In civil and mechanical engineering, the design process is done almost entirely by computer (think about the way bridges are designed). A long-held goal in radical chemical and materials product design is to shift from an ad-hoc, labour intensive and expensive process towards a more robust and adaptive computer aided paradigm.
Ultimately the goal is to enhance de novo product design, shorten time to market, provide an adaptive response to supply chain variability, and encourage the adoption of development for sustainability. The ability to design virtually allows for acceleration of R&D processes and the smoother development of new products, especially for high-value manufacturing markets where growth arises from high R&D intensive efforts.
To this end, we have set up a chemistry and materials programme focused on developing an integrated data analytics, modelling, and simulation capability that is able to address a broad variety of chemical and materials problems with world-class scale and accuracy. It includes multiple research activities that can be separated into two main threads, one focused on simulation and the other on data and cognitive computing.
The overall programme will exploit the most recent advances in HPC, modelling and simulation, big data and cognitive computing (or specifically cognitive chemistry as we call it in our team), and this will require step-changes in a number of interdependent fields. The Hartree research team, made up of STFC and IBM colleagues, will address challenging issues involved with enabling non-specialists to access HPC and simulation in an easy fashion. This part of the work is vital to enable scientists to harness the power of HPC and to perform in-silico experiments that will focus their research & development activities.
The programme has a number of broad work areas:
- Automated model parameterisation: A bottle-neck in the application of modelling and simulation for the chemical industry is in the quality of the models employed. We will explore novel ways of developing improved models and parameters to better represent real systems.
- HPC in the cloud: To derive maximum impact of our methods we need to put the power of HPC into the hands of the non-expert. We will develop interfaces and methodologies that allow industry to adopt HPC internally to help transform their R&D activities. Delivery by cloud and container-based solutions will allow for maximum flexibility.
- Smarter simulations: Traditional HPC has often applied a brute force approach in the search for answers. We are developing methods to facilitate smarter simulation, in essence helping to minimise the amount of simulation required whilst maximising the quantity of useful data.
- Big data for chemistry: What important insight can we determine from the large volumes of data generated by the chemical industry? We will develop and employ big data analytics methods to capture valuable information relating to product design and performance.
Over the coming months I will write pieces on our Hartree blog to let you know how our research activities in this field are progressing, and to bring you results of our projects within the chemistry and materials programme.
We are currently in the process of capturing the requirements of the UK chemical industry. If you want to know more about what we are doing, or how you can be a part of this work, then please don’t hesitate to contact us. We would welcome your views.
Thanks for reading!
Dr Rick Anderson