ABSTRACT

It is well known that the interpretation of tracer tests breakthrough curves can be multiple. It should be difficult to describe all the possible situations of wrong and inconsistent interpretations induced by ‘automatic’ calibrations solving the inverse problem without introducing any ‘hard’ or ‘soft’ geological data. When a 3D flow and transport model is applied to simulate tracer tests performed in 2D (depth averaged) conditions, a quasi infinite number of suitable parameters combinations allow to reach the calibration of the transport model. Geological data must provide informations on the different geological layers to be distinguished in the model. Inside these layers, the observed facies changes can motivate the choice of different values of the flow and transport parameters. Morphostructural analyse provides also informations on more fissured zones (in hardened formations) or on sedimentological features (in loose sediments), and shallow geophysical surveys provide new data or confirmation of the previous informations. In fact, one can not choose freely the parameter spatial distributions. Only few of the combinations are fully geologically consistent with all the collected ‘hard data’ and ‘soft data’.

Due to differentiated values of hydraulic conductivity, effective porosity and dispersivities distinguished in the different layers of a 3D model, computed breakthrough curves can represent any complex shape of the measured curves including double peaks, strong delays, etc.

Comparisons are made between the computed breakthrough curves obtained by calibration of a 2D model and by calibration of a 3D model for a same situation corresponding to tracer tests, performed in depth-averaged conditions, in the alluvial sediments of the Meuse River (Belgium). Conclusions in terms of under- and over-estimation of the interpreted parameters values are drawn.