Coordinated transmission expansion planning considering strategic data center companies: A game-theoretic approach

초록

The rapid growth of artificial intelligence (AI) services has accelerated the deployment of hyperscale data centers (DCs), whose concentrated loads reshape transmission congestion, locational marginal prices (LMPs), and long-term investment needs. In conventional transmission expansion planning (TEP), grid-cost-only placement ignores the private costs of data center companies (Datcos), whereas placement based on pre-DC prices can concentrate new load and trigger avoidable reinforcement; neither represents the strategic siting response within the planning framework. To address this gap, this study develops a game-theoretic coordinated design (CD) framework that endogenizes DC siting within TEP and market clearing. Formulated as a Stackelberg game, the framework represents the transmission company as the leader minimizing social cost while anticipating the Nash response of multiple price-anticipating Datco followers. The model jointly represents Datco workload routing, server activation, modular capacity, and communication links. A KKT-and McCormick-based single-level mixed-integer linear program generates search candidates; every reported CD profile is then certified ex post through exact integer best-response verification and market re-clearing. Case studies on IEEE 24-bus and 118-bus systems show that CD reduces social cost and total Datco private cost relative to the grid-driven and price-following benchmarks in all reported cases, while staying within 0.9% of best-found centrally planned social costs. These results show that coordination can avoid a zero-sum trade-off, and they highlight the value of coordinating DC siting, market clearing, and transmission expansion.

키워드