(486y) A Model Generator for Process Simulation and Optimization

Simulation allows the prediction of the behavior of a process under determined conditions, the result being the output variables for a given set of input variables and process parameter data. Basically, the idea is to build a square set of equations without any degree of freedom. There are two types of simulators, sequential-modular and equation-oriented (Biegler, 1989)

Sequential-modular approach allows solving the problem following the flowsheet structure. This approach suggests that a partition of the system and a definition of the precedence order for the calculations are required, thus solving equipment one at a time until the whole process is simulated. This is the way commercial process simulators work, HYSYS (Aspentech, 2009a) and the sequential-modular approach of ASPEN (Aspentech, 2009b). The user has access to a built-in library containing several equipment, that can be used in a straightforward way to simulate a process from simpler to very complex ones. The drawback is that the user has no access to the models themselves, then they cannot be modified, and the only way of adding user-defined processes is to write them down from scratch, with considerable effort.

Another drawback is the slow convergence of highly integrated processes, or the optimization, provided that the input data or parameters are not known and have to be calculated.

Equation oriented approach, on the other side, provides a model of the whole process where all the equations are solved for all the variables in a simultaneous way. The whole set of equations and variables are generally available to the users. New models can be added, existing ones can be modified or specifications changed in a generally straightforward way. There is no need to provide the main program with a subroutine with the system of equations and a calculation routine when new equipment is added. Highly integrated processes, of processes with recycles, can be solved more easily in an equation oriented environment. Optimization is also benefitted with the equation oriented approach.

Commercial software like GAMS (Brooke et al, 2004) utilizes the equation oriented approach for solving large sets of equations. The drawback is that the user has to provide the whole model, including thermodynamic calculations, task that must be carried out carefully in order to have a proper model of the whole process.

Having in mind both approaches, this work presents a model generator software. The user can add any plant, selecting one by one the equipment in an object oriented environment, obtaining the mathematical model representing the whole process that can be embedded afterwards in a equation oriented solver. The whole set of equations becomes available to the user, without the need of introducing the model equations one by one, because the software automatically make the whole set of equations available when selecting the equipment. The model can be easily adapted to the syntactic requirements of the solver used.

Previous experience of the research group with the development of instrumentation design software MODGEN (Vazquez et al, 2001), in this work it is proposed a redesign of this software. As a result, it will be obtained a user-friendly software oriented to process simulation and optimization, with intuitive usage features, that can be summarized as:

?Complete visualization of process plants.

?Object-oriented programming

?Rigorous mathematical model, as well as physical and chemical properties of the streams entering the process

?Model syntax easily tuned to any equation oriented simulator

?Flexible modeling options

?Consistency check and data validation

?Tools to simplify the input of plant topology

?Capable of dealing with industrial size plants

?Development tools for software developers, with precise instructions on how to include models and thermodynamic packages, as well as guidelines for the modification of existing models.

The object oriented approach used for designing the model generation software gives flexibility to the program, so the user can include any equipment model or thermodynamic package. The syntax was developed in XML language (Raik, 2001, Harold and Means, 2004) and it is clearly shown in the paper.

Each equipment and thermodynamic model is treated as independent block, which is included to the program interface.

The models generated with this software were tested with software GAMS (Brooke et al, 2004).

References

Aspentech, HYSYS, (2009a). Available at http://www.aspentech.com/hysys/

Aspentech Home Page, (2009b). Available at http://www.aspentech.com.

Biegler, L., (1989). Chemical Process Simulation. Chemical Engineering Progress, pp 50-61.

Brooke, A., D. Kendrick, A. Meeraus, and R. Raman, (2004). GAMS: A Userxs guide.

Raik, E, (2001). Learning XML. Ed. OxReilly.

Vazquez, G.E., I. Ponzoni, M.C.Sánchez, and N.B Brignole, (2001). ModGen: A Model Generator for Instrumentation Analysis. Advances in Engineering Software, 32, pp 37-48.

Harold, E., W. Means, (2004). XML in a Nutshell. Ed. OxReilly.