Abstract: Diesel generator power tunnels in data centers commonly employ a series airflow cooling system that combines overall ventilation of the main equipment room with localized forced cooling provided by cabinet radiators. Under this operating mode, the thermal load varies dynamically with the number of operating generator units, while complex heat dissipation paths produce strong air-thermal coupling effects. Consequently, fan operating strategies have a significant influence on both cooling performance and energy consumption. However, existing studies lack refined methods for calculating and applying heat-source loads in numerical simulations, and limitations remain in full-scale ventilation network modeling, network adjustment, and the coordinated design and operation of multiple fans. To address these challenges, a diesel generator power tunnel in a data center was investigated under a representative summer operating condition in which nine generator units operated simultaneously at full load. A quantitative method was developed to calculate the residual indoor heat load in the engine room and the localized forced cooling load associated with cabinet radiators. Based on the air-thermal coupled cooling mechanism, the relationship between heat load and required air flow was established. SpaceClaim and Fluent were subsequently employed to develop a numerical simulation framework based on steady-state governing equations and the k－ε turbulence model, with spatially distributed loading methods proposed for the two heat-source types. A full-scale, high-resolution ventilation network model was then established, incorporating spatial heat load allocation and fan configuration. The two modes were compared in terms of airflow characteristics, temperature distributions, and energy consumption. The results indicate that: (1) The proposed heat-source decomposition method, based on the series airflow cooling path, and the corresponding spatial loading approach provide a reasonable representation of thermal loads in power tunnels. The developed simulation model and ventilation network adjustment method enable effective optimization while accounting for constraints associated with existing generator and fan configurations. (2) Under the existing fan equipment configuration, optimized operating conditions for both modes satisfy the required environmental control and heat dissipation criteria. Although coordinated fan control provides a slightly improved induced-air cooling effect, it results in substantially higher total fan power consumption and relatively low cooperative ventilation efficiency. (3) Under full load operating conditions, independent operation of the main fans is recommended, with an appropriate increase in mainfan capacity to further improve cooling performance.

Key words: underground data center, diesel generator power tunnel, operational ventilation, fan system, heat load, Fluent simulation, air-thermal coupling, energy consumption analysis