TY - JOUR
T1 - Techno-economic optimization and social costs assessment of microgrid-conventional grid integration using genetic algorithm and Artificial Neural Networks
T2 - A case study for two US cities
AU - Nagapurkar, Prashant
AU - Smith, Joseph D.
N1 - Publisher Copyright:
© 2019 Elsevier Ltd
PY - 2019/8/20
Y1 - 2019/8/20
N2 - Through two case studies, a methodology is presented that assessed the techno-economic and environmental performance of microgrid-conventional grid integration scenarios for fifty homes located in US cities of Fargo and Phoenix. The microgrid was composed of seven components - solar photovoltaics, wind-turbines, lead acid batteries, biodiesel generators, fuel cells, electrolyzers and H2 tanks. Firstly, mathematical models that predicted the hourly power generation were developed for every microgrid component. Secondly, Artificial Neural Networks were utilized to predict hourly electricity demand and its results were validated with actual available data. Thirdly, through an electricity dispatch strategy and a Genetic Algorithm optimization technique, microgrid configurations were determined that had lowest levelized cost of energy, $/kWh. From peak power standpoint, four microgrid-conventional grid integration scenarios were examined, namely, microgrid possessing penetration level of 25%, 50%, 75%, 100%. Based on the environmental life cycle assessment of power generation, three carbon taxes were imposed -$12, $48, $72/tonne carbon dioxide emitted. Microgrid's electricity cost was found to be $0.43–0.86/kWh. Imposing carbon taxes barely showed any effect on microgrid's electricity cost nor its optimum configuration, but conventional grid's electricity cost was found to increase by 7–33% as its carbon emissions were five times as that of microgrid.
AB - Through two case studies, a methodology is presented that assessed the techno-economic and environmental performance of microgrid-conventional grid integration scenarios for fifty homes located in US cities of Fargo and Phoenix. The microgrid was composed of seven components - solar photovoltaics, wind-turbines, lead acid batteries, biodiesel generators, fuel cells, electrolyzers and H2 tanks. Firstly, mathematical models that predicted the hourly power generation were developed for every microgrid component. Secondly, Artificial Neural Networks were utilized to predict hourly electricity demand and its results were validated with actual available data. Thirdly, through an electricity dispatch strategy and a Genetic Algorithm optimization technique, microgrid configurations were determined that had lowest levelized cost of energy, $/kWh. From peak power standpoint, four microgrid-conventional grid integration scenarios were examined, namely, microgrid possessing penetration level of 25%, 50%, 75%, 100%. Based on the environmental life cycle assessment of power generation, three carbon taxes were imposed -$12, $48, $72/tonne carbon dioxide emitted. Microgrid's electricity cost was found to be $0.43–0.86/kWh. Imposing carbon taxes barely showed any effect on microgrid's electricity cost nor its optimum configuration, but conventional grid's electricity cost was found to increase by 7–33% as its carbon emissions were five times as that of microgrid.
KW - Artificial neural network
KW - Carbon tax
KW - Environmental assessment (LCA)
KW - Genetic algorithm optimization
KW - Microgrid
KW - Techno-economic assessment
UR - http://www.scopus.com/inward/record.url?scp=85065819344&partnerID=8YFLogxK
U2 - 10.1016/j.jclepro.2019.05.005
DO - 10.1016/j.jclepro.2019.05.005
M3 - Article
AN - SCOPUS:85065819344
SN - 0959-6526
VL - 229
SP - 552
EP - 569
JO - Journal of Cleaner Production
JF - Journal of Cleaner Production
ER -