Optimization Model-Based Reliability Assessment for Distribution System with Data Centers Considering Multiple Heterogeneous Faults

The integration of data centers (DCs) into distribution systems introduces new challenges for reliability assessment. It also provides operational flexibility through the spatial migration of computational workloads. However, existing reliability assessment methods have not sufficiently captured the joint effects of DC workload flexibility, post-fault network reconfiguration, and multiple heterogeneous faults. To address this gap, this paper proposes an optimization model-based reliability assessment framework for distribution systems with integrated DCs. A DC workload dispatch model is first developed by considering server operation, cooling demand, backup requirements, and inter-DC workload transfer limits. The post-fault operation process is then formulated as a two-stage optimization problem, including fault isolation and service restoration. The proposed framework co-optimizes network reconfiguration and DC workload migration under line faults, node faults, and remote-controlled switch failures. The resulting problem is reformulated as a mixed-integer linear programming model and solved using commercial solvers. Case studies on a modified IEEE 123-node test feeder show that DC workload migration can significantly reduce expected energy not supplied by exploiting the spatial flexibility of interconnected DCs. The results also demonstrate that neglecting node faults and switch failures may lead to overly optimistic reliability estimates. Additional tests on a modified 356-node regional distribution system further verify the applicability and scalability of the proposed framework. The proposed method provides a physically interpretable reliability assessment tool for active distribution systems with DC integration and heterogeneous fault mechanisms.