A Modular Modelling Environment for Computer-Aided Process Design

Abstract We present a modelling approach which allows to formulate and solve steady-state and dynamic process systems engineering problems in the programming languages Fortran and Python. Additionally, we have developed a property prediction software which can be installed with a container on a local machine or on servers rented by the user, research institution or industrial entity. The prediction tool comes with a web-interface and an application programming interface (API) which one can connect to and request predictions of fixed physical and thermo-physical properties directly into their routines. Further we used Pyomo to solve superstructure optimization problems with surrogate models to find the optimal process structure and point of operation to the given global constraints. On the superstructure level we connect a graphical user interface (GUI) with Pyomo to make it easier for the general user to work with an equation-based environment where initial estimates for all variables can be defined through the spreadsheet-style table of the GUI. This modular, multi-level approach also allowed us to connect commercial process simulators with our developed Python-COM interface to perform sensitivity analysis or retrieve surrogate models in e.g. Aspen or PRO/II. The process design, process flow-sheeting and superstructure optimization layer can individually be wrapped with lifecycle analysis tools. The modelling environment is platform independent (except the information retrieval from Aspen or PRO/II). All the necessary code examples in this environment is hosted and further developed in several different repositories freely accessible to scientific and industrial users. The code is published under a free license.