Frequency security constrained generation mix optimization with multiple heterogeneous resources

This is the authors' abstract. We don't add key points for this paper.

• This paper enhances the frequency performance in generation mix of power system. • Multiple resources’ frequency support including HVDC is considered in generation mix. • Proposing an adaptive piecewise linearization method to reduce frequency constraints. • Proposing a linearized reformulation method for the frequency constraints with HVDC. • Verifying the effectiveness and the superior of the proposed method. The increasing penetration of converter-interfaced generation has introduced significant frequency security challenges to modern power systems. From a generation portfolio perspective, this paper proposes a novel frequency security constrained generation mix optimization approach, which simultaneously considers the allocation and frequency support of multiple heterogeneous resources (i.e., synchronous generators, renewable energy units, and HVDC links). A linearized reformulation method combined with a high-accuracy bound tightening technique is developed to effectively transform the nonlinear frequency nadir constraint with multiple heterogeneous resources. An efficient adaptive piecewise linearization method for the frequency nadir is proposed to further reduce the number of frequency constraints. The case studies on the HRP-38 system and a practical power grid in China demonstrate that the proposed adaptive piecewise linearization method significantly reduces the computation time compared with the existing method, with an average improvement rate of 65.32% in various operation scenarios. The proposed generation mix model with emergency frequency control of HVDC can significantly reduce the investment cost of resources and capacity requirements of VSM by 81.6% and 222.2%, while ensuring frequency security. The bound tightening technique ensures much higher accuracy of the proposed linearized reformulation method while its solving time is 1/ 5.63 of the base method.