The goals of the PEER Seismic Hazard Simulation of Bay Area Highway Network Transportation Analysis are to develop appropriate analytical and computational methods for evaluating the impact of a transportation system on an urban area and to illustrate the utility of these methods through an application to a region. In general, risk analysis of highway transportation systems is performed with the objective to provide:

• appropriate information in the retrofit and disaster mitigation decision process (in the pre-event and post-event period, it is necessary to determine which bridges are to be retrofitted/repaired/replaced, in what order and to what design level); and
• models and tools for estimating the socio-economic impact of transportation systems that have been damaged by an earthquake, and the benefits (losses foregone) of current and proposed retrofit and reconstruction programs affecting transportation structures.

Linking models of network performance, transportation demand, bridge performance, and seismic hazard are the key to this effort.

One of the main objectives of this project is to apply the methodology to the Bay Area highway transportation system. During the first PEER Transportation Risk Analysis Workshop held in 1998, the workshop participants recommended that the application area be the San Francisco Bay region. The rationale for the selection was that the region has a very complex transportation network with limited redundancy. In particular, it is very likely that the major bridges in the region will be subjected to ground motions of similar severity due to their proximity to major faults in the area. All the long-span bridges, the San Francisco-Oakland Bay Bridge, Golden Gate, San Mateo, Dumbarton and San Rafael, are flanked by the San Andreas Fault to the west and the Hayward fault or its extension to the east.

PEER’s System Vision identifies three levels of decision making supported by Performance-Based Earthquake Engineering:

1. owners and investors of individual facilities
2. building system managers, including infrastructure authorities, and
3. public authorities considering regulatory costs and benefits.

The objective of this research is to determine how to make design decisions about transportation networks that reflect specific seismic performance objectives. The ultimate goal of this research is to support decisions in the second and third categories. In an infrastructure context, decisions of the second and third types interact, with infrastructure authorities in the executive branch making recommendations to commissions and legislatures, which in term make funding allocation and burden decisions in response to these and other recommendations.

The combined activities at USC and Stanford leverage the research done on buildings and bridges by tying these findings to seismic hazard and transportation network performance. The project is intended to serve as a template for leveraging research findings in the context of the Highway Testbed.

Nonlinear programming (for network equilibrium analysis and gravity model estimation), economic modeling, and trial and error.

Three major tasks have been accomplished over the past year. These are update of the MTC highway network, improvement of the traffic assignment model, and sensitivity study of the bridge closure criteria.

1. Update of Network and Demand Data
The previous study was based on 1990 MTC highway network and trip demand data. Over the past several years, MTC has updated the San Francisco Bay Area highway network. Some new links were added, and some of the links that were coded incorrectly were corrected. The travel demand between each origin-destination zone pair was also updated in 1998. We therefore updated our transportation network accordingly. The highway bridges were merged to the updated street network using GIS tools. This merging task made it possible to translate bridge damage into transportation link functionality.

2. Improvement of the Traffic Assignment Model
A major problem that was identified based on our previous work is that the conventional traffic assignment model represents normal situations well, but over estimates congestion levels in earthquake scenarios. This problem has been addressed. A variable-demand traffic assignment model was developed for the study area. In this model, traffic demand between any two zones is a function of the travel cost between the zone pair. This variable-demand model is calibrated with the conventional fixed-demand model in baseline scenario. The level of service over all the highway links in the study area are compared across the two different assignment models and across the baseline and an earthquake (M=7.5 on the Hayward Fault) scenarios. See Figure 1.

Figure 1: Comparison of v/c Ratios between the Fixed-demand and the Variable-demand Models in the Baseline and Earthquake Scenarios
Larger View

The modified network level of service in the earthquake scenario results to account for changes in the transportation demand between each origin-destination zone pair. Figure 2 shows the pattern of redistribution of the travel demand.

Figure 2: Proportionate Difference in Trip Production, Variable vs. Fixed Demand Models
Larger View

3. Sensitivity Study of Bridge Closure Criteria
The decision to open or close a bridge subject to a certain level of damage is mostly based on an inspector’s engineering judgment. It is reasonable to assume that bridges subject to severely damage or worse should be fully closed and bridges with minor damage can remain open. Bridges with moderate damage are most uncertain with respect to functionality following an earthquake. Two bridge closure criteria have been adopted. These are listed in Table 1. The resultant change of functionality, total number of assigned trips, and total vehicle hours are also listed in Table 1.

Table 1: Sensitivity Study – Comparison of the Two Bridge Closure Criteria

The ultimate product of our research is intended to be tools for decision support. Decisions concerning seismic risks to transportation networks must be made both before and after earthquakes. Pre-event decisions focus on allocation of resources to retrofit, construction, and network design options. Post-event decisions focus on resource allocation to repair, reconstruction, replacement, and capacity management options.

Retrofitting or reconstruction facilities prior to an earthquake can best summarized as an exercise in stochastic network design. Given scarce resources, which links in the network should be improved to best mitigate uncertain seismic risks to system performance? This is obviously a very challenging question. Deterministic network design problems involving such discrete investments are well investigated, and standard approaches have been developed. Stochastic network design problems have been stated, but have not been treated analytically beyond the level of toys. Still, public agencies in seismically active areas cannot afford to ignore this question. Every seismic event results in temporary opportunities for public authorities to intensify public investments in the mitigation of seismic risk, and this includes the stochastic network design decisions that are made every time a transportation structure is retrofitted.

Post-event decisions have the advantage of being deterministic, because they take place after the random events of an earthquake have been realized. They have the disadvantage of being very large problems. Sequencing repairs to hundreds or even just dozens of earthquake-damaged bridges in the SF Bay Area network is a far larger network design problem than is normally encountered in engineering practice. The challenge exceeds the reach of available methodologies. Still public authority cannot afford to ignore the problem, because the problem will confront us in the foreseeable future.

In previous research funded by NSF and PEER, we have specified a computable model that shows the effects of earthquakes on transportation networks, building stocks, and regional economic activities. This model accounts for interactions between the urban economy and the transportation network.

Previous research done with MCEER and FHWA has focused more on a probabilistic treatment of hazards and has suppressed representation of the urban economic activity system. Research done at PEER and MCEER has been metropolitan and intra-urban. Work done at MAE has been regional and interurban.

The project team will normally participate in scientific meetings organized by PEER, and acknowledge PEER sponsorship in any papers published as a result of this research.

Year 5 research activity will conclude March 31, 2003. Deliverables associated with PEER project 3222001 include brief reports submitted at intervals defined by PEER, and a comprehensive final report at the conclusion of the project. The final report will be coordinated with the final report document provided by the Stanford team.