System Control logic enhancements through Fluid-Mechanical valve dynamic transfer functions

Riccardo De Paolis

GE Nuovo Pignone


Turbomachines performances control is strictly depending from the interactions between the system control logic and devices dynamic (i.e. valves, drivers, actuators, etc…) which can act or promote modification of main fluid systems parameters over all working conditions. In order to seamlessly link the physical behavior of the system components, the functional instructions must be properly exchanged through the same language. This is possible describing the real components evolution with a system of equations that control logic can understand. Manageable input-output relations are often achieved from simplification and most of them are usually taken as they are, without any possibility of improvements. Approximations errors are inherited from previous applications and because of these, many situations arise where machine do not work as expected with the need of fine tuning in the field. In the worst scenario errors propagation in transient run may cause system response drift which cannot be recovered, with unknown consequences.
A similar issue arose in a Steam Turbine plant, where control logic became unable to manage the bulk steam control valve dynamic, resulting in output power oscillation.
This paper highlights how, with one-dimension CFD software such as FlowmasterTM, it is possible to improve the traditional approach on auxiliaries’ characterization through analytical multi physics system study. This will lead to confirm the goodness of current approximations or rather generate an equivalent transfer function of the entire system for additional studies.
In details, the previous mentioned steam control valve is modeled within Flowmaster software to properly reproduce item dynamic. Valve working conditions are strictly connected to feeding network behavior; the simulation will take into account these influences. Once model results are validated, advantages are expected in terms of simplification and other possible situations might be explored widely. Integration of the model in the control logic environment will make easily foreseen any related communication problems.