In this paper self-adaptive supervisory predictive functional control is studied for applications in a semi-batch reactor in which the optimal operation is to follow the reference trajectory without significant overshoot, and to minimize the overall batch time. The volume in the reactor varies, bending to increase with time, as does the heat-transfer surface inside the reactor. This causes variations in the batch-reactor dynamics which means serious problems with the reactor's core-temperature control. The semi-batch reactor is used in the pharmaceutical industry for the production of drugs, where quality control is extremely strict and it is necessary to follow the exact production recipe. The semi-batch reactor is an example of a hybrid process where there are some discrete on/off valves and a continuous mixing valve. The supervisory control is designed to cope with the constraints and the discrete actuators. A simulation study was carried out to investigate the possible use of the proposed self-adaptive control algorithm in a real application. (C) 2007 Elsevier B.V. All rights reserved.