1Urban public services rely on a considerable amount of infrastructure which, over the years, becomes part of a city's historical heritage. Today, rather than new facilities, new techniques are needed to operate these networks and maintain them in good working order.
2These needs can be met thanks to new data processing technologies. However, the setting up of management strategies, the detection and identification of problems, the setting of priorities for repairs and replacements, have brought to light the fact that the real issue is not so much an operational one, though this too is important, as it is one of obtaining data and elaborating methods for analysis.
3A network is a series of pipes connecting various structures and usually leading to processing plants. It includes sewage pump houses, storm spillways, sluice gates. Data can be gathered from rain gauges, by measuring flow rates, heights, speeds, silt levels, and by analysing the nature of these flows. Mechanisms can be regulated through automatic systems or by remote control. The structure of the network may also be adapted to varying needs.
4The operator of a drainage network is responsible for implementing programmes to preserve both the natural and urban environments against damage wrought by floods and pollution.
5Thanks to a partnership between the Greater Lyons area and the National Urban Engineering Institute (Institut national de génie urbain), a research group was organized to study this issue. The objective was decision support for the definition of strategies for the configuration of networks to avoid unnecessary flooding or polluting.
6The use of the term "unnecessary" reflects the awareness that in many cases, flooding and pollution could be avoided. It also shows that policy-makers encourage innovation as a means to solve these problems. Innovation is in particular made possible today by advanced computer technologies able to process previously nonexistent or isolated data.
7Implementing these technologies means learning about their potential, how to design local automatic systems and elaborate global strategies for running and monitoring these instruments. Training may be based on observation, on the operators' previous experience and on numerical experiments. In a context where tools are few, where there has been no previous overall monitoring experience, the last solution is preferable.
8A simulator was developed to evaluate the different network management systems in rainy weather (Fig. 1). This simulator is made up of a series of models for the evaluation of several indicators (place and duration of overflows, quantity of overflow, masses poured into the natural environment, impact of these flows on the environment, deterioration of the infrastructure, network and plant operating costs, etc.)
9The next stage in the project, once measurements were made on a site in the Lyons region and models adjusted, consisted in defining simulation procedures. The simulations of different network configurations, which were carried out on the basis of rain data taken over a period of four years, show that it is possible to reduce pollution risks to a considerable extent without increasing the risk of floods. Figure 2 shows the results obtained for the indicator "rejected DCO (chemical need for oxygen) masses".
10These results can only be achieved if the right network configuration is chosen in time. In cases where rain can be forecast, static configurations seem to be a priori sufficient. If these solutions are considered too risky, it is necessary to set up a dynamic configuration which can be adapted during rainfalls.
11Despite these satisfactory results, we must not forget that many problems remain as yet unsolved:
12- owing to the complexity of the phenomena under study and concerns about pollution risks, research has concentrated on qualitative models. However, "comparable complexity" models should also be envisaged, since an evaluation of the water flows must take into account an evaluation both of volumes and concentrations.
13- more information must be obtained to explain impact phenomena.
14- there is a lack of fixed, standard procedures for measurement systems, model adjustments, numerical experiments, and training.
15- there is an ongoing controversy concerning rain forecasts, even short-term ones.
16These unresolved matters show that we must maintain a certain amount of caution in problems concerning decision support and choice of strategies for drainage networks during rainy periods.
17As far as maintenance is concerned, two main problems have to be dealt with: obsolescence (due to the emergence of new needs and new standards) and deterioration (due to age). All in all, 5000 kms of networks will have to be rehabilitated in the next few years. Given the size of the task and the medium and long-term means necessary to achieve it, research in this field is greatly needed to help define priorities, aims, and means. In other words, we must determine what should be rehabilitated, why, and how rehabilitation should be carried out.
18In big cities, coping with deterioration raises many problems linked to the availability of the necessary data. Given the total length of the network (700 km of walkable network for the Lyons urban area), an early diagnosis system based on specific risks was developed to take into account risks due to malfunction and the probable impact of the environment on the network, other than the deterioration of structures.
19Before drawing up a list of priorities in order of importance, a series of descriptors, models, and criteria were identified and selected. First, special forms for the description of the physical state of the network, to be filled out by the operating staff, were elaborated. The second stage consisted in analysing the extent to which the urban environment is vulnerable to the effect of the deterioration of environmental engineering structures (structural importance of a given section, derivability, impact of pollution on water gathering devices, etc.) The third stage is the study of the impact of the environment on the network; the disruptions caused by environmental factors and the conditions in which they occurred are evaluated; their effects are then described in terms of increased deteriorations. The necessary tools are now being developed. They are already capable of providing specific information concerning the state of the network. At the early diagnosis stage, data can be manipulated to provide assessment rules and simplified procedures. The real issue concerns the validation of results. This validation can only be gradually achieved by confirming or developing hypotheses and parameters on the basis of the analyses carried out on each series of measurements and probes.
20A parallel approach consists in adjusting simplified models to complex but reliable models, if available. Many research projects in civil engineering laboratories deal with, among other subjects, the influence of ground forces on the distribution of stress around galleries, the development of new materials, the structural reinforcement of underground ducts, etc. This research provides very important data for early diagnosis models and contributes to the advance of both curative and preventive maintenance techniques.
21The term "operating" mainly refers to flows, and that of "maintenance" to structures. Flow and structure management policies require a lot of data which is not always available.
22In order to study such questions as the physical laws governing the progress of deteriorations, impact modelling, etc., not only is it important that researchers cooperate with professionals, but also with other researchers from close, though in practice often quite separate scientific fields. As far as decision support is concerned, the notions of preference, choice, hierarchy of priorities, and risk management require more knowledge on the subject; so far, only part of the necessary data has been collected. Incomplete, uncertain and imprecise data, an inevitable element of the decision-making process, must also be accounted for. Other sociological and economic factors must also be considered and there too, learning by simulation - in the presence of the various actors - will serve as a basis for cooperation.