Selection Criteria

Because the investment required to exploit simulation software is significant, care should be taken in the selection criteria used. Following are some of the issues that should be considered:

Steady-State vs Dynamic

Steady-state simulators are currently the most commonly applied to process systems, but also limited in the possible returns. They are mostly used for:

Dynamic simulators have been used in niche applications for some time but are increasingly replacing steady-state simulators. They can be used for any of the steady-state applications listed above and in addition they can be used for:

Sequential-Modular vs Equation-Oriented

Sequential-modular simulators perform their calculations one block at a time in sequence, usually in the general direction of flow of material through the process. Block calculations generally require known input streams and known design parameters and calculate output streams. While such calculations are "traditional" they suffer a number of serious efficiency problems, recycles have to be handled iteratively and dynamics can only be approximated

Equation-oriented simulators assemble the equations from all the blocks into one large equation set which is solved simultaneously using differential-algebraic-equation (DAE) solvers. This approach has a number of advantages. The "equations" and the "solving" are separated that makes model writing much simpler. Recycles have no effect on calculation efficiency. Dynamics are handled efficiently especially if adaptive step-length integrators are used.

Custom vs Application-Specific vs General-Purpose

Custom simulators are purpose built for a specific process or equipment item. They are convenient but expensive and inflexible.

Application-specific simulators incorporate models for a range of equipment for specific processes, such as gas plants or wastewater treatment. They usually also incorporate the terminology and practices of the specific industry which is attractive to some users. They are obviously restricted in application and some do not allow the addition of user models.

General-purpose simulators are by nature the most flexible. However you may have to write all the models required or you may be fortunate enough to be able to buy or access a library of models for your process. Handling unusual equipment items is not a problem.

Model Architecture

With the exception of custom simulators, process models are usually assembled by connecting sub-models selected from a library. This is the most effective approach and GUI interfaces for assembling models make model construction convenient. It is highly desirable to be able to add your own sub-models to the library as processes generally have their own peculiarities.

Transparency

All aspects of the simulator should be transparent to the user, in the sense that you can see and understand everything that is happening. You should know what the model equations are, what assumptions were made, what parameters are built in, what is happening when the model solves and especially when it does not solve. Black-box simulators inevitably lead to frustration and are frequently shelved.

Numerical Solvers

The numerical solver is the heart of a simulator and efficient and robust numerical methods can make the difference between a used and a shelved simulator. Look for solvers where you specify an accuracy rather than a step size and which adapt the step size.

Hybrid Solvers

Modern processes are increasingly batch or combined batch and continuous processing. Simulators with hybrid solvers, those that can handle combined continuous and batch (discrete-event) models, can be a key selection criteria.

Optimisation

Simulators with optimisers or that can be simply integrated with optimisers should be considered an advantage. Not only is optimisation of process conditions an important application area, but data reconciliation and parameter fitting are often required in using and validating process models.

Data Presentation and Analysis

Convenient graphical presentation of data and simple exchange of data with graphics software and spreadsheets increases the usefulness of a simulator. The ability to drive a simulation from process data profiles is also an advantage.