Search papers, labs, and topics across Lattice.
This paper introduces a novel electricity-computation-water (ECW) nexus framework that integrates virtual water impacts into power system dispatch for data centers. It achieves this by embedding a differentiable dispatch optimization layer within a deep learning architecture, enabling end-to-end learning of coordination policies while maintaining operational feasibility. The framework demonstrates a 3-5% reduction in generation-related freshwater withdrawals in case studies, showcasing its ability to guide workload relocation and power dispatch to mitigate water stress.
Data centers can cut freshwater withdrawals by 3-5% simply by optimizing power dispatch with a new deep learning framework that accounts for "virtual water" embedded in electricity generation.
The expansion of data centers (DCs) drives a sustained increase in electricity demand and associated water withdrawals at generation sites. These withdrawals occur at generation sites and are virtually allocated to demand based on network power flows. Consequently, the actual water footprint of a specific load varies dynamically with generation dispatch and network conditions. Existing approaches typically rely on static statistical accounting to quantify these water footprints. However, such static methods fail to capture how dispatch optimization and workload relocation dynamically affect water withdrawals. As a result, static statistical accounting approaches remain decoupled from the optimization process, rendering them incapable of guiding workload relocation or power dispatch to mitigate water stress. To address this limitation, this paper develops an operational electricity-computation-water (ECW) nexus framework that internalizes virtual water impacts directly into power system dispatch. The framework represents dispatch optimization as a differentiable optimization layer embedded within a deep learning architecture, enabling efficient end-to-end learning of coordination policies while preserving operational feasibility. Combined with fixed-point coordination, the framework enforces consistency between virtual water attribution and physical generation-side withdrawals. Case studies on the IEEE 30-bus and 118-bus test systems demonstrate reliable convergence, exact power-water consistency, and reductions of approximately 3-5% in generation-related freshwater withdrawals under water-constrained conditions.
