To cite this article: Woo, W. Q. et al. (2017). A novel methodology for molecular design via data driven techniques. J. Phys. Sci., 28(Supp. 1), 1–24, https://doi.org/10.21315/jps2017.28.s1.1
ABSTRACT
The design of chemical products that satisfies customer requirements commences with the identification of desirable properties for a specific application. Molecular design techniques have been traditionally used to find the products that meet the identified properties. Conventionally, product design based on properties is done based on the assumption that property prediction models are available for the target properties. However, in many design problems encountered in industry, such prediction models may not be readily available. In this paper, we have developed a systematic framework for the design of chemical products by targeting the attributes defined by customer even if there are no property prediction models available for target properties. In addition, a methodology has been developed for the understanding of the global interactions between properties and their impact on environmental and technical performance. Different kinds of chemometric techniques have been used to define the needs of the customer in terms of physical properties. In the next step, computer aided molecular design techniques have been integrated with the data driven methodologies to design the optimum products. To illustrate the applicability of the developed framework, a case study has been discussed to design environmentally benign chemical products to satisfy property requirements of a biofuel additive. This has been achieved in conjunction with the developed property models that represent consumer defined attributes of biofuel additives.
DOWNLOAD
