
	[About Me] [Ph.D. Project] [Schering Price 2000] [Picture Page] [Disclaimer] [Statistics]

Description of doctoral research project of Tobias Jockenhoevel:

“Dynamic Optimization of complex power plants and chemical processes.”

The focus of this work is research on optimal start-ups and load shifts of power plants and chemical processes in real-time. The optimization methods include SQP Active Set and Interior Point methods as well as a stochastic approach using Simulated Annealing. In the course of this work, a user-friendly software package for dynamic optimization, the OptControlCentre (OCC), was developed. The OCC is based on MATLAB and has a powerful graphical user interface (GUI).

Dynamic or stationary nonlinear optimization can be conducted with the simultaneous optimization of both control profiles and control parameters. OCC can also be used for dynamic System Identification. OCC has a couple of powerful features such as Automatic Grid Refinement, Flexible Time Elements and Auto Integration Error Determination.

The OptControlCentre can conduct fully automatic optimizations of dynamic systems in real-time (Online Optimization). The state of the plant is periodically identified by a nonlinear dynamic system identification which provides estimation values of unmeasured system variables. System Identification (with or without noise) is either done for the entire dynamic system or for various subsystems in a parallel approach, which is much faster. The system identification is followed by a nonlinear dynamic optimization which provides updated optimal profiles of the control variables. The course of the online optimization can be supervised by displaying the system variables. The user can conduct studies on scheduling, multiple objectives and fault detection.

It is often desired to find optimal control trajectories and control parameters without intensive modifications or rewriting of existing simulation models. To accomplish this task, a stochastic approach using Simulated Annealing was developed. Random control trajectories are generated and dynamic MATLAB/Simulink models are used as “Black Boxes” to determine the objective value. The algorithm was modified to allow simultaneous global optimization of control profiles and time-invariant control parameters. The offline solutions provide excellent starting values for the online-optimization.

The industrial partner for this research project is SIEMENS Power Generation (KWU). This research project is funded by a personal Ph.D. scholarship from the Ernst von Siemens-Foundation.