Prioritization of bridges and tunnels in earthquake risk mitigation using multicriteria decision analysis: Application to Lisbon

Abstract

This paper presents the development of a multicriteria value model enabling the prioritization of bridges and tunnels according to their structural vulnerability and strategic importance for the formulation and implementation of civil protection policies, both for retrofitting and emergency management, in face of seismic events. An interactive structuring process was developed with a group of key-players to carefully define the evaluation criteria and the MACBETH approach was extensively used (i) to facilitate the assessment from the group of the judgmental information necessary to build value functions and (ii) to establish relative weights for the criteria. The model was subsequently explored to prioritize the bridges and tunnels of a zone in Lisbon with high seismic hazard.

Introduction

Strong earthquakes in urban areas may damage several critical structures, such as bridges and tunnels. Their loss of functionality has a considerable detrimental effect on after-event emergency response. Investing in prevention actions, such as the application of retrofitting measures and upgrading to current seismic design codes, aims at contributing to reduce the risk and extent of earthquake damages. However, as evidenced in seismic evaluation studies in USA, Japan and New Zealand [1], the civil engineering processes of retrofitting bridges and tunnels are extremely costly and time consuming, whereas the authorities responsible for their maintenance and repair are constrained by limited financial and human resources. Consequently, the availability of a model allowing a quick identification and prioritization of the structures to be used in the context of an emergency management, and that need to be retrofitted, is extremely important [2].

As stated by Phillips and Bana e Costa [3], “the correct basis for prioritization, the one that ensures that best value is obtained for the available resource, is risk-adjusted benefit divided by cost.” However, most bridge-upgrading “prioritization schemes are developed in terms of relative risk without an attempt at quantification of cost or benefit” [4]. The primary objective of this paper is to show, through the description and discussion of a real-world case, that a multiple criteria decision analysis is adequate to model the benefit component of a benefit/cost ratio for prioritization. A fundamental step is to identify and rank all deficient bridges and tunnels in a specific urban region living with the ever-present possibility of a catastrophic seismic occurrence of high magnitude, as is the case of Lisbon.

The estimate of the safety level of each structure within the area by statistical analysis correlating the level of damage, observed after the occurrence of earthquakes, with the structural attributes of the infrastructures and local geological characteristics [5], is not possible to be performed in Lisbon, due to lack of data. Therefore, detailed seismic inspection of each bridge and tunnel is necessary to decide which structures need retrofitting and, also, to calculate the respective costs for an acceptable level of risk [6]—note that high retrofitting costs may justify the alternative of building a new structure, depending on the age of the seismically deficient structure and other cost-benefit considerations [7]. Because of the significant number of structures and the limited resources available, the structures must also be ranked for inspection, for which purpose the multicriteria model presented in this paper is also useful.

Basöz and Kiremidjian [8] review the best known approaches for prioritizing bridges for seismic retrofitting, some of them implemented in several states of USA, in Japan, and in New Zealand. A critical analysis of typical methodological and practical shortcomings is offered by Maffei [1]. This author also observes that the prioritization approaches “typically need data for each bridge on (a) the level of seismicity, (b) the importance of the bridge, and (c) the seismic vulnerability due to soil and structural characteristics” [1]. The higher the levels of seismicity, importance and vulnerability, the higher the overall benefit of retrofitting a structure. Some approaches mix the intervention cost with the benefit dimensions. In fact, only Maffei and Park [7] separate benefits from costs and propose the methodological correct use of a benefit/cost ratio for prioritization, although they do not use any multicriteria analysis in the benefit component of their ratio.

The majority of the approaches use rating scales aggregated by average sum, with a few exceptions, like the mostly multiplicative models developed by the California Department of Transportation [9], [10] and by the Japan Ministry of Construction [11]. The multiattribute additive rating system developed by the Applied Technology Council [12] is one of the most widely used in USA. Two other models follow the principals of Multiattribute Utility Theory [13]. These are the additive utility model developed for IDOT, the Illinois Department of Transportation [14], and a similar model proposed by Basöz and Kiremidjian [8]. The latter is an improved conceptual model, using hypothetical utility functions and weights, whereas the former is a complete real-world application of MCDA to bridges prioritization: single-attribute utility functions were developed by expert opinion surveys and scaling constants (“weights”) based on value trade-offs assessed by an analyst from five representatives of IDOT [15].

These models offered us insight in designing an effective multicriteria approach to the problem of prioritizing bridges and tunnels for earthquake risk mitigation in Lisbon. As in other engineering management contexts in which we have offered our decision analysis and process consultation expertise [16] to the Lisbon local and regional authorities [17], [18], there was a clear need from the Lisbon authorities for a decision support tool that would also avoid the pitfalls and drawbacks associated with ad-hoc incorporation of expert judgements in public decisional processes. The second key objective of this paper is to show that experts can make judgements about “differences of attractiveness” [19, p. 212] between consequences in a complex engineering context.

In model building process, we made use of MACBETH (Measuring Attractiveness by a Categorical Based Evaluation Technique), an approach founded on difference measurement, that requires only qualitative judgements about differences of attractiveness to quantify the relative attractiveness (priority, in our case) of options (the bridges and tunnels). The approach, based on the additive value model, aims to support interactive learning about the problem and the elaboration of recommendations to prioritize and select options in individual or group evaluation processes. The theoretical foundations of MACBETH and an extensive list of references to real-world applications can be found in [20] and a straightforward introduction to the approach and the M-MACBETH software is offered in [21]. As remarked by Belton and Stewart [22] and Bana e Costa and Chagas [23], asking only for qualitative judgemental information enables one to avoid the difficulty of making direct numerical judgements. Technically, the use of MACBETH to construct value functions and define weights clearly distinguishes the additive value model described in this paper from the other multicriteria models above mentioned.

The model was built in direct interaction with a group of eight key players (two civil engineers and four experts from the Civil Protection Department of Lisbon and two technicians from the Medical Emergency Institute) and tested in an urban zone of Lisbon classified as a critical area of seismic risk in the Emergency Plan for Seismic Risk of the Lisbon Municipality [24]. The zone is composed of historical residential and commercial districts and it includes a main road (a lifeline belonging to the primary network connecting the suburban area in the north of Lisbon to the city centre) and two infrastructures that support emergency response (the Santa Maria Hospital, one of the highest capacity hospitals in Lisbon, and the Av. Rio de Janeiro Fire Brigade Department, where the ambulance fleet used to rescue earthquake victims is based). The transportation network in the zone includes a total of 10 bridges and tunnels (S1–S10), eight of them belonging exclusively to the road network while the other two are part of the train and underground networks of Lisbon. Section 2 describes the structuring component of the multicriteria process and Section 3 is devoted to the MACBETH model building phase during which priority functions and criteria weights were defined. In Section 4, the model is explored to rank the 10 infrastructures of the selected zone. Some conclusions are presented in Section 5.