Critical analysis of the use of white-box versus black-box models for multi-objective optimisation of small-scale biorefineries

Biorefinery systems that are embedded in their local setting provide an attractive framework for the valorisation of locally available food- and other bio-waste streams. They can aid in the provision of local bio-waste processing facilities as well as the targeted revalorisation of local bio-waste feedstocks by converting them in locally desired biorefinery products. Since food- and other bio-waste feedstocks are often diffuse feedstocks, small-scale biorefineries that are tailored for their local setting are the most suitable biorefining system for their processing. Whereas small-scale biorefineries cannot rely on the economy-of-scale to be an economic sustainable endeavour, they need to be meticulously optimised according to multiple sustainability objectives. These objectives can be of economic, societal, or environmental nature. A commonly used optimisation criterion in these problems is the energy requirements of the entire biorefinery system. For many commonly used biorefinery processes mass balance models are available (which are often mechanistic models), however, energy balances are difficult to obtain. Chemical process simulators, like Aspen Plus, provide an extensive toolkit to easily model the mass- and energy balances of a multitude of chemical processes. However, especially in the context of multi-objective optimisation, the obtained white-box models are too complex to simulate the considered processes efficiently consecutively. Therefore, in this contribution, a critical analysis is presented of the use of white-box versus the black-box models in the context of the multi-objective optimisation of a small-scale biorefinery. An in-house developed biorefinery network is re-modelled in Aspen Plus and used as a digital twin for the development of a surrogate model. Eventually, the modelled biorefinery network is optimised using both models and a comprehensive evaluation is drafted.