Adaptive predictive control for the steam generator of a fossil power plant

In this paper, an adaptive predictive control scheme for the steam generator (or boiler) startup of a fossil power plant is discussed. This scheme is proposed in order to determine the fuel flow required to force the downcomer water temperature and the main steam temperature to track their respective references. The computed control actions are not applied automatically to the plant; instead of, they are displayed to the human operator as a recommendation. This is due to the fact that this scheme is to be embedded into an operation support system (OSS), which supports the human operator whenever he needs on-line assistance. The proposed scheme consists of two components: the first one performs a control function whose goal is to track the reference of the downcomer water temperature; the second one has a similar goal, but in this case the variable to be controlled is the main steam temperature. For both components, the control variable is the fuel flow; however, they do not operate simultaneously; the least one starts after the first algorithm has achieved its goal. When applied in a full-scale simulator of a fossil power plant, the main performance operation indices were improved with respect to a standard manual startup. The scheme also improves other non adaptive schemes tested in similar conditions by ourselves.