Abstract
Cottonseed oil mills are electricity-intensive, and most of their demand comes from motor-driven hulling, conditioning and screw-pressing units whose specific energy consumption (SEC) depends strongly on raw-material properties that vary from batch to batch. Conventional energy accounting in such plants relies on monthly meter readings and cannot respond to these variations. This study develops and evaluates a real-time energy management approach for a cottonseed oil mill in Kokand, Uzbekistan, that links Internet of Things (IoT) sensing with process-level control. A five-layer IoT–edge–PLC–SCADA–cloud architecture samples electrical and technological variables at 1 s intervals. Energy characteristics of the main aggregates were identified by regression: the SEC of the MP-3000 screw press has a minimum of 19.1 kWh/t at a cake output of 0.68 t/h, and the residual oil in hulls reaches its minimum at seed moistures of 12.1%, 11.3% and 12.5% for grades I–II, III and IV. These relations are embedded in a moisture-aware SEC model that normalises throughput to conditioned mass and accounts for seed oil content, whose variation between 15% and 23% changes the SEC per tonne of oil by 53% at the press and by 36% at plant level. A damped Newton–Raphson optimiser computes the moisture and press-torque set-points in real time from a model re-identified on line, and the PLC enforces them through the seed conditioner and variable-frequency drives. In MATLAB/Simulink simulations parameterised with plant data, the closed loop narrowed moisture variability from ±2.4% to ±0.3%, press-torque fluctuation from ±18% to ±4% and load dispersion from ±15% to ±5%, and reduced the SEC of the pressing line from 53.1 to 49.2 kWh/t (7.3%). At plant level the implementation is credited with saving 142.2 MWh of electricity per year (128 million UZS) for an investment of 405 million UZS, giving a simple payback of 3.2 years and an internal rate of return of 29%. The results show that coupling IoT data with raw-material-aware process models turns energy monitoring into an active control variable in oilseed processing.