitation strategies involved in the program. The priority for each action is given in Table 5. Assuming that the available budget is $2.10 million, work programs that cost more than the available budget are not possible, which means that work programs 2, 3, 5, 6, 11, 12, 14, and 15 must be eliminated. Work program 9 has the highest total weight of 1.35 and a total cost of $2 071 056, and work program 18 has the second highest weight of 1.20 and a total cost of $1 800 431. The decision makers can compare these two efficient work programs to select the more attractive one for the agency. Analysis of the above case study has shown that 27 possible work programs are available for three projects with three possible actions. The number of possible work programs will increase significantly as the number of bridges and available rehabilitation actions increases. For instance, 20 projects with
Published by NRC Research Press

Abu Dabous and Alkass

1237 Branco, F., and de Brito, J. 2004. Handbook of concrete bridge management. American Society of Civil Engineers, U.S. De Brito, J., and Branco, F. 1998. Computer aided lifecycle costs prediction in concrete bridges. Engineering Modeling, 11(34): 97106. Frangopol, D., and Liu, M. 2007. Maintenance and management of civil infrastructure based on condition, safety, optimization, and life-cycle cost. Structure and Infrastructure Engineering, 3(1): 29 41. doi:10.1080/15732470500253164. Frangopol, D., and Neves, L. 2003. Life-cycle maintenance strategies for deteriorating structures based on multiple probabilistic performance indicators. Edited by F. Bontempi, System-Based Vision for Strategic and Reactive Design, Vol. 1. Sweets and Zeitlinger, Lisse. Frangopol, D., Kallen, M., and Noortwijk, J. 2004. Probabilistic models for life-cycle performance of deteriorating structures: review and future directions. Progress in Structural Engineering and Materials, 6(4): 197212. doi:10.1002/pse.180. Gopal, S., and Majidzadeh, K. 1991. Application of markov decision process to level-of service-based maintenance system. Transportation Research Record, No. 1304, Transportation Research Board, Washington, D.C., pp. 1217. Hawk, H. 2003. Bridge life-cycle cost analysis. NCHRP Report No. 483, Transportation Research Board, National Research Council, Washington, D.C. Huang, Y., Adams, T., and Pincheira, J. 2004. Analysis of life-cycle maintenance strategies for concrete bridge decks. Journal of Bridge Engineering, 9(3): 250258. doi:10.1061/(ASCE)10840702(2004)9:3(250). Jiang, Y., Satio, M., and Sinha, K. 1988. Bridge performance prediction model using the Markov chain. Transportation Research Record, No. 1180, Transportation Research Board, Washington, D.C., pp. 2532. Keeney, R., and Raiffa, H. 1993 Decision with multiple objectives preferences and value tradeoffs. Cambridge University Press, New York, N.Y. Kong, J., and Frangopol, D. 2003. Life-cycle reliability-based maintenance cost optimization of deteriorating structures with emphasis on bridges. Journal of Structural Engineering, 129(6): 818828. doi:10.1061/(ASCE)0733-9445(2003)129:6(818). Kulkarni, R.B., Miller, D., Ingram, R.M., Wong, C., and Lorenz, J. 2004. Need-based project prioritization: alternative to cost-benefit analysis. Journal of Transportation Engineering, 130(2): 150158. doi:10.1061/(ASCE)0733-947X(2004)130:2(150). Madanat, S., and Ben-Akiva, M. 1994. Optimal inspection and repair policies for infrastructure facilities. Transportation Science, 28(1): 5562. doi:10.1287/trsc.28.1.55. Madanat, S., Karlaftis, M., and McCarthy, P. 1997. Probabilistic infrastructure deterioration models with pannel data. Journal of Infrastructure Systems, 3(1): 49. doi:10.1061/(ASCE)1076-0342 (1997)3:1(4). Mirza, S., and Haider, M. (2003) The state of infrastructure in Canada: implications for infrastructure planning and policy, A Report Prepared for Infrastructure Canada, McGill University, Montreal, Que. Saaty, T.L. 1980. The analytic hierarchy process. McGraw-Hill, New York, N.Y. Saito, M., Sinha, K., and Anderson, V. 1988. Bridge replacement cost analysis. Transportation Research Record, No. 1180, Transportation Research Board, Washington, D.C., pp. 1924. Vanier, D. 2000. Advanced asset management: tools and techniques, innovations in urban infrastructure. Seminar of the APWA International Public Work Congress, Louisville, KY, pp. 3957.
Published by NRC Research Press

three possible rehabilitation actions will yield 3 486 784 401 work programs. Developing this large number of work programs and evaluating them will require relatively long processing time. This limitation of the proposed methodology needs to be further evaluated to study options to eliminate its impact.

Conclusions
Allocating limited available budget to maintain bridge networks and selecting effective MR&R strategies for each project are among the most challenging tasks for managers and decision makers. The research investigated the current decision making process used for managing bridge infrastructure. Based on that, a decision support methodology is developed that integrates multi-criteria decision making methods to rank and prioritize project for intervention and evaluate the available rehabilitation strategies. The methodology enables the decision makers to include their judgments and inputs into the decision making process and modify the decision making attributes and criteria based on their requirements. The research proposes a technique to develop a recommended work program that can be implemented under budget constraints. The technique uses the simulation to develop all the possible work programs and to evaluate effectiveness of the rehabilitation strategies associated with each program. During the course of this work, future research directions arose. First, data and cost estimating models for bridge rehabilitation actions are scarce or inaccessible as compared to building cost data. More data needs to be collected to extend the cost estimating models for bridge deck rehabilitation and to refine the proposed models accuracy. In addition, more models are needed for the different bridge elements including joints, girders, abutments, and piers. Second, methods to quantify enhancements from implementing rehabilitation strategies and work programs need to be further investigated and developed. Third, multi-year programming for budget allocation is another essential and complex issue. Further research is needed to develop a sequence of recommended work programs over an extended planning period that can ensure the most efficient use of resources over the life cycle of bridge networks. Fourth, the proposed methodology needs to be extended to other bridge components and can be implemented into a full-scale computerized bridge management system.

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by ETH ZUERICH on 06/11/12 For personal use only.

Acknowledgement
The partial financial support from the Natural Sciences and Engineering Research Council of Canada is appreciated.

References
AASHTO. 2008. Bridging the Gap [online]. Available from http:// www.transportation1.org/bridgereport/scarce.html, Abraham, D., and Wirahadikusumah, R. 1999. Development of prediction model for sewer deterioration. In Proceedings of 8th Conference on Durability of Building Materials and Components, National Research Council (Canada), Ottawa, pp. 12571267. Abu Dabous, S. 2008. A decision support methodology for rehabilitation management of concrete bridges. Ph.D. thesis, Department of Building, Civil and Environmental Engineering, Concordia University, Montreal, Que. ASCE. 2009. Report card for Americas infrastructure [online]. Available from: http://www.infrastructurereportcard.org/ [cited 7 February 2011].