Parametric Models for Forest Industry Transformation in Energy Efficiency: Machine Learning Approach

Research output: ThesisDoctoral ThesisCollection of Articles

Abstract

This thesis is based on industrial projects with Pulp and Paper industry in a Nordic country. The main focus of the thesis is on the energy efficiency development of the thermomechanical pulp (TMP) mill and optimal integration of the TMP mill and paper machine through heat recovery and the concept of an Energy Hub. Advanced statistical approaches and machine learning methods have been employed to develop refining identification models and advanced energy-saving refining optimization methods for the TMP process. Results prove that an accurate refining identification model could be developed through advanced machine learning methods. The refining identification models to predict the refining energy (such as specific energy consumption) and final pulp quality (such as freeness and fiber length) can be further used to develop a refining control and optimization strategy. The developed optimization strategy based on the integration of Machine learning methods and Genetic optimization algorithm confirms an average reduction of 14 % for the total refining-specific energy consumption. In the following, the optimal integration of the TMP mill and paper machine has been investigated through the Energy Hub (EH) concept. The proposed approach for the cost and energy-efficient design and operation of EH is based on the integration of thermo-economic analysis, reliability and availability analysis, and EH load prediction. The proposed approach was first introduced and evaluated for the energy and cost-efficient design of a combined cooling, heating, and power (CCHP) system that provides the hourly thermal demand of a high-rise residential building. Results prove that by utilizing the proposed method, the system's average total cost could be reduced by 16% during the system's lifespan. As the presented method has shown to be effective in residential EH applications, this method was examined in a second case study (the forest industry) to determine the optimal integration of TMP mill and paper machines. The proposed design method offers a robust design that isn't impacted by penalty rates of unsupplied demand. Depending on the penalty rates, the total system cost could decrease by 14%-28% utilizing the proposed design method.
Translated title of the contributionKoneoppimismallinnus metsäteollisuuden energiatehokkuuden parantamisessa
Original languageEnglish
QualificationDoctor's degree
Awarding Institution
  • Aalto University
Supervisors/Advisors
  • Syri, Sanna, Supervising Professor
  • Laukkanen, Timo, Thesis Advisor
  • Holmberg, Henrik, Thesis Advisor
Publisher
Print ISBNs978-952-64-1302-0
Electronic ISBNs978-952-64-1303-7
Publication statusPublished - 2023
MoE publication typeG5 Doctoral dissertation (article)

Keywords

  • pulp & paper Industry
  • thermomechanical pulping Process
  • machine learning
  • energy hub

Fingerprint

Dive into the research topics of 'Parametric Models for Forest Industry Transformation in Energy Efficiency: Machine Learning Approach'. Together they form a unique fingerprint.

Cite this