Large-equal temperature difference design for water-mediated air-cooled data center cooling systems

Water-mediated air-cooled cooling systems remain widely used in data centers, but their chilled-water and cooling-water temperature differences are still commonly selected from conventional design values. This study represents the cooling process as a series heat-exchange network and uses heat-capacity-flow-rate matching to guide large-equal temperature difference (LETD) design. A steady-state annual model was developed with an R134a chiller correlation that accounts for both part-load ratio and pressure ratio, and the model was supported by prototype component tests and a system-level testbed. In the ideal equal-temperature-difference analysis, the nine-city annual optima are 12.1–19.4 K and the corresponding minimum cooling load factors (CLFs) are 0.0788–0.1432. When the server-room air-temperature span is fixed at 10, 12, or 14 K, the optimum common chilled-water/cooling-water temperature difference becomes 10.5–12.0, 11.1–12.0, and 11.6–11.8 K, respectively. Variations in CRAH, plate-heat-exchanger, and cooling-tower performance give real-point optima of 9.4–13.5 K, confirming that 12 K is a useful nominal reference. Relative to the traditional 6 K system, the 12 K LETD system reduces annual cooling-system electricity by 15.32%–19.10% across nine cities, and the normalized investment model gives base reductions of 6.94%–13.66%. The results provide a system-level design method that links ideal temperature-difference matching, practical room-side constraints, and equipment-dependent annual optimization.