Search

At an ever increasing rate innovative chemistry and technology platforms are reshaping manufacturing environments to become factories of the future by being more productive, lean and flexible. The use of Life Cycle Assessment (LCA) in early process development phases has been challenged many a time to assess whether or not this willingness to strive for innovation is an environmentally sustainable one, often due to a lack of process data. This study takes the streamlining of LCA one step further and proposes an optimal complexity of modelled product systems in terms of their optimal set of predictor variables. Out of more than 2,800 Basic Operations (BOs) in pharmaceutical synthesis steps, candidate predictor variables were identified to forecast the environmental burden (in this case natural resource consumption) of a production step per unit of output. By means of backwards stepwise linear regression modelling, combinations of candidate predictors were tested and evaluated based on their predictive power (R²) and the model uncertainty. It was proven that at least the amount of organic solvents used, the molar efficiency and the time duration of the synthesis step should be included in the model (R² = 0.87) as being the most significant predictor variables. Including additional predictors however imposes no guarantee to contribute to the predictive power and eventually weakens the model interpretation and its simplicity. The results of the study were evaluated in the light of the product-specific versus product group approach debate. Should LCAs be generalized to such an extent that an extensively diversified product group is to be represented with an averaged burden, while fairly simplified and streamlined methods can represent product-specific impact assessments with a reasonable need for data, time and knowledge? The trade-off between simplicity and accuracy will be dealt with quantitatively in the oral presentation. Ideally, an organization should be able to derive its environmental impacts from readily available Enterprise Resource Planning (ERP) data, linking supply chains back to the cradle of resource extraction, excluding the need for an approximation with product group averages. While this study has taken a step in that direction, more research should be conducted especially on how efforts towards sustainable development should be addressed with care to valorise them efficiently in the supply chain and its sectors, beyond any company borders.