Optimisation of Marine Boilers using Model-based Multivariable Control

Traditionally, marine boilers have been controlled using classical single loop controllers. To optimise marine boiler performance, reduce new installation time and minimise the physical dimensions of these large steel constructions, a more comprehensive and coherent control strategy is needed. This research deals with the application of advanced control to a specific class of marine boilers combining well-known design methods for multivariable systems.

This thesis presents contributions for modelling and control of the one-pass smoke tube marine boilers as well as for hybrid systems control. Much of the focus has been directed towards water level control which is complicated by the nature of the disturbances acting on the system as well as by low frequency sensor noise. This focus was motivated by an estimated large potential to minimise the boiler geometry by reducing water level fluctuations. Strategies for achieving such a goal, based on model predictive control, are suggested while pressure control is achieved by using a multivariable control setup.

The thesis further directs attention towards control of the boilers in load situations requiring on/off switching of the burner. A new strategy for handling such situations based on a generalised hysteresis control approach is presented. The solution is optimal according to a specific performance and the new method prove promising results compared to traditional methods and existing optimisation based methods (finite horizon model predictive control). In the thesis the pressure control is based on this new method when on/off burner switching is required while the water level control is handled by a model predictive controller.