Data-based energy optimization of mechanical pulping process — A data visualization technique

In the pulp and paper industry, mechanical pulping (MP) process is widely used. However, the MP process is extremely energy intensive. Together with the increase of electricity price and environmental concerns, efforts are desirable to find one way to optimize energy consumption while maintaining the pulp quality. As a continuous process with multiple operating variables, sequential operations will impact one after another and eventually affect the final product quality. Hence, the interactive and overall impacts of operating variables should be considered while performing optimization, which inspires the development of this systematic strategy.

As a solution, a real-time optimization framework and a feasible operating region visualization technique are proposed. First, the entire MP process is separated into multiple segments according to the production workflow, namely, wood chip treatment, two-stage high consistency refining, latency removal, screening and reject refining processes. Then, mainly focusing on the first three operating segments, a high-precision model has been established and validated on the real process data. Based on the model, a nonlinear real-time optimization strategy is employed to explore the feasible and the most energy saving operating regions, by considering all available operating variables. Finally, an effective dimension reduction technique is developed based on the Kiviat diagram, which allows to visualize operating regions in 3D space over time. This technique has been tested and validated on a real MP process dataset, with satisfactory performance. Moreover, the proposed framework can also be extended to evaluate larger process segments with a hierarchical structure as a future work.