Utilization of Predictive Analytics & Constrained Metaheuristics toAchieve Optimal Operation of aLiquid-fueled Industrial Process Heater

predictive analytics;process heater;efficiency;metaheuristics;optimization;thermal;oilfield;mechanical engineering;systems;integration;combustion;control;sensors;actuators;ann;ga;artificial neural networks;genetic algorithm;artificial intelligence;Mechanical Engineering

Prodaniuk, Cody L

University of Waterloo

关键词: predictive analytics;    process heater;    efficiency;    metaheuristics;    optimization;    thermal;    oilfield;    mechanical engineering;    systems;    integration;    combustion;    control;    sensors;    actuators;    ann;    ga;    artificial neural networks;    genetic algorithm;    artificial intelligence;    Mechanical Engineering;   

Others  :  https://uwspace.uwaterloo.ca/bitstream/10012/9330/1/Prodaniuk_Cody.pdf

瑞士|英语

来源: UWSPACE Waterloo Institutional Repository

【 摘 要 】

The operation of industrial combustion devices to improve fuel efficiency and reduce emissions, is necessary in many respects. Increasingly stringent environmental regulations and the present volatility of the price of fossil fuels make it necessary to control a combustion device at an optimal operating point while accounting for changes in the environment, heater load, and fuel quality.In this study, an adaptive and predictive optimal control methodology is developed and tested on a 2.05 MW diesel fueled oilfield process heater. To date, the majority of previous developments and control methodologies have focused on combustion devices either a magnitude larger or smaller, or on gaseous-fueled combustion processes. The present work, which focuses on maximizing the thermodynamic efficiency of an oilfield process heater, is divided into two parts; the first of which is the development of an algorithm that continually regulates and predicts the operating state. The algorithm is comprised of three iterative components: (a) an artificial neural network for adaptivity and prediction; (b) a genetic algorithm for optimization; and (c) a refinement of the search space to complement and restrict the other components. The second part of this work is the physical implementation of the sensors, actuators, and computing hardware necessary for the algorithm. Several challenges encountered during the implementation on the experimental apparatus are discussed, namely involving the diesel flow rate sensor and the actuator for modulating the damper position.Two experiments were performed, and data were collected and evaluated offline. The first experiment was to initially evaluate performance of the control methodology, and the second was to evaluate the iterative and refinement capabilities of the algorithm. The algorithm optimizes the operating state of the heater and the results agreed with observed trends discovered through the experimentation. Additionally it was discovered that there is a correlation between the optimal operating state and increases to the quantity of heat transferred to the process fluid flowing through the inner coil.