Completed: Data-driven optimisation framework for assessing energy and emission saving potentials in foundation industries

In collaboration with Palm Paper

This project focussed on the assessment of prospective energy and cost savings associated with foundation industries. A case study of Palm Paper was considered but the underlying concepts transcend this specific application. Palm Paper is a paper production company which produces newsprints and news paper grades from completely recycled materials who rely on natural gas and electricity for their energy sources.

The project took place over 3 phases (1): Building Datasets: this phase entailed the development of appropriate experiments required to capture dynamic or static properties of the paper production process. It encompasses data collection, processing and analysis which forms the basis of the system modelling. (2): Identification of Benchmark Representative Production Activities: This task required the identification of manipulated, controlled and measured variables based on available data and included interactions between the boilers, combined heat and power (CHP) systems, paper separation plants, and pulping and de-inking station. (3): Data-driven modelling: Each sub process was modelled and subsequently combined to represent the whole plant. Both static and dynamic models were considered depending on their suitability for predicting process outputs. (4): Optimisation: The impact of this project is specifically embodied in the optimisation phase.

A potential reduction in the gas and electricity consumed for paper production is projected with a Model Predictive Controller implemented. This translated to eventual processing cost reduction and energy savings, which are expected to range between 10% and 50%.

The research identified suitable process models describing subunits of a paper production plant for the purpose of variable predictions and energy optimisation. Models were developed which captured static and dynamic behaviours of the plant and results are being compiled for a more detailed journal publication.

Dr. Alessandra Parisio

University of Manchester

alessandra.parisio@manchester.ac.uk