Dynamic optimization toolset for efficient tuning and optimization of simulation models
Development of the simulation model corresponds to roughly 40% of the total time required for delivering an
industrial operator training simulator. A big part of this work is targeted to validate and tune the model. This
is required to increase the fidelity of their results. It is a procedure in which the model parameters are
iteratively adjusted until the behavior that best represents the physical system is found. This laborious task
requires domain expertise to select and adjust specific model parameters.
Process modeling requires much knowledge on process, chemical, physical and control. Even though skilled
engineers create models, it takes time to tune the models to improve their fidelity.
Semantum and Omega simulation have collaborated to develop a model optimization toolbox to
efficiently optimize and tune simulation model parameters.
The optimization function calculates the steady state optimal parameter values like cost minimum or benefit
maximization. Similarly, these functions calculate the dynamic optimization like transient state optimization or
minimal time control
with process constraints. These constraints are called simulated constraints. The optimization functions can
also be used for finding optimal PID control parameter.
On the other hand, the tuning functions are to tune systematically and automatically using multi-parameter
optimization methods. The multi-parameter optimization is based on fitting simulation results with the available
measurement data of the target system. This critically reduces time needed for model tuning and can be used to
find optimal parameter values that fit both steady state and dynamic behavior.
The co-developed tool reduces the tuning cost and contributes to develop high-fidelity process model.
The optimization algorithm was developed for tuning the process model and both steady state and dynamic optimization of plant operation. The tool can efficiently determine
the optimal parameters with only a few iterations.
The optimization toolbox has been integrated with Omega Simulation’s product Visual Modeler and some of these
features are planned to be included in the product as an optional function. Visual Modeler is a next-generation
plant simulation tool developed for delivering operation training solutions, for developing control systems, and
for creating online simulation systems for operation support.
Both steady-state and dynamic optimization have been achieved. The high-fidelity model for dynamic process
response is particularly crucial for plant operation and control. Additionally, there are no restrictions on
the process model, as the toolbox functions with the process simulator as a black box.
The optimization toolbox is a comprehensive tool set that aims to reduce development time and effort for model
parameter tuning and finding optimal operation. This results, for example, in shorter development of operator training simulators and other
simulation-based applications. It increases the level of fidelity of
simulation models and plant dynamic optimization.
