This article also appears in
Subscribe now »

Exploded view of the final design for the CFRP enclosure and TMS demonstration unit. Internally, copper fins were added to collect heat from the air for the lower power slices, while a HiK plate was designed for the high power chips on the motherboard. CCHPs were jogged through and flattened in walls. Externally, a single fin stack was designed with an attached fan to remove the heat brought through the walls by the embedded heat pipes.

Heat-pipe embedded CFRP enclosures for electronics thermal management

Modern military aircraft rely on an increasing number of electronic assemblies to achieve maximum performance and improve platform capability. A typical assembly consists of a collection of electronics, mounted on printed circuit board (PCB), and encased in an enclosure. These assemblies provide critical functions such as engine control, actuator control, and power distribution management. Failure of these electronics would result in reduced aircraft performance, at best, and potentially render the aircraft inoperable.

Along with the increased use of electronics come increased waste heat loads. Since electronic failure rises exponentially with operating temperature, management of this heat load is as essential to aircraft operation as flight-worthiness. Providing sufficient thermal management, however, is becoming increasingly difficult for several reasons.

Avionics designs are calling for increased power density to reduce platform mass and volume. In addition, designers are integrating multiple components into single, multifunction units. In some instances, these components operate in high-temperature environments that result in a significant amount of ambient heat gain. All of these factors lead to increased component heat flux. Without improvements in thermal management, higher heat flux results in larger temperature gradients between the heat source and sink, which leads to higher component temperatures.

Carbon fiber reinforced polymer (CFRP) has the potential to provide significant weight savings in aircraft but suffers from poor thermal conductivity. CFRP materials typically have thermal conductivities on the order of 5 W/(m·°C), while carbon steel is near 50 W/(m·°C) and aluminum is near 200 W/(m·°C). Currently manufactured CFRP enclosures are restricted to low heat load applications. The fabrication process employed produces a low thermal conductivity material that precludes any other use at this time.

Recently, engineers from Advanced Cooling Technologies and TE Connectivity developed a method for moving heat through a relatively low thermal conductivity composite structure to an active sink located on the exterior of the avionics package. Multiple heat-pipe-based assemblies were developed and demonstrated as part of a fully functional PC/104 avionics package. The PC/104 is a rugged form factor used in a wide variety of applications, including both Tier II and Tier III UAVs, as well as military ground vehicles.

A new CFRP enclosure was developed for the PC/104 form factor that was scalable for various card requirements. PC/104 circuit cards stack vertically, each one plugging in to the next to assemble all the required hardware for the particular embedded computer application. The enclosure design followed this stacking approach, so that the height of the box could be customized for the number of cards to be included for the particular application. Each slice consisted of four CFRP walls and four corner pieces. Typically one wall will have pass-through connectors to attach input and output cables.

A modular thermal management system (TMS) was designed to transport heat from within the enclosure through the poorly conductive CFRP walls to an external sink. The approach focused on developing a generic solution for the box, rather than specific solutions for particular circuit cards and chip placements.

The design consisted of constant conductance heat pipes (CCHPs) that penetrate through the CFRP walls. The CCHPs are capable of transporting significant heat energy through a limited cross-section with a minimal temperature penalty.

The modular design allowed for the TMS to be included only on slices that dissipate significant power. The TMS may be employed on one, two, or three walls, again depending on how much power is dissipated by the particular card in that slice. In this way excess mass of TMS is saved when it is not necessary. Mass of the TMS should be carefully considered, since weight reduction is the primary motivating factor to use CFRP in the first place. If the TMS results in a box that weighs the same or more than simply using an aluminum enclosure, then the entire purpose would be lost.

The TMS for the CFRP enclosure consists of three subsystems. The first subsystem is inside the box and transports heat from the highest power chips to the walls. An aluminum plate with embedded heat pipes, called a HiK plate, is secured to the motherboard where fins and a fan would typically be. The heat pipes extend from this plate and are attached to the walls of the box slice.

The second subsystem consists of the heat pipes embedded in and jogged through the composite walls. These heat pipes collect heat from the inside of the box and transport it through the wall. Pipes that are not directly coupled to a HiK plate may have fins attached to facilitate in collecting the heat from the air within the box.

The third subsystem is the external method for removing the heat. The testing demonstration used an aluminum fin stack with a fan that was held solidly to the wall with the exposed wall heat pipes. Depending on the conditions and requirements in an actual application, this external heat sink could be swapped out for a liquid-cooled cold plate, thermoelectrics, or a natural convection fin stack. To allow access to the internal electronics the slices cannot be permanently attached together. Allowing the external heat sink to be removed and reattached easily is necessary to maintain the modularity of the computer and enclosure design.

Thermal analysis of the design was performed using the commercially available FEA software CFDesign. Power was applied to a 0.5 x 0.5 in square on the PCB supported within the enclosure. No thermal interface resistances were modeled. The maximum dissipation from any of the cards in the test system was 32 W. Because the motherboard fills two slices, 16 W or half of the total dissipation was used as the applied power for this analysis.

Two baseline cases were run on a slice with no TMS. The first was with the CFRP box. Heat was transferred from the power application site via natural convection to the walls, which were then cooled on the outside by additional natural convection. Some heat was also conducted to the walls via the CFRP supports. This analysis predicted a maximum temperature on the PCB of 173°C. A second baseline case used the same geometry but the material properties were changed from those of CFRP to aluminum. The aluminum baseline analysis predicted a maximum temperature on the PCB of 154°C.

The manufacturer provided temperature limits on these electronics as a maximum allowable of 85°C, and the power supplies were de-rated above 60°C. So, while applying all 16 W of power on only 0.25 in² may be overly conservative, the baseline results indicated that even an aluminum box needs additional thermal management. This was confirmed by the fact that existing aluminum PC/104 enclosures often include finned exteriors and internal heat spreaders or fans. When performing mass analysis to compare weight savings between the CFRP and aluminum, the mass of the CFRP TMS is partially cancelled out by the similar TMS required in an aluminum case.

The TMS was then modeled. Aluminum blocks were placed around the internal CCHP to spread the heat across the evaporator and condenser sections. A second CCHP was jogged through the wall and kept flush on both sides. On the exterior of the wall a thin aluminum fin stack was modeled with fin spacing configured for natural convection.

The analysis predicted a maximum temperature on the PCB for this configuration as 58°C. These results were promising, since they demonstrated a solution that keeps the temperature below the 60°C point; here, the power supplies were de-rated.

However, the absolute hot spot temperature should not be taken as definitive because of the assumption made for the PCB heat load geometry and because interface resistances were neglected. The more reliable conclusion from these results is the comparison between the TMS analysis and the baselines. There is a 115°C reduction in the maximum PCB temperature from the baseline CFRP case and 96°C reduction from the aluminum baseline case.

This article is based on SAE International technical paper 2014-01-2189 by Andrew Slippey and Michael Ellis, Advanced Cooling Technologies Inc., and Bruce Conway and Hyo Chang Yun, TE Connectivity.

Continue reading »