VHTX: A code for simulation of steady state and dynamic response of single or multiple networked cross flow heat exchangers in data center thermal management systems

Abstract

Crossflow heat exchangers are key components of both centralized (e.g. CRACs) and decentralized (rear door, in-row) cooling equipment utilized in data center thermal management systems. Modeling of their behavior in steady state is well documented but transient or dynamic operation, whether by intent or as a result of a system failure, has not been well documented. In “smart cooling” scenarios, cooling should be modulated with heating (i.e. IT) load which can vary with time and space as IT load varies within a rack and within the data center room. Cooling is optimally utilized when the cooling load follows or even anticipates the heating (IT) load and as such the heat exchanger operates in a dynamically controlled mode. In data center operation, there is also interest in understanding their behavior in case of system malfunctions such as pump or chiller failures which results in transient operation. The MATLAB™ simulation code VHTX was developed in order to simulate the performance of crossflow heat exchangers in both steady and dynamic operation. It is a standalone code for simulation of heat exchanger networks and core code elements are also being embedded into or coupled with other simulation environments such as MATLAB SIMULINK™ for control investigations, VTAS for data center system thermodynamic and energy analysis, and CFD codes for room simulations. This paper describes the basic formulation of the VHTX solver and its validation against research quality data on heat exchanger cores. It is shown that the code can accurately predict the coolant flow distribution within the heat exchanger core and its dynamic response to temporal events such as modulation of the coolant flow rate or temperature to match the air side thermal load. A case study simulating a typical rear door heat exchanger is presented as an example of the use of the code in a data center simulation.