14.1 Introduction
14.2 Rack Cabinet Design
14.3 CFD and CAD Modelling
14.4 Boundary Conditions
Boundary condition | Symbol | Equation | Value |
---|---|---|---|
Inlet air flow velocity | U | – | m/s |
Heat load | \(\dot{Q}\) | \(\dot{Q}= \dot{m }Cp \Delta T\) | 40 kW |
Inlet temperature | T-in | – | 40℃ |
Outlet | P-out | – | Pressure outlet |
Server temperature rise | \(\Delta T\) | \(\dot{Q}= \dot{m }Cp \Delta T\) | 16℃ |
Heat transfer coefficient | h | \(\frac{\dot{m }{c}_{p}({T}_{air,u}-{T}_{air,d})}{A}\) = \(h \left({T}_{air,d}- {T}_{ext}\right)\) | 283.9 W/m2.K |
Heat exchanger loss coefficient | \({k}_{L}\) | \(\Delta p={k}_{L}\frac{1}{2} \rho {v}^{2}\) | 18.6 |
Heat exchanger temperature | T-hx | – | 14℃ |
Room temperature | T-room | – | 25 ℃ |
14.5 Results and Discussions
14.6 Conclusion
-
Inlet air temperature has great influence on the temperature rise within the cabinet, which could affect the thermal performance of the heat exchanger.
-
Temperature distribution in the cabinet does not exactly represent the realistic scenario of the 40 kW server rack. This will be optimized in near future work by implementing porous condition paraments to represent the pressure drop across the cabinet.
-
The simulations have provided a benchmark study of implementing energy efficient water-cooled data centre server racks. In the future, this simulation procedure will be further improved by conducting and validating experimental investigations.
-
Finally, rather than imposing the Boussinesq approximation to simulation air density, the compressible flow will be considered in the near future numerical investigations.