A Network-Feasible Automatic Market Maker for Scarcity Pricing Under Congestion: Benchmarking Against LMP in a Stressed Network

Scarcity pricing must remain operationally meaningful under network constraints: dispatch must be feasible, congestion must be respected, and prices must not exhibit pathological escalation when marginal feasibility is dominated by load shedding. Conventional locational marginal pricing (LMP) provides a principled benchmark via OPF/SCED dual variables, but scarcity regimes can induce extreme or highly variable prices in stressed networks, motivating bounded and state-dependent alternatives. This paper proposes a network-feasible Automatic Market Maker (AMM) that couples OPF-based dispatch with a bounded, state-dependent scarcity controller. The mechanism is formulated as an iterative AMM-OPF loop: each iteration solves a standard DC-SCED (with reserves and VoLL-based feasibility) and updates an internal scarcity state using measurable physical stress indicators (served share and reserve shortfall). This state is mapped to a published scarcity price component that is bounded by design and scaled in monetary units using the contemporaneous bid stack. We establish feasibility and boundedness by construction and show convergence to a stable operating point under mild assumptions on the update map. A constraint-activation stress test benchmarks AMM outcomes against an LMP baseline under identical physical inputs, sweeping selected interface limits to induce unconstrained operation, persistent congestion, and scarcity. Results demonstrate that the AMM produces stable, bounded scarcity signals anchored to system costs while preserving network feasibility and comparable system-level energy balance, even though the resulting feasible redispatch can differ and thus distributional stress metrics (e.g., proximity to binding limits) may change.