Abstract
A paradigm shift is currently underway in the wastewater sector from considering wastewater as a waste to be treated to a valuable source that has an increased potential for energy production and resource recovery. New systematic model-based methodologies are needed to comply not only with the changing
design objectives of the industry, but also with the increasingly stringent effluent quality regulations. In this work, we develop a novel simulation-based optimization framework for process synthesis and design of complex engineering systems whose accurate modeling requires a full-scale simulation, such as wastewater treatment plants (WWTPs). At the first step of the proposed three-step-framework, we postulate a superstructure comprising of alternative treatment technologies (including also the newly arising innovative technologies) and generate alternative plant networks using factorial combination along with expert knowledge. The second step identifies promising plant networks with the use of exhaustive Monte Carlo simulations, which are highly parallelized to align with the high-performance computing environments. The final step of the novel framework employs a model-based constrained derivative-free method to optimize the energy production of the plant using stochastic Kriging metamodels whose heteroscedastic variance information is used to represent inherent system uncertainties that are propagated with Monte Carlo simulations. The framework is applied to a case study of designing an energy surplus WWTP and the results show a potential for an increase in the energy production by about 20% compared to designs obtained from the second step without compromising the effluent quality.
design objectives of the industry, but also with the increasingly stringent effluent quality regulations. In this work, we develop a novel simulation-based optimization framework for process synthesis and design of complex engineering systems whose accurate modeling requires a full-scale simulation, such as wastewater treatment plants (WWTPs). At the first step of the proposed three-step-framework, we postulate a superstructure comprising of alternative treatment technologies (including also the newly arising innovative technologies) and generate alternative plant networks using factorial combination along with expert knowledge. The second step identifies promising plant networks with the use of exhaustive Monte Carlo simulations, which are highly parallelized to align with the high-performance computing environments. The final step of the novel framework employs a model-based constrained derivative-free method to optimize the energy production of the plant using stochastic Kriging metamodels whose heteroscedastic variance information is used to represent inherent system uncertainties that are propagated with Monte Carlo simulations. The framework is applied to a case study of designing an energy surplus WWTP and the results show a potential for an increase in the energy production by about 20% compared to designs obtained from the second step without compromising the effluent quality.
Original language | English |
---|---|
Publication date | 2019 |
Number of pages | 2 |
Publication status | Published - 2019 |
Event | Foundations of Computer-Aided Process Design 2019 - , United States Duration: 14 Jul 2019 → 18 Jul 2019 |
Conference
Conference | Foundations of Computer-Aided Process Design 2019 |
---|---|
Country | United States |
Period | 14/07/2019 → 18/07/2019 |
Keywords
- Simulation optimization
- Derivative-free optimization
- Monte Carlo simulations
- WWTPs