 |
|
|
|
|
|
|
|
|
|||||||||
 |
|||||||||||||||||
|
|||||||||||||||||
|
©2000 - 2002 Dipl.-Ing. Tobias Jockenhövel. All rights reserved. |
|||||||||||||||||
 |
|||||||||||||||||
 |
|||||||||||||||||
|
|
|
|
|
[DAE 2 NLP] [OCOMA] [NLP Solvers] [Flexitime] |
|
|
|
|
The goal of dynamic optimization is to find the optimal control profile of one or more control variables or control parameters of a system. Optimality is defined as the minimization or maximization of a objective function without violating given constraints. For example: Find the optimal trajectory of the control variables in a power plant to start-up the plant as fast as possible (objective Min. time t) without violating the maximal thermal stress allowed in certain parts of the power plant equipment. Dynamic optimization can be applied to any dynamic system. However, the focus of this work are dynamic processes in chemical and energy engineering. A dynamic optimization problem in this field consists of a DAE system of the process, additional equality and/or inequality constraints, initial values and an objective function. Solving dynamic optimization problems directly (analytically) is not an easy if not impossible task for large scale systems. Several methods have been developed to solve dynamic optimization problems of which the method of the full discretization of state and control variables is used here. The full discretization of dynamic optimization problems result in a large-scale sets of nonlinear algebraic equations which require very powerful NLP solvers. If these problems are solved once, the term “Offline Optimization” is used. Subsequent optimizations scheduled periodically in real-time are called “Online Optimization”. The OptControlCentre is capable of doing both in a very user-friendly manner. The following links give more detailed information about the method used:
|