Created: 5/21/95 Updated: 8/18/98

Direct Forced-air Thermal Modeling of Commercial Electronics

Project Number : C93-10

Point of Contact:
Dr. Michael Osterman
CALCE EPRC
email: osterman@calce.umd.edu
Phone: (301) 405-8023
Fax: (301) 314-9269
Objectives Background Work Accomplised

Objectives

Historically, empirical methods have been used to analyze the cooling design of commercial off-the-shelf (COTS) products. However, the growth in the availability of computational fluid dynamics (CFD) programs provides an attractive alternative to reducing construction of prototype hardware. Although CFD programs exist, there is a need to evaluate the accuracy and limitations of such modeling capabilities. The main thrust of this effort is to correlate computational results with experimental test data.

 

Background

Recent changes in military equipment acquisition policies permit the use of ruggedized commercial off-the-shelf (COTS) equipment for applications previously requiring fully military qualified equipment. COTS technologies typically employ forced and natural convection cooling modes in their thermal management plans. With increased power dissipation expected from more powerful microprocessors and associated equipment, and with decreasing time to market, the design community faces the challenge of providing sound thermal management and timely design review. The use of computational models is the primary method of meeting this challenge. It is critical that the modeling procedures be experimentally validated.

 

Work Accomplished

The CALCE Center has conducted experimental tests using the test fixture provided for the Westinghouse AWACS experiments. In these experiments, board and component temperatures were measured under active convection cooling. Other test parameters included the volumetric flow rate of the working fluid and board spacing. The test boards contained 144 14-pin DIP components placed on a regular 12x12 array. Air channel heights used in these experiments were half an inch and one inch. The air-flow velocities varied between 100 ft/min to 700 ft/min. The data obtained from the experiments was used to approximate the convection heat transfer coefficients. Along with the experimental work, computational models were established using both the CALCE and FLOTHERM software packages. The results of the experimental tests were compared with the established computational models. These comparisons demonstrated the acceptability of the convection heat transfer coefficient model used by the CALCE Thermal Analysis Tool. In addition, the FLOTHERM software proved to be quite accurate. A detailed summary of this work available in the form a paper published in the IEEE Transactions..

 Experiments using the test vehicle provided by IBM were performed on a 20-slot VME (6U) chassis populated with one 6U VME card and 19 cards with patch heaters. IBM also provided an air mover capable of delivering the volumetric flows to be tested and a laminar flow element to measure the flow rate, the VME chassis, and three populated VME 6U cards. The CALCE EPSC constructed 20 dummy cards with patch heaters and prepared the chassis for testing.

In addition to this experimental work, computational models were established for both the CALCE and FLOTHERM software packages used to simulate the experiments.