ENGINEERING MANAGEMENT OF LIFELINE SYSTEMS UNDER EARTHQUAKE RISK

The state-of-the-art of lifeline system management under earthquake risk is discussed. After a historical overview and specific features of lifeline earthquake engineering, elements of lifeline earthquake engineering practice are presented. They include outline of lifeline protection technologies, lessons from recent urban earthquakes particularly Kobe 1995, and seismic code developments. Then two topics are addressed specifically. First seismic reliability under system interaction is discussed with its general scheme and an analytical methodology. Next, criteria for lifeline performance is discussed with an emphasis on quantification of qualitative issues that are needed particularly in implementation processes. After presenting a general framework, two cases, one from Kobe on water customers' response, and one from California regarding funding procedure for seismic improvement. Finally, future orientations of earthquake engineering and its relation to lifeline earthquake engineering is addressed. It is emphasized that we should establish a third generation earthquake engineering where structural and geotechnical mitigation, crisis management, and sustainable development should be integrated through multi-disciplinary developments. Some examples of multi-disciplinary research initiatives are introduced.


2827
Under the worldwide trend of urbanization, earthquake protection of urban regions is a key issue.Conceptually, the target may be evaluated from the following two distinct points of view.
(l) Seismic safety of urban regions.
(2) Seismic reliability of urban regions.The seismic safety is related to a state of being free from casualties and property losses, whereas seismic reliability stands for a state of being free from unacceptable loss of urban functions.
The major role of earthquake engineering in a traditional trend has been realization and enhancement of seismic safety.The research and technology developments have been focussed on structural and geotechnical issues to make engineered structures earthquake resistant through appropriate design and construction.Such definitive efforts have lasted for more than three quarters of this century.In the case of Japan, the development of modern earthquake engineering was initiated upon the experience of the devastating Kanto Earthquake of 1923, where it was demonstrated that seismic design is useful, effects of geotechnical conditions on ground motion intensity is important, etc.This aspect of earthquake engineering may be called "structural earthquake engineering." Whereas seismic safety remains a major role of earthquake engineering, seismic reliability has grown an issue important as well in the recent decades.There are two principal reasons for this.(1) In the trend of growing urbanized societies, the requirements for high seismic reliability of urban functions under seismic risk has become a key issue.As modern urban functions rely on lifeline systems much more heavily than they used to, Lifeline systems are characterized by their wide spatial distribution.An entire lifeline system forms a network configuration that is composed of various types and a large number of link components and node components.Lifeline facilities must be constructed throughout their service areas regardless of the soil conditions of their sites.Because of the fact that many big cities and their metropolitan areas worldwide are situated in coastal zones and/or near river mouths and are inevitably subject to poor soil conditions due to soft sedimentary deposits, the lifeline systems serving such regions involve inherent vulnerability in terms of their seismic risk.
Another aspect that are concerned with earthquake vulnerability of lifeline systems is that they generally constitute highly hierarchical systems.The hierarchical systems may be characterized as: hierarchy level of components number of facility units high small low large effects of disruption large small Because of the large number of units and relatively lower effects of their disruption on the entire system functionality, it is often difficult, not feasible under economic constraint, to provide the components at a low hierarchy level, e.g., terminal water distribution network with small diameter pipes, with uniform earthquake resistance regardless of soil conditions, or high resistance same as that of other components at higher hierarchy levels.
The specific feature of lifeline systems characterized by their wide spatial distribution and their highly hierarchical system structure has led lifeline earthquake engineering to consist of the following major branches dealing with technologies for: (1) enhancement of structural performance of individual lifeline components, (2) seismic reliability management of lifeline networks, and (3) methodological developments for efficient post-earthquake operations.A comprehensive combination of the outputs from these branches is a distinctive feature of lifeline earthquake engineering, while other facilities like buildings, dams, bridges, etc. primarily rely on the technologies from structural earthquake engineering.Systematic observations of this aspect were made by many authors including Nojima, Kameda and Shinozuka [1990], and Kameda and Nojima [1992].

Outline of Lifeline Protection Technologies
On the basis of better understanding of seismic behavior of lifeline systems, earthquake countermeasures have been developed and implemented in actual systems by many lifeline sectors.Earthquake countermeasures for lifelines are aimed at abating outcome of an earthquake; specifically, they are (1) preventing and mitigating physical damage, (2) minimizing service malfunction, (3) mitigation of secondary disasters, and (4) executing an expeditious recovery.
To realize these objectives, various disaster mitigation measures have been established, which may be classified in the following four categories: (1) upgrading structural performance of network components.
block separation systems, information management systems, monitoring systems, mutual-aid agreements, expert systems On which categories of these four emphasis are placed depends on the system characteristics of individual lifelines.It is noted that the set of earthquake countermeasures are the consequences of basic studies in lifeline earthquake engineering as well as creative technology developments in the each lifeline sectors and related industries.Lifeline interactions unfortunately played a pivotal effect in disaster chain in the aftermath of the Hanshin-Awaji earthquake disaster.Interaction related problems are summarized in issues related to i) physical damage propagation, ii) functional damage propagation, iii) recovery interruption, iv) back-up functions of substitutive systems, and vi) compound damage propagation.(3) Functional restoration required a very long period of time.Because of the extensive damage as described previously, restoration work for social infrastructure needed an enormous operation.In addition to the activities by the lifeline sectors directly in charge of the damaged systems, nation-wide operation aid teams were organized on the governmental as well as professional association bases.Yet the functional restoration needed much longer times, five -ten times, than those we experienced in recent major urban earthquakes, namely, 82 days for water supply, 85 days for natural gas supply, 7 days for power supply, 20 months for highway bridges, and 6 months for major lines of railways.As an example, Figure 1 compares the restoration curves for water supply systems in recent urban earthquakes.Quantitatively very long restoration period caused qualitatively very severe users' difficulties as will be discussed in a later chapter.

Days after
Despite difficult situations of long lasting lifeline disruption, there were important findings regarding lifeline earthquake engineering practice, which may be summarized as follows.
(1) Under this heavy structural and geotechnical failures of lifeline facilities, modern lifeline engineering technologies proved to be promising in various ways.High-performance pipes including steel pipes with modern butt welded joints and ductile-iron pipes with seismic joints performed well with no substantial leaks even in areas where extensive liquefaction-induced permanent ground displacement occurred, while old cast iron pipes broke easily even under relatively firm soil condition, see photographs in Figure 2. Development of modern earthquake engineering has, thus, proved to be in a right track.
(2) Methodologies of lifeline management under earthquake emergency were tested in the true earthquake event.
Among them, emergency shut down valves installed in distribution tanks for the Kobe municipal water supply system functioned well, and reserved water of 40,000 m 3 which was used for emergency delivery.
Block separation system of the Osaka Gas Co. was activated to separate the heavily damaged area from areas with slight damage or no damage, which enabled the repair teams concentrated in the limited disaster area, while service was continued in the other areas around.These operational methods generally proved useful to meet earthquake disaster emergency needs.
(3) In reconstruction of structures for civil infrastructure systems, efforts were taken to incorporate new visions to initiate the next generation of earthquake engineering technology.Highway bridge reconstruction was performed with a new design load to consider near field ground motions from intra-plate earthquakes.Some highway viaducts were reconstructed on the basis of comprehensive structural system design concepts with base isolation rather than component design of individual structural elements (Figure 3).Urban highways were rebuilt in a way to improve an environmental problem of noise from traffic vehicles (Figure 4).In reconstruction of harbor facilities for Kobe Port, strategic decisions were made to design liner berths at selected important locations for design seismic loads that are higher than the others.These experiences practiced in the reconstruction activities constituted bases for the seismic code revisions, retrofit of existing structures, and related activities that followed.
Figure 2: Buried pipelines under earthquakes -ductile iron pipes with seismic joints surviving large permanent ground deformation (left) and brittle failure of cast iron pipes (right) Figure 3: Nineteen-span continuous girder-frame Figure 4: New constructions for improvement of structure with base isolation at pier foot adopted environmental problems from traffic noise for reconstruction of Benten viaduct in Kobe adopted in reconstruction from disaster (Ashiya)

Seismic Code Developments for Lifeline Systems
The seismic codes for civil infrastructures including highway bridges, railway facilities water and sewer facilities, buried gas pipes and high pressure gas facilities were revised following the Hyogoken-nanbu (Kobe) Earthquake.The general concepts for these revisions reflect the JSCE Recommendations on Earthquake Resistance of Civil Engineering Structures [JSCE 1996a], which was proposed as the output from a Special Task Committee of Earthquake Resistance of Civil Engineering Structures organized in March 1995.Major decisions made in the code revisions have following common features.
(1) Two levels of earthquake motion, i.e., Level I and Level II design motions have been incorporated.In principle, the Level I motions account for moderate seismic loads experienced at sites from random sources, while the Level II motions stand for near field ground motions immediately from the causative fault.The Level II motion is to be determined directly from scenario mechanism of susceptible active faults.For the cases where the fault survey is difficult because of thick alluvial and diluvial deposits, standard Level II motions may be specified on the basis of recorded near-field ground motions.
(2) The concept of performance-based design and management has been adopted extensively.The importance of structures and facilities are classified into several categories, and required seismic performance is defined from the relation of ground motion levels and their probability of occurrence.
(3) For the design against the Level II motion, it is basically required to estimate the plastic deformation and ultimate strength of structures including RC and steel structures as well as soil structures including embankments, quaywalls and foundations.Buried pipes in liquefiable grounds are subejected to 1.2 -2.0% ground strain.1998 1997 1997 1995 1996, (2000) 1996 1995, (2000) Thus, to consider specific active faults for Level II design is a new direction that emerged in the seismic codes of civil infrastructures.This does not necessarily mean that the scenario earthquake-based ground motions should be developed for each lifeline system.A better way is to generate hazard conditions for general regional disaster management plan uses, and apply them to lifeline system designs.
In doing so, ground motions should be generated by using most advanced simulation techniques that take into account the source mechanism such as asperity distributions and rupture propagation, source-to-site propagation effects for high frequency components by using semi-empirical methods, three dimensional topographical effects on low frequency components, nonlinear soil amplifications, quantitative evaluation of modeling uncertainties and uncertain parameter, etc.Such an example is found in Irikura [1998) with an application to Osaka basin.In this way, a definite orientation is perused for obtaining a common basis of science-based regional hazard assessment to cope with earthquake disasters caused by strong near-field ground motions under rupture of specific active faults, that are regarded as "low probability but high impact disaster." As a matter of fact, there are differences in the ways this concept is incorporated among different lifeline facilities as shown in Table 1.This depends on the actual structural and system performance of lifelines under the Kobe disaster as particularly to how well their modem technology worked.The concept is already in practice, but its feasibility is not uniform throughout the country, depending on the tectonic characteristics and levels of information available.In the following paragraphs, two typical examples of code revisions, water supply systems and railway facilities, are introduced where advanced concepts have been perused intensively.
( 1) JWW A Guidelines for Water Supply Facilities The Japan Water Works Association (JWW A) revised its Seismic Design and Construction Guidelines for Water Supply Facilities [JWW A, 1997;Matsushita, 1999a].General design flowchart is shown in Figure 5.Its specific features are characterized by the "importance level" and "system improvement" boxes.
System components are classified according to their importance levels A and B. Corresponding to each importance level, different performance requirements are specified for Level I and Level II design seismic loads, on which basis design calculations are executed The descriptions are such as "no damage" (A&I), "functionality maintained under minor damage at individual components" (B&I), "no serious threat to human life / functionality to be maintained under minor damage at individual components" (A&II), and "functionality of entire system to be maintained with structural damage and malfunction at some individual components" (B&II).
It is also a feature of the JWW A guideline that options between structural improvement and network redundancy  enhancement are recognized to be equally feasible as means for realizing system performance requirements.
(2) Design Standards for Railway Structures Another example is the new Earthquake Resistant Design Standards for Railway Structures, which became effective in December 1998 after a four-year activities based on comprehensive studies of the disaster in Kobe, research on structural and system response and failure under very strong near-field ground motions, theoretical as well as experimental [RTRI 1999, Nishimura 1999).Its major feature may be recognized in its most advanced performance-based-design oriented format that have sound scientific and engineering bases.
Quantitative links are provided among (a) seismic load definition (Level I & Level II), (b) structural and system performance requirements in three levels (Pl, PIT & PIil), and (c) damage states corresponding to each performance level (structural member damage & foundation stability).On this basis, the consistent design calculations and evaluations are conducted as shown in Figure 6.Structural responses are determined from dynamic analysis.Damage states and stability are evaluated from push-over analysis.

Scheme of System Interactions
It was pointed out in 2.2 that extensive cascade effects and system interactions occurred in the 1995 disaster in Kobe.Not only the case of Kobe, system interactions are a reality in many urban earthquakes.The interactions take place in various manner.On the basis of information from actual earthquake disaster, they can be shown in a form of "interaction matrix" indicating the relations between impacting and impacted systems [Kameda, 1994).The system interactions that were observed in the Great Hanshin-Awaji (Kobe) Earthquake Disaster are found in Nojima and Kameda (1996).On these bases, the system interactions can be categorized into four groups; i.e., (a) physical damage propagation, (b) functional damage propagation, (c) recovery interruption, and (d) back-up functions of substitute systems.These items are summarized in Table 2.

Analytical Framework of Functional Damage Propagation
As an example of seismic risk assessment of lifelines under system interaction, a multi-system seismic risk analysis is discussed.This corresponds to dealing with problems pertaining to Item b in Table 2.The analytical procedure is schematically illustrated in Figure 7.It shows steps for risk assessment for a multiple system.Construction of event-trees of system interface in the stage of system modeling is a key step in dealing with system interactions.
As an illustrative example, seismic risk analysis for a water supply system linked with power supply was analyzed.Since electric power is essential to pivotal facilities of water supply systems, focus was placed on assessing and reducing the effect of loss of power supply.An illustrative example is shown in Figure 8 which is a plot of the probability of loss of water supply along a specific delivery path of pipelines.The horizontal axis is the distance from the water intake facility."basic model" represents the overall seismic risk of loss of water supply in the original system state.It is generally observed that the probability of malfunction increases with the distance from the source node.However, the probability of malfunction at hierarchically lowest five nodes are seen to be particularly large.These five nodes are located on hillsides and they rely on a booster pump which require additional electric power for which the contribution of the probability of powr loss becomes very high.In order to reduce the probability of malfunction of the water supply system, two kinds of measures against power failure were examined.One is (a) installation of additional power feeder line to reduce the probability of power failure, and the other is (b) installation of back-up power unit.Additionally, (c) construction of system interconnecting pipeline to upgrade network redundancy was considered.Figure 9 shows the effects of these measures on the improvement of system reliability.In this particular example, measure (a) and/or measure (b) are effective in reducing the risk to the level of 0.25 probability.If further risk reduction is needed, these countermeasures against power failure should be applied in combination with the measure (c) to improve the redundancy of the pipeline network.

General Framework of Lifeline Performance Criteria
Figure 9 shows a general framework of post-earthquake performance of lifeline systems.The upper part of the figure is based on purely engineering aspect of performance evaluation.Figure 9(1) shows that structural performance of a group of a large number of components that constitute fine lifeline networks are evaluated by using fragility curves, while critical components should be evaluated by event tree-, fault tree-and reliability assessment for individual subsystems.For the post-earthquake recovery phase, Figure 9(2) shows that "restoration curves" are common tools for overall post-event system performance evaluation, and engineering measures for supply to critical facilities under earthquake emergency are key factors for dealing with specific lifeline components.
A practical criterion is how to minimize the area above the restoration curve within constraints of resources for post-earthquake activities.Optimization of functional restoration works in this context has been discussed [lsoyama et al, 1985;Nojima and Kameda, 1992] as part of efforts for research aimed at efficient lifeline recovery planning for post-earthquake periods.It should be noted the target reliability for earthquake protection of lifelines cannot be determined only from engineering measures.Questions are often raised from engineers of lifeline sectors that a rational basis should be provided by which the post-earthquake performance of lifelines can be evaluated in a way that societal consensus or acceptance is realized.To do this, it is indispensable to incorporate evaluation of lifeline performance in an appropriate manner from the users' side.
Figure 9(3) is a conceptual illustration of a scheme to evaluate users' inconvenience under lifeline disruption [Kameda, 1994].In this context, we also use restoration curves, but we need restoration curves for various levels of lifeline services.Conventional restoration curves are normally defined as the time variation of the proportion of the served section in the entire service area, where the situation is judged from "served" or "not served."It should be noted that considering different levels of service will lead to different restoration curves.
Consider a case of water supply.We may classify the levels of water supply into the following typical cases.(a) 3lpcpd (litters per capita per day) is specified as a minimum standard of water supply under disaster emergencies.In the Japanese post-earthquake activities, local governments conduct emergency water deliveries aimed at supplying at least this amount of water.(b) I 0-1 00lpcpd is said to be a minimum requirement to be able to cook foods and prepare daily meals.This level is considered as a target of supply during the second stage of restoration.(c) In the third stage, water supply is aimed at delivering 100 -180/pcpd, which is not enough for normal activity levels but basic demands are met.(d) Then, finally, normal levels are to be recovered at which water consumption in Japanese urban regions is 200 -400lpcpd.These four levels may be typically defied as the corresponding service levels, level O -level 3, respectively, and corresponding restoration curves for these levels can be drawn as in Figure 9(3).They will be a comprehensive representation of the progress of restoration on the side of the lifeline sector.
While these restoration curves are an engineering measure, the users' perception of lifeline disruption may be represented by the time variation of the degree of acceptance as shown in the lower half of Figure 9(3).People can stand lifeline disruptions for some period immediately following the earthquake.But as a reduced level of supply is sustained, their demand will increase and the individual users' "acceptance curve" decreases with time, and converges to a certain lower level.The speed of decrease and the reduced level of convergence would be different for different service levels.It is emphasized that the acceptance curves should be understood as a measure not only of lifeline users' physical environments but also their psychological situations.Therefore, definition of the acceptance curves must be done on the basis of a rational combination of evaluations both from engineering and social psychology.
Once the restoration curves and the acceptance curves are defined for various service levels, the rate of acceptance, or the "overall acceptance" by the people in the entire service area, denoted by D(t) can be defined as the weighted mean of these curves; i.e.,

L, R; (t)A; (t)
D(t)=-; _ _ _ TP in which R;(t) = restoration curve at service level i, Ai (t) = (individual users') acceptance curve, and Tp = duration of the total post-earthquake period.The summation is performed for all supply levels at time t.
A schematic illustration of the overall acceptance curve D(t) is given in the right-hand side of Figure 9(3): a single curve that can be used as a comprehensive measure of evaluation of post-earthquake performance of lifeline systems.The area of the section above the overall acceptance curve D(t) and below the horizontal line of unity, denoted by S2, may be used as an utility index of lifeline performance incorporating users' evaluation.Applications to actual earthquake disasters are found in Kameda [1994).

An Experience in Kobe on Users' Difficulties in Water Shortage
As we saw in Figure 1, a very long period of time, 82 days, was needed for recovery of water supply in Kobe, 1995.The experience was quite severe if compared to the cases of other earthquakes.For this reason, the time needed for functional restoration played a role of a quantitative index for measuring water customers' difficulties coming from sustained water loss, which are basically qualitative or judgmental issues.An evidence this situation is reflected is seen in the telephone calls from individual customers to the Municipal Water Supply Headquarters that were made throughout the restoration period [Matsushita, 1999b].
There were a total of approximately 2,400 calls.The number of calls per day generally decreased with time, as restoration works progressed and the total number of people without water supply decreased.In contrast, the number of calls per day normalized by the number of customers without water remained nearly constant during the first four weeks of restoration period, and increased drastically after four weeks had passed.
This demonstrates that the people who had to spend more than four weeks without water had excessive difficulties of leading their lives.Indeed, the contents of conversation in these calls varied from those of inquirytype in earlier stages to strong complaint, anger and desparation in the post four-week phase.On the basis of this experience, the Kobe Municipal Waterworks Bureau has established a long-term reconstruction plan for its system in a way that emergency recovery to all customers, say the service level 2 as discussed in 4.1, should be finished within four weeks under any heavy urban earthquake disaster.Construction of redundant and robust transmission and distribution networks is the major feature of the reconstruction plan.
The example discussed herein would suggest that quantification of the users' acceptance level can be a feasible concept.By doing so and combining them with engineering technology, it is expected that reasonable framework and methodologies be established for general risk management of lifeline systems under earthquake environment.

Development of Consensus on Funding for Disaster Mitigation
Funding seismic upgrade programs for lifeline systems is a critical issue for executing the programs successfully and complete them within a required period of time.The question is "who pays and how?".An impressive example of resolving this issue is observed in the Seismic Improvement Program for the East Bay Municipal Utility District (EBMUD), California [Diemer, 1998].
As a wholesale agency supporting municipal water supply systems in the East Bay area of the San Francisco Bay Region, EBMUD, serving 1.2 million customers, has an important role of maintaining water supply functions under earthquake emergency.There are two seismogenic active faults crossing its service area: Hayword Fault and Calaveras Fault.
On the basis of experiences of the 1989 Loma Prieta Earthquake, EBMUD began its Seismic Improvement Program, consisting of seismic evaluation, development of improvement strategy, and implementation planning.A 10 year program was established with a total cost of $189.0 million.The program is composed of seven stages including reservoir upgrades, pumping plant upgrades, tunnel improvements, construction of a loop pipeline, transmission system upgrades and pipe fault crossings, building upgrades and equipment anchorage, and water treatment plant upgrade.
Besides engineering technologies incorporated in the program, it was a key issue to establish consensus for funding.They needed to find its financial resource by charges on the customers water bills.Extensive procedure was taken for public reach and education.The stakeholders were city councils, planning commissions, neighborhood groups, business groups and associations, service clubs, retiree groups, critical care providers, etc.Through these efforts, a favorable conditions were created, finally to be supportive to a charge of about $20 for an average customer per year for the next 30 years.
While technology developments in earthquake engineering is important, their implementation is equally important.Recent urban earthquake disasters continue to give us lessons by letting us notice a simple fact; i.e., earthquake engineering development does not immediately mean that the urban region is safer and more reliable.Implementation is a key factor.Efforts taken for this purpose, such as that introduced herein, should be highly recognized as indispensable steps for realizing safer urban regions.

Generations in Earthquake Engineering Developments
The history of earthquake engineering development may be outlined as shown in Table 3.Its first generation is represented by static design at elastic limit, the second generation was a movement toward the dynamic design and ductility design as well as liquefaction issues and lifeline earthquake engineering, whose effectiveness was essentially confirmed in the Great Hanshin-Awaji (Kobe) Earthquake Disaster.

Figure 1 :
Figure 1: Functional restoration of water supply systems in recent urban earthquakes Figure 5: Seismic design procedure for water supply systems (JWWA, 1997 Guideline)

Figure 6 :
Figure 6: Procedure of earthquake resistant design for railway structures [after Nishimura, 1999]

Figure 7 :
Figure 7: Flow chart of seismic risk analysis of lifelines under system interaction

Figure 9 :
Figure 9: Performance of lifeline systems under earthquake environment Their typical examples are found in the development of ductile iron pipes with earthquake-proof joints, construction of redundant networks, zonal block separation for isolation of damaged networks,