Tools for Computer Aided Analysis and Interpretation of Process Simulation Results Marion ALLIET GAUBERT and Xavier JOULIA Institut National Polytechnique de Toulouse (INPT) Ecole Nationale Supérieure d’Ingénieurs de Génie Chimique (ENSIGC) Institut du Génie des Procédés (IGP) Equipe Analyse Fonctionnelle des Procédés 18, chemin de la Loge - 31078 TOULOUSE Cedex tel. : (+33) 5 62 25 23 55 ; fax : (+33) 5 62 25 23 70 ; e-mail : Xavier.Joulia@ensigct.fr 1.INTRODUCTION The use of process simulators is still a complex task for process engineers. It requires know-how, considerable handling and a good knowledge of modelling and simulation. The user has many problems to face: poor communications abilities at the pre-processor level, such as global process representation from predefined models; at the executive level, data transmission hazards (from simple copy-paste to simulation parameter estimation from raw data), and potential convergence difficulties; and, finally, at the post- processor level, problems of analysis and interpretation of the results. These problems occur because simulation software is highly complex. In addition, little practical help is available. The user is required to carry out many tasks unaided. Optimised use supposes a good understanding of the software’s capabilities and even an in-depth knowledge of how the software actually works (algorithm details and implementation). Some of these problems may be put down to the technical nature of the actual simulators, for instance in numerical or modelling aspects. These problems will be solved as fast as technical developments proceed. On the other hand, there are many problems that ought to be able to be solved by a better representation and exploitation of the available knowledge. The goal of our work is to build an environment to facilitate simulator use. Assistance is based on including in the environment more information than strictly necessary for simulation purposes. It also relies on an appropriate representation and use of two kinds of available knowledge: 1- general knowledge about fundamental physical and chemical laws or heuristics, 2- user-specific knowledge about the user's process captured in an appropriate structure of the environment. A good representation of this knowledge should be transparently clear for the user. The environment is devised from an object representation of the process and uses graphical and interactive techniques. In this paper, we will emphasise the use of these techniques for the analysis and presentation of simulation results. Comprehension is facilitated by the graphical interface. Interactivity provides flexibility, while an object- oriented approach is used to ensure a better adapted representation of knowledge. Our second part presents some recent progress in results visualisation. This is already able to help analysis by offering easy and concise access to results. The third part describes the modelisation concepts. This is the link which helps to bridge the gap between simulation results and the engineers’ ideas about the process. The concepts developed of path, aggregation and functional model are shown to improve visualisation. Moreover, this modelisation is important for the analysis tool presented in the fourth part, since it allows easy communication and better use of the tool. The fourth part introduces an analysis tool which enables results and certain expected values to be compared. The expected values are either given by the user or are some general and usually admitted chemical engineering criteria. Throughout the paper, the results analysis is applied to the process of hydrodealkylation of toluene (HDA). All the necessary data for simulation can be found in Douglas (1988). In the fifth part, a summing-up of this example shows how the concepts developed can be used. 2.VISUALISATION Graphical interfaces have come a long way in recent years and users are now much more demanding. Some commercial process simulators (ASPEN, ProVision, ...) enable the user to give the input data through adapted and interactive windows, use the graphical interface to show the simulation convergence on the flow-sheet and give access to results from the flow-sheet (a click on a stream or a unit will give its characteristics). However, the pre-processor aspect is usually more developed than the post-processor one. Some academic work (for example, Preece et al, 1991), where simulation is associated with design, is capable of producing data- sheets. Malone (Malone and Dye, 1996) proposes to show flow-rates with proportional bars and to look at aggregated units according to Douglas's hierarchical view (Douglas, 1988). These are real improvements from the old user’s interface which used to be closed to