La prévision de l'effet des aménagements fluviaux sur les nappes. Analyse de deux cas concrets

Abstract

River bed and bank improvement works frequently affect flow in aquifers related to the river. The major difficulty in investigating such effects is that they must be predicted to scales ranging from that of the actual site to that of the entire aquifer. Two examples are discussed, which illustrate typical problems and show how hydrogeologists are equipped to deal with them. A - LARGE-SCALE CANALISATION OF THE RIVER ESCAUT In order to convert the lower part of the river Escaut in France to large international size standards, diversion channels were required along a number of reaches and some locks had to be rebuilt, in addition to reshaping the river profile as a whole. Several schemes were compared with a view to reducing large civil engineering structures to a minimum (especially locks) whilst ensuring minimal effect of the conversion on flow in the chalk aquifer associated with the river over its entire course. Regional hydrogeology factors were decisive for the preliminary project in that this involved comparison of the effects of various proposed schemes on overall groundwater behaviour. Close accuracy of absolute values and detailed analysis of conditions at structural sites were not required. This comparison was made with the aid of a conductive paper model (Fig. 3). For the final project design, at which stage final structural data for the river sections was available from the preliminary design, however, a detailed study had to be made of the effects of stagewise implementation of the proposed engineering work on groundwater flow conditions, which were expected to last for several months, i.e. for long enough to require effective counter-measures. For this problem, models whereby both regional effects and local effects in the vicinity of the structures could be simulated with the same accuracy were an obvious requirement. A variable-mesh mathematical model developed by the BRGM for this type of study was used, which provided a general square mesh system allowing simulation of regional effects and, when subdivided according 10 requirements, simulation of local effects. Two versions of this model were used in order to ensure adequate representation of the hydrogeological conditions. One was a single-layer model covering the entire downstream part of the plain, and the other was a double-layer model of its upstream part, which was required because of the geological structure of the considered area (Fig. 5). The effects of development work on each river section were studied separately on these models, in which full use was mode of the possibility provided by the program of only subdividing meshes actually required in each case. Test conditions and the order in which the simulations were run are shown in Table I. Figs. 4 and 6 show examples of the results obtained, basically in the form of a map of areas endangered by groundwater emergence on the surface, with indications of probable drainage discharge requirements. B - DYKING SCHEME FOR AN INDUSTRIAL AREA This example relates to a smaller area and a different purpose, namely to dyke a river in order to provide an area for industrial development. The hydrogeological context is summed up in Figs. 7 and 8. The river floods produced overpressures in most of the general aquifer, which were expected to cause vertical flow through the more loamy surface formations. Suitable outlets had to be provided for this flow, the behaviour of which depended on ground altitude and configuration, and the approximate rates of flow required establishing as design data for the drainage system (sec Fig. 8). A programme of precise ground data measurements was put into effect, followed by two different design calculation procedures : - 1. Initial evaluation of piezometric level variations (i.e. uplift and flow emergence probability) in the considered area, as related to water level in the river Seine (Fig. 9). This was done on a very simple analytical model, neglecting inter-layer exchanges and assuming one-dimensional flood wave propagation at right angles to the river. 2. Simulation of groundwater behaviour on a two-layer mathematical model featuring an uniform square-mesh system. Piezometrie data and vertical flow distributions were determined for the following :- a) Various zone sizes. b) Natural ground elevation and banked up to 5-5.5 m above natural ground level c) Various flood levels. The overall model boundaries are shown in Fig. 9 and the results in Fig. 10. CONCLUSIONS By combined use of overall regional hydrogeological investigation procedure, test and field survey measurements and suitable simulation techniques (especially mathematical models), hydrogeologists can give river engineers reliable warning of probable major effects of proposed engineering works on groundwater flow.