This research award is to develop a framework of accurate reduced-order models that when combined with optimal model-based regulation will enable improved real-time control of high dimension, complex and nonlinear distributed parameter systems. The objectives of this research include: (i) develop more accurate computational models of distributed parameter systems that will attenuate the problem of adapting the model and (ii) investigate closed-loop stability of the proposed reduced-order model framework. These objectives will be achieved using a framework that (i) employs a particular reduced-order model generated by an accelerated Karhunen-Loève expansion; (ii) applies both probability and possibility theories to generate a random-fuzzy variable description to account for both systematic and random errors associated with uncertain variables and parameters; (iii) uses an efficient stratified sampling technique, Latin hypercube Hammersley sampling, to enable effective uncertainty propagation; and (iv) determines the optimal inputs for future predictions using a reduced-order, model-predictive, optimal control formulation.
We will use existing reservoirs and wells as a demonstration platform to prove out the proposed framework. It is expected that this framework will enable maximum production potential of reservoirs and wells by employing methods that mitigate environmental damage and improve our understanding of how to manage our existing and future reservoirs and wells. From a broader perspective, the overarching research needs of accurate modeling and control technologies are applicable to countless other complex processes whether fossil-based, biological-based, or otherwise. This award will promote a significant integration of research, education and industry. It will provide important research topics to graduate students and educate them as promising engineers and academics in future.
