Southwestern France's Temple-sur-Lot Dam, a gate structure built in 1948, has been subjected to continuous alkali-aggregate reaction (AAR)-induced displacements since 1964, despite nonsignificant residual swelling test results and low and relatively constant alkali content in the concrete. It has been assumed that this long-term behavior could be explained through a substitution process between calcium and alkali in the alkali-aggregate reactive gel. The calcium substitution phenomenon cannot be detected through a conventional residual swelling test since it is very slow, so an original AAR kinetics and residual swelling capability assessment method is proposed. A laboratory test dealing with silica consumption kinetics is first involved in this methods, and a numerical finite element inverse dam analysis, including laboratory measured consumption kinetics, is involved as a second step in this method. At a given period, only one observed structural displacement rate fit the final swelling amplitude. Comparison between instrument point displacement predicted by the calculations (not used for the fitting) and dam measured variations provided model prediction capability validation. Finally, future dam displacement and damage field prediction calculations were performed.