TY - JOUR
T1 - Energy Cost Optimization of Globally Distributed Internet Data Centers by Copula-based Multidimensional Correlation Modeling
AU - Lasemi, Mohammad Ali
AU - Alizadeh, Shahin
AU - Assili, Mohsen
AU - Yang, Zhenyu
AU - Baboli, Payam Teimourzadeh
AU - Arabkoohsar, Ahmad
AU - Raeiszadeh, Amin
AU - Brand, Michael
AU - Lehnhoff, Sebastian
PY - 2023/12
Y1 - 2023/12
N2 - The high operating costs of Internet Data Centers (IDC) are a major challenge for their owners worldwide. Therefore, more attention has recently been paid to the energy and cost management of IDCs. This paper investigates the optimal operational strategy for minimizing the electricity costs of a group of globally distributed IDCs in different locations under various day-ahead electricity markets, and each is equipped with a high-performance energy storage system. For this goal, optimal workload dispatching and optimal energy management of the storage units of all IDCs are simultaneously perused by the proposed problem. The system is modeled regarding power balancing constraints, battery costs, and quality of service (QoS). For more practical results, a penalty function is also considered when QoS constraints are not perfectly met, and the impact of the batteries’ depth of discharge on the cost of energy storage is also modeled. Moreover, the cross-correlations between the traffic of IDCs are also considered by the multidimensional copula function. The proposed energy cost optimization is linearized for increasing the accuracy of convergence. The results show that not only the power consumption pattern of the IDCs is significantly improved, but also the cost of power consumption is reduced by 34%. The results also prove the positive effect of battery discharge on workload dispatch and represent a compromise between battery costs and electricity cost savings.
AB - The high operating costs of Internet Data Centers (IDC) are a major challenge for their owners worldwide. Therefore, more attention has recently been paid to the energy and cost management of IDCs. This paper investigates the optimal operational strategy for minimizing the electricity costs of a group of globally distributed IDCs in different locations under various day-ahead electricity markets, and each is equipped with a high-performance energy storage system. For this goal, optimal workload dispatching and optimal energy management of the storage units of all IDCs are simultaneously perused by the proposed problem. The system is modeled regarding power balancing constraints, battery costs, and quality of service (QoS). For more practical results, a penalty function is also considered when QoS constraints are not perfectly met, and the impact of the batteries’ depth of discharge on the cost of energy storage is also modeled. Moreover, the cross-correlations between the traffic of IDCs are also considered by the multidimensional copula function. The proposed energy cost optimization is linearized for increasing the accuracy of convergence. The results show that not only the power consumption pattern of the IDCs is significantly improved, but also the cost of power consumption is reduced by 34%. The results also prove the positive effect of battery discharge on workload dispatch and represent a compromise between battery costs and electricity cost savings.
KW - Correlation modeling
KW - Electricity market
KW - Energy management
KW - Energy storage system
KW - Internet data center
KW - Quality of service
KW - Workload distribution
UR - http://www.scopus.com/inward/record.url?scp=85143806588&partnerID=8YFLogxK
U2 - 10.1016/j.egyr.2022.12.033
DO - 10.1016/j.egyr.2022.12.033
M3 - Journal article
SN - 2352-4847
VL - 9
SP - 631
EP - 644
JO - Energy Reports
JF - Energy Reports
ER -