Ranking Low-GWP Refrigerants for Data-Center Chillers Across Climate Zones and a Decarbonizing Grid

A Total Equivalent Warming Impact Analysis

Abstract

From 1 January 2027 the American Innovation and Manufacturing Act caps refrigerant global-warming potential (GWP) at 700 in new data-center cooling equipment, forcing a choice among several low-GWP chiller refrigerants. Existing Total Equivalent Warming Impact (TEWI) comparisons cannot settle it: they use a single operating point or a generic comfort-cooling profile, and a static grid emission factor. A data-center chiller instead runs at near-constant load year-round, rejects heat under climate-dependent conditions, and spends 15 to 25 years on a decarbonizing grid. This paper develops a transparent TEWI framework closing all three gaps. Real-fluid coefficients of performance for eight refrigerants, from CoolProp, are coupled to two generic heat-rejection archetypes bracketing practice, a wet-bulb-driven water-cooled condenser and a dry-bulb-driven air-cooled condenser, over annual climate bins for five U.S. stations and three control strategies, with the indirect term summed year by year along regional grid decarbonization trajectories rather than held constant. Two findings follow. The compliant low-GWP fluids differ by only a few percent in lifetime TEWI, within the model's uncertainty, so the fluid is better selected on safety, charge limits, and cost; single-stage transcritical R-744 the one clear exception, 19 to 81 % worse depending on climate and control. What does move lifetime impact is the heat-rejection architecture and its control, which alter plant energy by up to 57 % and leave tower water largely unmoved. Built from open thermophysical, climate, and grid data, the framework lets operators select refrigerant, architecture, and control jointly, not in isolation.