This paper describes laboratory-scale testing conducted to evaluate Selective Dissolution of cesium from non-radioactive Hanford tank 241-S-112 salt cake simulant containing the primary chemicals found in the actual tank. A modified axial dispersion model with increasing axial dispersion was developed to predict cesium removal. The model recognizes that water dissolves the salt cake during washing, which causes an increase in the axial dispersion during the wash. This model was subsequently compared with on-line cesium measurements from the retrieval of tank 241-S-112. The model had remarkably good agreement with both the lab-scale and full-scale data.