Loading...

Agent-Based Optimization of Integrated Energy and Product Networks

Kheirkhah Ravandi, Zahra | 2023

101 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 56655 (06)
  4. University: Sharif University of Technology
  5. Department: Chemical and Petroleum Engineering
  6. Advisor(s): Bozorgmehry Boozarjomehry, Ramin; Pishvaie, Mahmoud Reza
  7. Abstract:
  8. The industrial paradigm shift toward intelligent and sustainable management stimulates policymakers to leverage artificial intelligence for the decentralized planning of chemical industries with energy networks. By doing so, this work firstly presents a simulation framework to investigate the rigorous transient behavior of integrated systems comprising natural gas and power transmission networks, along with a chemical plant whose feedstock is natural gas. This framework entails dynamic models for the gas transmission network and the SynGas plant, and a continuous-time AC-power flow formulation with dispatchable loads. It addresses the following key challenges: (i) analyzing energy and chemical system interdependencies, and their impacts on each other’s supply reliability and security; (ii) providing an environment conducive to settling a critical question of how to prioritize the natural gas consumption as fuels of power plants or feedstocks of chemical plants based on sustainability and resiliency criteria. The framework provides an integrated benchmark enabling users to realistically study various planning, scheduling, and operation problems of these integrated systems such as control of interconnected systems, fault diagnosis, optimization, and environmental damage analyses. Secondly, this research proposes a collaborative platform for the optimization of the developed benchmark. Thus, a multi-agent framework reflecting AI-assisted hybridization is developed to autonomously handle interconnected optimization problems considering transient and large-scale operational challenges. Toward automated operations in specialized settings, several extensible behaviors and a dynamically-adaptable ontological information system are implemented to support agent interoperations and task specializations. The agents dynamically interact with the information system to use ready-made simulation models and solve their sub-problems. The case studies on an integrated system, in the southwestern zone of the United States, evidence the capabilities of the decision-making framework. According to the knowledge representation results, the semantically-enabled combined strategy allows automated model-building and simulation scaling. Furthermore, interconnecting chemical and energy systems through the consensus-based approach significantly enhances economic performance and supply reliability, compared to the uncoordinated conventional practices used in industry. Due to the lack of AI-oriented systematizations, traditional methodologies focus on either energy networks or statically viewed chemical process industries, neglecting the coupling constraints of both systems. This work, however, advances knowledge-driven optimal planning to attain dynamic material and energy flow balances between the interdependent systems. Such a flexible and extensible framework augments human judgments according to Industry 4.0 visions and materializes AI-sustainability symbiosis.
  9. Keywords:
  10. Multi-Agent Optimization ; Dynamic Simulation ; Dynamic Knowledge-Capturing ; Integrated Energy and Process System Benchmark ; Ontology-Based Information Management ; Interconnected Systems Optimization ; Multi-Agent Optimization Framework

 Digital Object List

 Bookmark

No TOC