Kaare Høeg, president of icold, set the scene at Dam Safety 98 (16-19 June, Barcelona, Spain) with a summary of the reasons why it is becoming more important to ensure dams are not only safe but are visibly safe. A good proportion of dams are now more than 50 years old, he pointed out; and a new generation of engineers has taken over responsibility for their safety. For these new engineers documentary evidence of safety practices is both informative and reassuring — as it is for society in general, for whom both the risks and benefits of dams are ‘imposed’. Høeg went on to note that in many cases there have been changes in river characteristics, whether upstream or downstream, and new flood estimates; even in cases where there have been few changes, there is a great deal of new data available to update the original safety case. Finally, he referred to the profound changes that are occurring in the dam and hydro industries. He pointed out that deregulation means that in some cases ownership of dams has been transferred, and with it the responsibility for safety, and warned that in cases where the dam was owned and operated by different groups there may be differences in their approach to safety. He suggested that in this rapidly changing industry, safety and environmental protection would be an important concern for the public, and would be likely to come under greater scrutiny.

Høeg stressed that the safety record of the dam industry was ‘good and improving’ but noted that there were still numerous small safety-related incidents and that the question of tailings dams had to be addressed. In his presentation he highlighted some of the initiatives that were being undertaken with regard to dam safety — themes that were frequently picked up by other speakers during the two days of paper presentations, and that were also important aspects of the tasks allotted to seven European Working Groups. During the conference the Working Groups reported on their progress in the following subjects: legislation; risk assessment; floods; uplift under concrete dams; roller compacted concrete; internal erosion; and geomembranes and other facing materials.

Risk assessment

Risk assessment, and its formalisation, are subjects much under debate in the dam industry, and questions over the usefulness and appropriateness of probabilistic techniques were addressed directly by Høeg and by J J Fry in his report from the risk assessment Working Group. Fry pointed out that engineers have always been required to assess risk: in the past, however, this was not formalised and was referred to by phrases such as ‘engineering judgement’. More and more, the assessment of risk is being quantified, and its components — the frequency or probability of an event and its consequences — are being considered separately. This allows the risk to be monitored, analysed and assessed.

Fry warned that risk assessment involved more than ‘modelling our ignorance’ and Høeg agreed; to show how it could be used to improve safety, Høeg offered some examples of fault tree analysis and explained how quantifying risk can allow resources to be directed most efficiently towards reducing risk at the most crucial points.

Questions over the use of probabilistic fault analysis and risk assessment often centre on the perception that quantifying risk offers a point of attack for dam opponents. Høeg pointed out, however, that ‘risk analysis does not create uncertainties; it exposes them’. This offers the opportunity to present risk as ‘credible, defensible and transparent’. He also pointed out that the Netherlands had used risk-based dam safety analysis since the 1950s. The growing acceptance of this useful tool was illustrated by the fact that it had been offered in recent ICOLD safety guidelines.

Trends and monitoring

Speakers throughout the meeting confirmed the importance of monitoring dams and of making good use of the resulting data. Høeg led the way by warning against complacency and noting that ‘rubber stamping’ codes was not the best way to ensure safety. Dams should be monitored not to check prescribed values but to assess trends in behaviour.

Discussing his paper on the Llosa del Cavall dam M Fernandez agreed that monitoring was important. He suggested that it could be carried out by specialist companies but agreed that it was the use made of the data that was the most important. Fernandez said that the designer should also be involved in long term surveillance of dam behaviour. The designer ‘has an interest far beyond the design phase,’ he said; he knows ‘not only the design details but also the new data incorporated, the assumptions made and the decisions taken during construction, which may not appear in the log book’.

Several speakers during the meeting supported initiatives to create common databases, on a national or international basis, to list safety-related incidents. The data would be saved using a common framework.

Safety responsibility

Where will responsibility for dam safety lie, at a period when the industry is rapidly changing? Speakers were clear that the basic responsibility belongs to the owner of the dam. In P le Delliou’s report from the Working Group on legislation this allocation of responsibility was a constant, although the philosophy and practice of dam regulation vary widely from country to country. A Goubet, chairman of France COLD, reiterated the position, but sounded a warning note: the body which has responsibility for maintaining the safety of a dam — the owner — should be separated from the body who has responsibility for licensing the dam. This point may become most apposite where both dam and licensing authority are in public hands.

Much discussion of the issues raised at Barcelona can still be anticipated. After two days of session, perhaps the 500 or so delegates felt keenly the responsibility of becoming the ideal safety engineer, laid down by the Austrian authorities and passed on by R Melbinger of Austria’s Federal Dam Supervision section. The dam safety engineer should be a civil engineer, should be experienced, should be able to consider both the whole context and the detailed problem, should be familiar with the structure, should have detailed knowledge of the technology, should have appropriate power and — last but not least — should be available when required.