Twenty years ago, producing a dynamic simulation usually meant writing your own program, including all the integration algorithms, and so was a very formidable task for anything other than small systems. This is no longer the case and a wide range of software is available to assist engineers in model building. The available software divides roughly into three categories: equation based systems; block based systems; and process unit based systems. There is, however considerable crossover between the categories.
Equation based dynamic simulation software expects users to enter their models in the form of equations. The software then takes care of calculation sequencing and usually provides a choice of several integration algorithms. This type of software is very flexible and gives the user considerable control in the structure of the simulation models used. The downside is that it is time consuming and error prone to formulate and then enter the equations for big simulation models.
ACSL ( Advanced Control and Simulation Language) was commonly used equation based system. It is actually a FORTRAN pre-processor which converts the equations the user enters into a FORTRAN program which is then compiled and linked. ACSL runs on a variety of computer systems from PC's to mainframes. Linking into data bases like PPDS is relatively straightforward, but does require some FORTRAN programming. Equation based software has now almost disappeared as a standalone product, although more advanced systems like Aspen Dynamic do include equation based elements to allow the coding of unusual systems.
This type of software allows users to build up models in a graphical form by linking together a series of blocks. The blocks might contain things like : an integration calculation; a function; or a look-up table. The big advantage of this technique is that it is easy to visualise the connections in the model, when the models are relatively small. As the models get bigger, however, the graphical representation starts to look like spaghetti, although most block oriented systems have means of creating sub-models and this makes things much clearer. Block based models suffer from the same disadvantage as the equation based systems in that the user has to have the model equations available in the first place. A block based language does, however, make it easier to maintain a library of previously developed sub-models which can be re-used as required.
There are several block oriented systems available on the market at the moment. VisSim, the system used for dynamic simulation in this course is one of the industry leaders (its main competitor is a system called Simulink).
At first glance this type of software seems to be ideal. With process unit based modelling, the modeller is presented with a number of predefined models. The idea is that all he or she has to do is to tell the system the dimensions of the equipment and the materials being handled and the model is complete.
In reality things aren't quite as simple. If the process being modelled pretty standard, then a reasonable predefined model is probably available and the system works as advertised. In most cases, however, the models aren't quite right for the process being modelled and need to be altered. Most of the unit based systems allow users to adjust models and even enter completely new ones. The problem in doing this is in trying to link the models you build with the 'general-purpose' models supplied. In many cases, the time spent doing this can be much greater than that which would be required to build the model from scratch using a block or equation based system.
Aspen dynamic is pretty much the chemical industry standard for this type of system, and the MEng students will be using it next term for a control design project. Hysys which is owned by the same company that sells Aspen is the standard for the oil industry. Although coming from different backgrounds, then two systems are now converging and Aspen users should be able to convert to Hysys quite quickly.