This paper presents the analysis, design and implementation of a model based control strategy on a copper cementation process. The cementation process is a part of the electrolyte purification section at Falconbridge Kidd Creek Division, Ontario, Canada. At Kidd Creek, zinc is produced by an aqueous electrowinning process. Before the electrowinning, the zinc solution is processed in an electrolyte purification section where impurities like copper, cobalt and cadmium are removed. In the cementation reactor, zinc dust is added to the solution to reduce the aqueous copper to metallic copper. The cementation process is in principle simple, but analysis show that there is a recognizable varying stoichiometric efficiency and a transport delay in the zinc dust that is added. The measurements are also sparse and delayed which turns this process into a challenging control problem, The cementation reactor is modelled as a dynamic CSTR and the identified time delays are incorporated in the model. This model runs on-line in parallel with the process, and an algorithm estimates the stoichiometric efficiency factor of the process. This parameter is used in a feed forward control strategy. In addition a feedback compensation is computed from either the measured or the predicted effluent copper concentration. This strategy has shown to be able to compensate for the severe disturbances in the process, and hence decrease the variability of the process.