IN A RECENT publication of the international-commission-on-large-dams (ICOLD) on Benefits and Concerns about Dams (ICOLD, 1999), the following statements were given, which can be taken as a justification for a comprehensive dam safety evaluation and rehabilitation project:

• To ensure the continued and dependable delivery of benefits from a dam, the owner must have a comprehensive plan for operation, maintenance and rehabilitation. As dams become older, safe performance becomes a concern. This requires more attention in the form of inspections, evaluations, modifications and upgrades of the older dams so they meet current technology, statutes and regulations.

• Dam safety activities include monitoring structural performance, developing emergency action plans, training of dam operators, exercises involving the local officials and population and implementing risk reduction actions.

As residential and commercial development expands in a river basin, the hydrologic and earthquake risks of existing dams grow rapidly. Risk is estimated by the product of the probability of occurrence of an adverse effect and the consequences (ICOLD, 1999). To limit the risk of large storage reservoirs, structural and non-structural measures have to be implemented. As modifications in existing projects are very costly, it is necessary to identify those projects, which are deficient in their capability to sustain the loads arising from extreme events and which therefore represent the highest risk. Several studies have shown that the greatest risks from the natural environment derive from the hydrological and seismic hazards. Hydrological safety problems are on one hand caused by man-made changes in the hydrologic characteristics of the river basins, and on the other hand by changes in the hydrological design criteria, which nowadays require that a large flood (probable maximum flood (PMF) or 10,000 year flood) can be released without the overtopping of a dam.

During the past two decades, hydrological re-evaluations have been carried out for many dams and, as a consequence, the spillway capacity of several dams has had to be increased to cope with these new hydrological safety demands. As floods occur quite frequently, it is relatively easy to convince the parties responsible for the safety of dams to support a programme for the reassessment and upgrading of the hydrological safety of the existing dams. Nevertheless, for the design of cofferdams, return periods of design floods are sometimes rather short. For example, at Rogun dam site in Tajikistan (above, right), blocking of one of the diversion tunnels, debris flows from upstream tributaries, and an extreme flood led to the overtopping of the 40m high upstream cofferdam, which was conceived as part of the 335m high main dam (Sirozhev et al. 1993)

In the case of earthquakes, however, convincing the same people of the need for the seismic re-evaluation and subsequent upgrading of their dams has turned out to be a rather difficult task. This is particularly true in countries or regions with moderate to low seismicity, where destructive earthquakes have return periods of hundreds or even thousands of years. This is a well-known problem all over the world. Nevertheless, in regions of high seismicity, earthquake safety of existing dams is a major concern. For example, in the 1990s a comprehensive seismic reassessment and upgrading of the existing dams was carried out in California. The seismic safety of the 1200 dams under the authority of the State and 175 privately-owned dams was evaluated (Babbitt, 2003). As a consequence, safety improvements at 116 dams – mainly embankment dams – were necessary. As a temporary measure, storage restrictions were imposed (i.e. lowering of the reservoir) until permanent improvements were implemented. The structural improvements included: increase in freeboard, widening the crest, flattening the shoulders, or replacing unsuitable materials in the foundation. Several cases of seismic safety upgrading of dams are discussed in Wieland (2003) and in Vol. 3 of the Proceedings of the 21st ICOLD Congress in Montreal. Seismic strengthening is needed where dams have been built on liquefiable foundation layers, which were ignored at the time of construction and in cases where the seismic actions have been grossly underestimated or ignored at the time of construction.

Dams are built to store water for use in irrigation, power generation, water supply and flood control. Therefore, once a dam project has been completed, the safe operation of the dam is concerned with water resource management and efficiency of the hydro and electro-mechanical equipment. These are the tasks of hydrologists, hydromechanical and electrical engineers, i.e. people who have a prime interest in the hydrologic safety of the dam and in the uninterrupted safe operation of the facility to meet the demands of the population in the downstream area.Therefore, despite the costs involved, they do not question the need for improved hydrological safety.

But in the case of the earthquake hazard, there exist many reservations and even prejudices especially in regions of low to moderate seismicity. Here, the same people who support the hydrological upgrading oppose or seek arguments against the seismic reassessment of their dams, or try to delay such investigations, as the probability of occurrence of a strong earthquake is often very small. These people lack the ability to visualise and foresee the consequences of a strong earthquake striking their facility. Earthquake, structural and geotechnical engineers, who have a direct interest in the seismic safety of a dam, are usually not involved in the team assuming responsibility for operating the dam, i.e. structural safety and geotechnical engineers are insufficiently represented as compared to engineers of other disciplines, a fact that results in the situation explained above. These arguments shall, however, not be taken as an excuse for neglecting the seismic risk but as a motivation for increasing the public awareness in the seismic safety of dams.

The nuclear industry takes a very serious view of earthquake safety. For practically all nuclear power plants, the seismic action is the governing design load case. Nuclear power plants must be able to cope with the so-called Safe Shutdown Earthquake (SSE). This is an event with an average return period of 10,000 years. Large dams have to resist the so-called Maximum Credible Earthquake (MCE), which is an event that is at least as strong as the SSE. Therefore, it is obvious that if the current seismic design criteria for dams are followed, it is very likely that the seismic action would also become the governing load case for the design and safety assessment of most large dams. But the fact is that older dams constructed before, say 1970, were designed for earthquake resistance by means of a seismic coefficient, k, which represents a fraction of gravity and, when multiplied by the weight of the dam, gives a force (in horizontal direction) representing the action of the earthquake. This approach is usually called pseudo-static analysis. It is still practiced today in regions of low seismicity and for dams with a very low risk. The value for k is often assumed as 0.1 corresponding to an acceleration of 0.1g. However, recent investigations of the maximum seismic event in low seismic regions have demonstrated that the MCE can be considerably larger than originally assumed at the time the dam was designed and that the k-value may well be too low. Therefore, with the seismic information available today, some of these dams may not have sufficient resistance to withstand the MCE pertinent to their area. This is an important new fact that cannot be ignored in dam safety issues.

This clearly demonstrates that in the coming years, public awareness has to be focused on the seismic safety of existing dams. This awareness is also needed to counteract the still growing number of action groups opposing the construction of new large dams.

Therefore, as far as dam safety is concerned, an impeccable record on the condition of all the components that could be affected during an earthquake is a prerequisite to fend off any criticism.

Seismic safety

Quality of design and expected damage

Up to now no incidents have been reported in the literature where people were killed due to the failure of a well-engineered dam during an earthquake. However, this does not mean that dams are inherently safe against earthquakes. For example, during the Bhuj earthquake of 26 January, 2001, in Gujarat Province, India, 245 earth dams were severely damaged and needed repair and/or strengthening. Because the reservoir levels were extremely low during the time of the earthquake, no catastrophic release of water took place from the reservoirs of the severely damaged dams (left, top).

It has to be recognised that many of these earth dams in the Bhuj region had a height of less than 30m and were built mostly by local communities and organisations with little experience in dam design and construction. Hence, these were probably not well-engineered dams and their seismic and also hydrologic safety is unknown and may have been deficient in many cases.

Moreover, there are very few large well-designed dams, which have been exposed to ground motions that may be expected during the MCE. Sefid Rud dam in Iran was subjected to an earthquake which may have been very close to the MCE, namely the 1990 Manjil earthquake.

It suffered considerable damage but resisted the earthquake in that there was no catastrophic release of water although the reservoir was almost full (left, bottom). The dam was designed pseudo-statically with a seismic coefficient of k=0.25 (Indermaur et al., 1991).

Regulations for seismic safety of dams

Earthquake regulations for buildings exist in most countries. However, they apply to new structures only and do not require an upgrade of existing structures where necessary. The same is true for dams. Since the majority of the older dams were built using methods of seismic analysis and seismic design criteria, which today are considered as obsolete or outdated, it is not known whether an old dam complies with current seismic safety guidelines published by ICOLD (1989).

According to the current ICOLD guidelines, large dams have to be able to withstand the strongest ground motion that could occur at a dam site due to the MCE. Withstanding an MCE-level earthquake means that the dam may suffer a certain degree of damage but this must not lead to an uncontrolled release of water from the reservoir. It must be pointed out here that uncontrolled release of water can also occur from slope failure into the reservoir in the vicinity of the dam. Large volumes of soil and rock can generate a wave surge that can overtop the dam causing extensive damage.

Risk analysis

Recent approaches to quantify dam safety made use of risk analysis and applied it to several dams in industrialised countries. They have shown that the failure of a large dam, and the resulting flood wave from the uncontrolled release of water, can cause a large number of casualties and huge economic and environmental damages exceeding billions of US$. It is highly important to know whether earthquakes can produce damage on certain older dams, which could also lead to flood disasters. Deficiencies in dams that could lead to such events are often not known previously but must be detected from a thorough seismic evaluation of the dam structure and its foundation. The statistics on dam incidents show that quite a number of deficient dams fail during the first few years after construction but the causes are not of seismic origin.

Design ground motions

The trend in seismic analysis of dams becomes more conservative; it goes towards higher intensities and stronger ground motions. The earthquake ground motion is usually characterised by a peak ground acceleration (PGA), which is different from the pseudo-static acceleration expressed by the seismic coefficient, k. If a dam in a region of low seismicity was designed pseudo-statically with a coefficient k= 0.1, but recent analysis of seismic data led to a PGA of, say, 0.5g (from an MCE of about 6), the dam may in fact suffer considerable damage. (The pseudo-static acceleration of 0.1g may correspond roughly to a PGA of 0.15g). Old dams may be overdesigned and the dam would probably not fail completely so the consequences would remain small. However, these aspects must be investigated such that the dam’s safety can be assessed and necessary measures taken.

Hence, in areas where the level of the PGA during an MCE has increased significantly, the earthquake safety of most existing dams is unknown and some may even be unsafe.

If a dam is found to be unsafe, then the easiest way to comply with safety standards would be to lower the reservoir level or to decommission a dam. Because there are very few viable alternatives to dams in many developing countries, decommissioning of dams or lowering the reservoir level would be the last resort.

Dam safety and social responsibility

The failure of a dam nearly always causes loss of human life and inflicts great damage to infrastructure in the downstream area. Agricultural fields may not be fertile for several years. A storage facility whose structures show signs of distress that may eventually lead to failure needs corrective measures to satisfy the requirements for safety. Dam surveillance is an activity that enables an assessment of the seriousness of any deficiency. A well-designed and carefully constructed dam is not likely to give rise to major corrective actions during its lifetime and the general maintenance cost can be kept small.

The dam owner has an obligation to prevent foreseeable failure scenarios by applying suitable corrective measures or otherwise abandon the facility. For example, inadequate flood handling facilities can lead to overtopping of the dam, which in case of embankment dams is particularly serious because rapid erosion can lead to a breach. Similarly, if the alluvial foundation of an embankment dam has revealed layers or lenses of sandy soils prone to liquefaction during, say, a magnitude 6 earthquake, the foundation needs to be treated.

Much depends on the dam safety legislation and third party liability. For example, in the US, a victim can ask for very high compensations if the owner of a project has been negligent. As insurances do not cover gross negligence, no special legislation is needed in such a legal environment as the dam owners may face very high fines and, therefore, are interested in owning a safe dam.

In most countries the aforementioned self-regulating forces and mechanisms do not exist. There are also problems when dams are owned by government agencies. They cannot be threatened by lawsuits and, therefore, may neglect maintenance and rehabilitation works. But regardless of who the owner is, it is their responsibility to ensure sustainability in enforcing safety rules and protect people, property and the environment from damage caused by natural hazards.

Benefits

Benefits from a constructed facility can be in the form of monetary gains. This is the most welcome form of benefits, for example power generation, water supply, irrigation, although the actual profit to an investor may be small (except for hydro power projects). Many dams are of the multipurpose type; they are a source of water for irrigation, for domestic and industrial use, power generation, and fishery. They may also have some recreational benefits. All these functions improve the quality of life of a large number of people and provide additional income. However, some of these benefits (cleanliness of the water, recreation, etc.) are difficult to quantify in monetary terms because revenues can usually not be collected from those who benefit from these amenities.

What are now the benefits from implementing a dam safety programme to a storage facility? How can investments be justified? A dam safety programme has the objective to guarantee sustainability in development and to help mitigate risks associated with deficiencies in the dam and its foundation such that it can cope with severe natural hazards.

An investment for the purpose of corrective measures to improve and ensure safety of a facility is similar to coverage by insurance but the concept is different. With insurance cover, the owner accepts the probability of failure or destruction of a structure (for example, by fire) and pays a yearly premium. This investment guarantees a partial or full compensation of the economic losses incurred during the disaster. The impact of the failed structure on the environment usually remains small.

With a dam, the impact of a failure is not restricted to its immediate vicinity and may extend over tens of kilometers downstream. The investment into dam safety aims at decreasing the probability of a failure to a very low level or in other words, the event of failure or any other serious incident is most unlikely to occur. This represents an economic benefit, not in terms of financial gains but in terms of preventing the loss of present day benefits generated by the operation of the dam facility.

Hence investment in dam safety measures, i.e. implementing corrective measures necessary for the continued safe operation of the facility, does not bring any form of increased revenue, but it prevents the presently achieved revenue from being lost because of detrimental natural hazards, mainly floods and earthquakes, which have a low but real probability.

Nobody is willing to invest in an area that is exposed to possible large floods. For example, spillway channels with discharge capacities way below the PMF discharge can be considered a hazard for nearby residents. In relatively flat terrain, it may not be possible to escape the flood in time.

The probability of occurrence of strong seismic events in areas of low tectonic activity is in the order of several hundreds or thousands of years. But also other events (e.g. piping through a foundation) affecting a dam’s safety have a low probability of occurrence. Quantifying such events and deriving economic benefits is extremely difficult and will involve not only a great deal of judgement but also substantial uncertainties. Rare events have to be included when safety is being addressed and loss of life is a possibility. But any economic analysis becomes questionable and (emotional) arguments raised by social action groups cannot be answered satisfactorily by technical people.

If word goes around (and it will) that a storage facility may have safety problems, either from deficiencies in the existing structures (embankment dam with potentially liquefiable foundation soils, spillway with inadequate flood handling facilities, or a concrete dam with insufficient resilience to absorb the earthquake-induced strains, etc) the anti-dam lobby will be quick to mobilise public opinion against further operation of the storage facility. Such actions cost the owner valuable time, possibly legal fees and may even damage their reputation as a conscientious dam operator. It will certainly affect their overall economic performance.

A storage facility that has been identified as unsafe, either for seismic (structural) or for hydrologic reasons, may be required to operate below full storage level until the deficiency has been rectified. If this is the case, the revenues are reduced, i.e. less water for irrigation, power supply, etc. Hence, if the facility is brought back to acceptable safety standards, the investment will certainly generate an economic benefit.

A quantitative value of the benefits can be calculated based on a risk analysis. Such an analysis would have to be carried out for each dam. Simple models would have to be defined and agreed upon first, before proceeding with the analysis. Such models will have to include a number of debatable assumptions, for example, the value of a human life, and one has to realise that the accuracy of the estimated benefits will be limited. This is because there are often insufficient reliable good quality data to estimate realistically the required probabilities. Uncertainties in the order of factors of two to five can be expected if different models and assumptions are considered.

If one is only interested in the relative benefits between different dams, then simple risk models can be suitable and the corresponding effort is limited. However, if we want to achieve a high degree of accuracy then cumbersome and expensive cost-benefit and risk analyses have to be carried out.

Conclusions

  • There is a lack of seismic safety awareness among dam owners particularly in regions of moderate to low seismicity. Seismic safety issues must become part of every dam safety programme. The objectives of a dam safety programme are sustainability in development, safety of the people downstream of the reservoir and proper functioning of all the vital components of a dam and its appurtenances.
  • There is a need for the re-evaluation of the seismic safety of older dams, i.e. those which were designed using pseudo-static analysis and where the maximum level of ground shaking expressed by the MCE has been underestimated. A re-evaluation consists of a state-of-the-art re-analysis and a visual inspection of the dam structure with emphasis on possible damage and failures that could occur during the strongest earthquake in that region.
  • Investments in dam safety measures, including monitoring and surveillance, will not increase the revenue derived from a storage facility, but they will minimise the probability of failure and thus will sustain the level of economic benefits which otherwise may decrease drastically in the event of flood or earthquake.
  • Dam owners, both state enterprises and private share holding companies, have clear social commitments. It must be realised that investment in the safety of existing dams is also a social obligation and not a mere commercial investment.