Modem industrial processes are often very complex and it is almost impossible to evaluate every alternative by experiment alone. The process steps nearly always depend on each other, and a change in operating conditions in one step affects the performance of other units. Computer simulations can be used to identify problem areas which require further exploration. However, computer simulations can never replace experiments, but constitute an important tool when planning and evaluating laboratory or pilot-plant experiments.

Computer simulation of industrial processes requires mathematical models that are able to predict, for example, reaction rates, yields, vapour-liquid equilibria and energy requirements. The simulation can be performed either with a tailor-made program for the process of interest, or by using a general flowsheeting program, such as Aspen Plus (Aspen Tech, USA), ChemCad (Chemstations Inc., USA), or Process (Simulation Sciences Inc., USA). Another possibility is to use spreadsheet programs such as Excel or Lotus to create the simulation models. Most process simulators are not capable of handling complex solid and heterogeneous materials such as wood or yeast. These materials cannot be characterised in terms of well-defined physical properties, e. g. molecular weight or enthalpies. However, these problems are encountered when simulating the production of ethanol, and the components must be characterised in terms of average physical properties for the material.

We have chosen to use the commercial process simulator Aspen Plus to perform simulations of different process configurations. Aspen Plus is capable of handling heterogeneous and solid materials, and it is also possible to incorporate user-defined components and unit operation modules into the program. When defining user modules, results from experimental investigations, such as laboratory-scale or bench — scale trials, are fitted to empirical or mechanistic models and included in the process simulation.

Cost estimations are also important to determine the economic feasibility of a process. It is valuable to obtain the distribution of costs within a process to identify high-cost steps so that research and development efforts can be directed towards the most expensive process steps. By means of sensitivity analysis it is possible to predict how changes in a specific parameter influence the yield and the economy of the overall process. A simulation program, BioEconomics, developed by von Sivers and Zacchi (48) is used to estimate the production cost for ethanol from lignocellulosic materials.

In the following, three examples of how techno-economic simulations combined with experimental investigations in the bench-scale unit can be used for process development are presented.