Adaptive predictive control of superheated steam and economic performance

Rui Neves-silva, DEE Group Author

Research output: Chapter in Book/Report/Conference proceedingChapter


This chapter addresses the problem of superheated steam temperature regulation in thermoelectric power generation units. The link between improving control quality and economic performance is explained together with a controller that achieves this purpose.
In thermoelectric power generation units the economic performance is directly proportional to the superheated steam temperature. There is however a threshold that should not be surpassed under the penalty of increasing the risk of damage and reducing the life time of the plant. Since regulation is not perfect, the temperature set-point should thus be selected low enough so that the fluctuations do not overpass the safety limit. With an improved controller, the fluctuations are reduced and the set-point may be increased, while keeping the same risk (i. e. the probability of surpassing the threshold). This results in an increase of the plant efficiency and therefore of its economic performance without a further reduction in plant remaining useful life.Controlling the superheated steam temperature poses a number of significant difficulties such as the existence of a long delay that varies with power load and an high degree of uncertainty in dynamics. In order to meet these difficulties, an adaptive model predictive controller is described. Due to its predictive action over an extended horizon, this algorithm is able to tackle the problems related with varying input/output transport delay. A distinguishing feature consists in the fact that the different predictive models upon which control synthesis rely are separately estimated from each other. This introduces a redundant feature that improves the ability of the controller to tackle un-modeled dynamics and moderate nonlinearities.The approach proposed is demonstrated with actual plant experiments in a boiler able todeliver between 100 and 150 tons of steam per hour, both for electric power production and industrial usage. The experiments shown include the effect of algorithm parameters,response to disturbances induced by load changes and the superior performance with respect to a control structure based on the cascade of two carefully tuned PID controllers.
Original languageUnknown
Title of host publicationPower Plant Application of Advanced Control Techniques
EditorsPal Szentannai
Place of PublicationVIENNA / Austria
PublisherProcessEng Engineering GmbH
ISBN (Print)978-3-902655-11-0
Publication statusPublished - 1 Jan 2010

Cite this