TY - GEN
T1 - Intelligent joint spatio-temporal management of electric vehicle charging and data center power consumption
AU - Sun, Zhonghao
AU - Kong, Fanxin
AU - Liu, Xue
AU - Zhou, Xingshe
AU - Chen, Xi
N1 - Publisher Copyright:
© 2014 IEEE.
PY - 2015/2/10
Y1 - 2015/2/10
N2 - A data center is designed to have the capacity matching spike workload from a geographical region. This design concept comes with a significant waste of expense on power peak charged by the electrical grid, since workload fluctuations cause large variations in data center power demand. On the other hand, electric vehicles (EVs) have been emerging as major electricity consumers due to their large power demand for battery charging. In this paper, we propose a ValleyFill method that explores EVs to fill power valleys of geographically distributed data centers without increasing their power peaks. Leveraging geographical diversities of workload processing and temporal flexibilities of EV charging, this method determines routing scheme for workload and charging schedule for EVs to improve cost efficiency on the peak charge. We evaluate the proposed method with real-world workload traces and EV arrival patterns. The result shows that our method significantly improves the cost efficiency and saves up to 6% on total electricity bills. We observe that a data center with larger gap between its power peak and valley leads to less charging time for EVs and less workload migration.
AB - A data center is designed to have the capacity matching spike workload from a geographical region. This design concept comes with a significant waste of expense on power peak charged by the electrical grid, since workload fluctuations cause large variations in data center power demand. On the other hand, electric vehicles (EVs) have been emerging as major electricity consumers due to their large power demand for battery charging. In this paper, we propose a ValleyFill method that explores EVs to fill power valleys of geographically distributed data centers without increasing their power peaks. Leveraging geographical diversities of workload processing and temporal flexibilities of EV charging, this method determines routing scheme for workload and charging schedule for EVs to improve cost efficiency on the peak charge. We evaluate the proposed method with real-world workload traces and EV arrival patterns. The result shows that our method significantly improves the cost efficiency and saves up to 6% on total electricity bills. We observe that a data center with larger gap between its power peak and valley leads to less charging time for EVs and less workload migration.
UR - http://www.scopus.com/inward/record.url?scp=84924389953&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84924389953&partnerID=8YFLogxK
U2 - 10.1109/IGCC.2014.7039141
DO - 10.1109/IGCC.2014.7039141
M3 - Conference contribution
AN - SCOPUS:84924389953
T3 - 2014 International Green Computing Conference, IGCC 2014
BT - 2014 International Green Computing Conference, IGCC 2014
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2014 International Green Computing Conference, IGCC 2014
Y2 - 3 November 2014 through 5 November 2014
ER -