Company: Syngenta
Student/ Degree: Emilio Guazzotti - Cheminal Engineering
Manager: Stephen Pearson
Project title: Producing a plant wide python model
Project summary
Combining data science with my background in engineering to understand and dissect a chemical production process in order to assist with the production of a plant wide python model, ensuring its behaviour accurately mirrored reality.
The project involved learning how to code simulations in python, utilising previously foreign data analytics software to build on understandings and visualise differences between the model and its tangible counterparts, thus, requiring detailed problem solving and a wide range of analytical techniques.
Project Results
The model is to help guide large capital investments, to which it is in its final stages and close to completion. Presently, the model has highlighted key areas for potential process improvements and detailed why limitations have occurred or where they may occur, concentrating both capital and human resources in specific analytically targeted areas.
What have you gained from your placement? (one paragraph)
My personal results and learning outcomes include: learning python code from scratch and understanding how to read and model a discrete event simulation, understanding the nuances behind statistical mathematics, developing effective communication and understanding of analyses through graphical displays and interactive models, and improving on methods of problem solving. My placement has developed me as an individual and provided a deeper understanding into the field of data science and engineering.
Company Manager’s Comment
The data science role requires good working knowledge of a second domain such as chemistry or engineering as well the ability to analyse complex systems, write code to describe them and visualize the results.
Emilio did well to combine his engineering training with the data science approach. He has shown great self-motivation and determination to learn an entirely new domain and apply it effectively.
The python modelling detailed in Emilio’s submission is a complex problem, even for someone with years of experience in the area. His contribution was vital to ensure we had the correct understanding of how the process was controlled and that the modelled behaviour accurately reflected reality.
His input was key in finding several programming errors, which resulted in the model giving wrong recommendations but which on the surface appeared to be correct. Given the size of the capital investment this model is helping to guide, he has help us to ensure the model is fit for purpose.
