Last year significant earthquakes struck Izmit, Turkey, and Chi-Chi, Taiwan. Although the Turkish event had a higher cost in terms of lives lost and economic damage, the Taiwan quake offers multinational corporations, insurers, and reinsurers more insight into earthquake risks. A detailed investigation of the disaster provides many clues for improving risk management programmes.

Chi-Chi, at 1:47 am on September 21, 1999, was the largest earthquake to hit Taiwan in recent history. It had a magnitude of M7.6. The epicentre was approximately seven kilometres north-west of Chi-Chi, a small town bordering a mountainous resort 155 kilometres from Taipei. The duration of severe ground shaking was about 40 seconds, and was felt over the entire island. High ground accelerations, on the order of 0.5g to 1.0g, were recorded in the epicentral region. One of the most spectacular aspects of the earthquake was large vertical ground displacements along the fault rupture, with offsets measuring three to six metres in many regions. Nearly 15 metres of vertical offset occurred through the face of the dam at Shihkang.

In the ensuing five days many aftershocks occurred, several from magnitude M6.0 to M6.8, but the global economic aftershocks could have been much worse. Taiwan's high-tech industry produces much of world's supply of certain electronic computer components. It is concentrated in the city of Hsinchu, just outside the damaged area. Hsinchu and the companies located there came very close to experiencing major damage, and did suffer a near-total loss of power for about a week.

According to code

Taiwan has one of the most advanced earthquake building codes in the world, based on the codes used in California and Japan. Yet thousands of buildings, many of which were built in the last five to ten years, collapsed completely or were total losses. The primary reason was lack of code compliance due to poor engineering and poor construction.

Poor engineering played the much larger role, as many of the failed structures were designed without the required earthquake resistance features. A number had been constructed with “softer” ground floors with large store-front windows or open areas. The need for fewer columns and walls in these “soft-story” areas was not offset by more rigorous attention to seismic details of the few walls and columns. Collapses occurred in these weakened areas, allowing the buildings to pancake. It should be no surprise that buildings with these types of problems collapsed regardless of the quality of their construction. The ruined structures were designed and built without adequate inspection by the government authorities.

Taiwan construction practices mean that even if a building has been designed and built to code, there is no guarantee it will remain in compliance for long. A building owner has the freedom to significantly alter the framework of a building to meet changing needs. For example, walls can be knocked out, or bracing removed to make different interiors, without the requirement to obtain reviews and approvals from officials knowledgeable of seismic safety. A significant number of failures can be attributed to these post-construction modifications.

The shortcomings of codes

These are not unique problems. The same situation exists to some extent in virtually all the world's seismic regions, including the entire Mediterranean, particularly Israel, Greece, and Italy, the Caribbean, most of Asia, and parts of the US. However, in Taiwan a number of buildings that did meet code and had not been modified were total or near total losses, because modern building codes are intended to protect people, not investments. They represent the absolute minimum, legally-acceptable level of building performance. In other words, even for structures built truly to the code, the expected design-basis level of earthquake shaking may very well damage them sufficiently to make them total constructive losses, but if the buildings do not collapse and kill someone, the intent of the code has been met.

In addition, building codes do not address business interruption. That problem is left to the owner. If improved performance is needed or wanted, the owner must require a higher level of design. Nowadays this is called “performance-based design,” and is not legally mandatory. However, risk managers, CFOs, and facility owners who understand the limitations and the expected performance of structures built only to comply with modern building codes recognise the value of performance-based design.

Serious business interruptions were common after the Chi-Chi earthquake, even when buildings were undamaged, primarily because building codes did not require that manufacturing equipment, raised floors, suspended ceilings, and other “non-structurals” be protected. Very lengthy delays occurred when these items were damaged in clean-room environments.

The worst example occurred in the Science Park at Hsinchu. High-tech equipment failed at very low levels of shaking, levels which did almost no damage to the building structures. The most serious damage entailed breakage of quartz tubes inside vertical diffusion furnaces. Most of the silicon wafers contained in these quartz “boats” were cracked when they were tossed about during the shaking. There were many cases of process equipment shifting and disrupting crucial alignments, breaking attachments and fittings and the like. Sprinkler piping leakage caused heavy damage at some facilities, including damage to steppers and large sections of clean rooms. Had the shaking been even slightly stronger or of longer duration, damage and the consequent business interruption would have been much more extensive. Such damage to typical high-tech facilities occurred in the 1990 Philippine earthquake.

In high-tech industries equipment usually accounts for over 90% of the value of the plants, yet only minimal protection typically exists to guard equipment from earthquake damage. There are various excuses offered for not providing this protection; the need for easy access by maintenance workers, and the need to move equipment about as processes change are often cited. However, cost-benefit analyses consistently show that adding this level of protection is good business. Again, this is not a problem unique to Taiwan; it is also the norm in the rest of the world, including California and Japan.

In Taiwan business interruptions were the major contributors to financial losses for industry. These were often secondary losses, caused by damage to interrelated systems. For example, the earthquake damage to distant 345kV transmission towers and a switching station made it impossible for the Science Park to receive power from the usual steady source. Some facilities were able to maintain emergency power through the use of generators, but at least one wafer fabrication plant sustained a large loss when the generators failed after running continuously for 40 hours. With the loss of standby power the facility's fans were no longer able to maintain the clean-room environments.

Location, location, location

Hundreds of structures collapsed completely because they were directly on top of the faults. This is especially tragic, since their locations were known. Prudent property management would dictate the examination of available fault location data prior to construction, which should be the first step in a facility development or relocation project, as it is almost impossible to engineer safety into a building that straddles a fault. As is typically the case, much of the damage was exacerbated by soft soils beneath the buildings amplifying the ground motions. Unfortunately, many industrial facilities, and increasingly many commercial buildings, are located in such areas, including Silicon Valley and most major cities in Japan.

A commonly-held misconception meant many people were quite surprised at the location and severity of the event. Taiwan's seismic hazard is characterised in its building codes by three seismic zones labelled low, medium, and high. The earthquake occurred in the low zone. Unfortunately even the low seismic area of Taiwan, in absolute terms, is at least as seismically active as the most active locations in California. This misconception may have contributed to a more lax attitude towards seismic preparedness, and resulted in more damage as people and business were not vigilant in minimising their exposure.

The key lesson for earthquake risk management, learned many times before and once again from this earthquake, is that being built-to-code is not enough to prevent major losses. It fails to address business interruption and damage and loss to equipment, and does not address weaknesses in code enforcement, local interpretations, or local construction practices. Insurers and reinsurers which require more rigorous performance-based criteria for the risks they accept will realise significantly lower loss ratios over time.

The Chi-Chi earthquake has adversely affected production of computers and peripherals in Silicon Valley and elsewhere, as many critical components are manufactured in Taiwan. If the earthquake had been on one of the numerous faults close to or in Hsinchu, or near Taipei, the losses would have been staggering. BI losses alone could have easily reached $15 billion. Component shortages could have affected computer makers worldwide for six to twelve months. This clearly demonstrates the need for companies to understand the risks to their own businesses through the vulnerability of other businesses in their supply chain.

Results from investigations into damage at the Hsinchu Science Park indicate that the building and equipment vulnerabilities there were not too different from those existing today in other high-tech parks such as Silicon Valley, where the earthquake threat is just as great. The industrial losses, including those due to business interruption, were almost all avoidable. In this earthquake there were no significant lessons with respect to structural behaviour. Experienced structural engineers could have pinpointed and remedied most building and equipment configurations that led to damage and business interruption.

Improvements in hazard mapping and use of sophisticated catastrophe management software will soon enable Taiwan to improve insurance availability through risk-based catastrophe policy pricing, and software is already commercially available that can alert insurers and reinsurers to the magnitude and probabilities of the potential portfolio earthquake losses in Taiwan.

Peter I. Yanev is President of EQE International; Robert A. Philbrick is Regional Vice-President

of its subsidiary EQECAT, Inc. EQECAT has been engaged by Taiwan's Central Reinsurance Corporation to assist in a feasibility study for a voluntary residential earthquake insurance pool.