NTWIST and Sherritt Technologies assessed and proved the viability of an artificial intelligence (AI) process optimization solution for HPAL autoclave operations in 2018. The AI model identified operating conditions that increased potential profit during approximately 40% of the total operating hours. In light of the findings, Sherritt Technologies and NTWIST are collaborating to provide the AI process optimization solution to external customers.

Problem Statement

Autoclave operators have insufficient information to actively maintain optimal operation.

Solution

We used historical data to create a data-driven ‘Digital Twin’ model of the autoclave circuit which enabled us to predict process parameters based on current inputs. We then created an AI-powered optimizer/recommender which utilized the digital twin model to estimate the optimal set points and provide guidance to the operators.

Benefits and outcomes

Unlike human operators, our solution takes into account all available input data, and estimates the unavailable data. The AI has access to up-to-date economic data and takes into account downstream costs. Over the six-month evaluation period, the AI model identified operating conditions that increased potential profit in approximately 42% of the total operating hours; even with the limited operating data available to build and train the models. The AI was also able to identify the setpoints required to bring adverse operating conditions to within their target limits in approximately 17% of the total operating hours.

The digital twin model can also be used for personnel training and as an analytics tool.