Energy consumption optimization in wastewater treatment plants: Machine learning for monitoring incineration of sewage sludge

Biomass management in terms of energy consumption optimization has become a recent challenge for developed countries. Nevertheless, the multiplicity of materials and operating parameters controlling energy consumption in wastewater treatment plants necessitates the need for sophisticated well-organized disciplines in order to minimize energy consumption and dissipation. Sewage sludge (SS) disposal management is the key stage of this process, such that incineration due to the high costs of drying remains a matter of concern. Thus, a combination of experimental investigations and data analysis is required for an efficient plant design. Herein, we propose an intelligent tool based on Machine Learning (ML) algorithms (A: Parallel, B: Artificial Neural Network (ANN), and C: Chained, ML models) by employing SciKit-Learn library in Python, followed by hyperparameter tuning and the k-fold cross-validation implementation. The optimizer receives simulation data from ASPEN PLUS software, and imitates the behavior of system outputs (namely, Yi: fluidized bed temperature, steam heat transfer rate, and dryer residence time in the SS) to yield optimal changing variables (namely, Xi: feed temperature, air temperature, fume temperature, steam flow rate, moisture content in the feedstock, and steam inlet temperature to dryer). The authenticity and precision of our intelligent optimizer was validated in terms of optimum heat transfer amount (the higher the better) and dryer residence time (the lower the better) by data collected from wastewater treatment plant in Gdynia (Poland), demonstrating excellent predictability of the algorithm. The R-2 values for A, B, and C ML models were 0.85, 0.94, and 0.91, respectively. The B model, though slightly revealed better prediction than the C model, estimated the outputs in much lower time than the former. Thus, C model was selected as the computational tool for the optimization purpose. Overall, we claim that the methodology developed herein takes the advantage of ca. 6% saving in the total amount of energy required for incineration unit of SS disposal plant, which is well justified considering the energy crisis raised by the geopolitical issues in the area and also the high cost of energy worldwide.

Automated summary

Biomass management for energy consumption optimization is a recent challenge in developed countries. In this study, we propose an intelligent optimizer based on machine learning algorithms to minimize energy consumption in wastewater treatment plants. By using experimental investigations and data analysis, our method can efficiently design plants and reduce energy consumption.