Solving Complex Problems in Formulation Processing by Building a Pragmatic Multi-scale Model
Multi-scale and/or multi-disciplinary approach to process-product innovation
Multiscale Modeling (T3-5)
Keywords: multi-scale,pragmatic,complex, modelling
Traditional modelling techniques tend to model processes using complex mathematical formulae and equations but it is often not practicable to employ such techniques when dealing with complex systems as it would be expensive, time consuming and most importantly for many such systems the quantitative data required for the models to work are lacking. Traditional modelling tends to concentrate on either the molecular level (performed by chemists) or the macroscopic level (performed by engineers) of the problem and there is a limited of linkage between them which contributes to making scale-up a problem.
The methodology adopted in this work is to build a pragmatic multi-scale model using the observed phenomena and known basic science of the system. The model will be used to drive the design of specific experiments to explore the system and could also be potentially used as a basis for a mathematical model.
The methodology is exemplified using a manufacturing process typical of personal care products and typical pharmaceuticals which involves lyotropic liquid crystal. The use of this process is appropriate as there are multiple phenomena involved at different length scales which include soft solid phase formations, phase transformation, mixing and mass transfer among others. At a micro level, little is known about the kinetics of the process and the microstructure of the products. At a macro level, there is a lack of reliable scale up parameter and scaling up of the process has been a long problem.
A first step in the process is to obtain or “extract” information and understanding of observed phenomena from people who have been working on the process. The methodology involves development of a preliminary collection of information that identifies main transformations and captures a high level process map of the process. This is followed by Literature gathering and gap analysis to bring together both literature and experimental results of the system on the macro and micro level in order to identify the gaps in the knowledge. The data are in general represented as causal graphs and pictorial representations of phenomena linked to quantitative data and models where available. The model is extended and validated by gathering experts who have worked with the system, both the scientists and the engineers, to explain fully or share explicitly their ideas or belief about the system and test whether the model explains the observed system behaviours, and a working model of the system is developed. This model is used to design specific experiments to prove or disprove the certain part of the model. With the new information from the experiments the model is updated and again is presented to the panel of experts who would suggest improvement to the model. The process is an iterative one whereby the model is tested and the model modified as new results are obtained.
By employing the methodology we were able to design specific experiments which enable us to gain better insights in to the process which were previously not known. This potentially represents a significant improvement on the widely used approach in such systems which are to use statistical design of experiments.
See the full pdf manuscript of the abstract.
Presented Tuesday 18, 09:25 to 09:45, in session Multiscale Modeling (T3-5).