Predictive LCA – a systems approach to integrate LCA decisions ahead of design
P. Karka, S. Papadokonstantakis, A. C. Kokossis
Computer Aided Chemical Engineering, Volume 46, pp.97-102, 2019
GW; Pearly stage process design
Bio-refineries are promising production options of chemicals production, capable to produce a wide range of fuels and chemicals equivalent to the conventional fossil-based products. To establish bio-refineries as mature choices and achieve the commercialization of their technologies, the application of sustainable solutions during the design and development stages are crucial. The innovative character of bio-based production and therefore data availability and access on process modelling details, is a challenging point for decision makers to move towards this direction. Considering the environmental dimension out of the three aspects of sustainability, Life Cycle Assessment (LCA) is a suitable methodology for the evaluation of environmental impacts of bio-based processes because it highlights the stages with the greatest impact along a production chain. LCA studies require large amount of information, usually extracted from detailed flowsheets or from already completed pilot plants, making this procedure, costly, time consuming and not practical to act as a decision- support tool for the development of a bio-refinery. The aim of this study is to develop predictive models for the assessment of LCA metrics and use them to highlight sustainable design options for bio-refineries. Models require the least possible information, which can be obtained from chemistry - level data or early (conceptual) design stages. The modelling techniques used in this study are decision trees and Artificial Neural Networks (ANN), due to their easily interpretable structure and high computational capabilities, respectively. Models are based on the extraction of knowledge from a wide dataset for bio-refineries (it refers to 32 products that is, platform chemicals (e.g., syngas, sugars and lignin) and biofuels (e.g., biodiesel, biogas, and alcohols), starting from diverse biomass sources (e.g., wood chips, wheat straw, vegetable oil)). Input parameters include descriptors of the molecular structure and process related data which describe the production path of a study product. Models are able to predict LCA metrics which cover the most critical aspects of environmental sustainability such as cumulative energy demand (CED) and Climate Change (CC). The average classification errors for decision- tree models range between 17% (± 10%) to 38% (± 11%) whereas for ANN models the average R2cv values (coefficient of determination) range between 0.55 (± 0.42%) to 0.87 (± 0.07%). Demonstration of models is provided using case studies found in literature. Models are used to rank options in various design problems and support decisions on the selection of the most profitable option. Examples of such cases are the selection of the appropriate technology or feedstock to produce a desired product or the preliminary design of a bio-refinery configuration. The proposed approach provides a first generation of models that correlate available and easily accessed information to desirable output process parameters and assessment metrics and can be used as pre-screening tools in the development of innovative processes, ahead of detailed design, thus saving time and money.