Outdoor enclosures for housing electronic circuit boards are widely used in a variety of technologies, including telecommunications, industrial and military applications. These enclosures protect the equipment against a wide variety of environmental hazards, such as sun, moisture, dust and debris. As electronic components have become more powerful and complex, power dissipation from them keeps rising and thermal management has become a critical issue. In addition, solar loading further complicates this problem depending upon the size of the enclosure, surroundings and orientation with respect to the sun. Ignoring the effect of the sun can result in excessively high internal enclosure temperatures causing equipment reliability problems or even failure.
A large variety of cooling techniques have been proposed and used to cool outdoor electronic enclosures. These include conventional techniques, ranging from passive natural convection to the use of commercial air conditioners or heat pumps and concepts using thermosyphons and phase change materials (PCMs). The internal heat is transferred primarily by convection to the inside surfaces of the enclosure by conduction through the walls of the enclosure and then by convection to the external ambient.
CFD SIMULATION CONFIGURATIONS
Figure 1 shows a schematic of typical thermal resistances related to an enclosed box with internal heat generation. For such a heat dissipating box, CFD simulations were done with a commercial code  dedicated for electronics cooling. A total of 12 different cooling configurations were simulated. Out of these, nine configurations were created by making combinations of three types of commonly used airflow management options and three types of enclosure surface coatings. Besides these three, special configurations were selected, including a solar radiation shield, double-wall enclosure and a heat exchanger; using a typical aluminum enclosure measuring 300 x 300 x 400 mm.
There were eight printed circuit boards (PCBs) inside, each dissipating 12.5 W uniformly, bringing the total internal power dissipation to 100W. The distance between each PCB was 30 mm. The PCB measured 240 x 180 mm and 3 mmthick. There was a gap of 35 mm between the extreme PCBs and the side faces of the enclosure. Three types of enclosures were selected – the first was completely sealed; the second was sealed with internal circulation fans and the third had vents to allow air exchange with the outside.
Three varieties of outside coating were examined, including white, black and no coating (plain aluminum finish). These coatings were selected based on their radiation characteristics, i.e., solar absorptivity (α) and radiation emissivity (ε). The white oil coating had a low value of α (0.25) and a high value of ε (0.91), making it very favorable for cooling under solar heat loads. The black coating had a high value of α (0.88) and a high value for ε (0.88). The plain aluminum finish had low values for α and ε (0.08, 0.09). In order to minimize the effect of solar loads, various options were analyzed, including employing a radiation shield, a double-walled enclosure with air circulation and a heat pipe based air-to-air heat exchanger (Figure 2). Table 1 summarizes all enclosure configurations considered for evaluation.
The sealed enclosure with internal fans case was modeled as two fans placed above the PCB assembly and oriented in suction mode pulling the air through the PCB assembly slots. The fan selected was 80 x 80 x 20 mm and had a maximum flow capacity 0.022 m3/s and a maximum pressure capacity of 61Pa. The enclosure with louvered vents case was modeled as two opposite enclosure walls perpendicular to the PCBs having cut-outs at the bottom and top side. The double walled enclosure had an air gap of 20 mm between the inner and outer walls. A fan was placed at the top of the outer wall sucking the air between the inner and outer walls through four openings at the bottom of each outer side wall (Figure 3). For this case, the fan selected was 120 x 120 x 25 mm, having a maximum flow capacity 0.053 m3/s and a maximum pressure capacity of 107 Pa.
For the case with radiation shield, an umbrella was used to shield the sunlight. In the simulation, this was modeled with zero solar loads on the enclosure and no radiation heat transfer from the top wall. The heat exchanger system consisted of two parallel heat pipes with condenser ends coming out of the enclosure (Figure 2). Fins were attached to the heat pipes inside the enclosure as well as outside the enclosure. Fans were used to blow air through both sets of fins. Fans selected for this case were 80 x 80 x 20 mm and had a maximum flow capacity 0.022 m3/s and a maximum pressure capacity of 61Pa. Figure 2 also shows a schematic of the air circulation pattern and the heat flow from inside the enclosure to the outside.
Simulations were run with solar heat load as well as without. For the cases with solar load, the maximum load was assumed to be 600 W/m2 for Pune City (India) in March, which is one of the hottest months of the year. The sun’s rays were assumed to be incident on three adjacent surfaces – the top surface and two sides. This works out to a total solar load of anywhere between 0 and 200W, depending upon the solar absorptivity of the outer surface. This shows that the solar load can be of the same magnitude as the internal thermal load.
CFD SIMULATION RESULTS WITHOUT SOLAR LOAD
Initially, a set of simulation runs were carried out without any solar heat loads. Figure 4 depicts the simulation ∆T values for middle PCB surface. The air inside the enclosure follows essentially the same trend (see  for more details).
It can be seen that, by just having an internal circulation fan, the internal temperatures can be significantly reduced. It is also well below the configuration having vents. In fact, it can be seen that vents have a relatively small impact on the temperatures. Having a black or a white coated surface can also be very effective. The case having a white coated enclosure with heat exchanger had the lowest ∆T values. In this case, a significant amount of heat is dissipated to the outside through the heat exchanger. In the cases of double-walled enclosure and radiation shield there isn’t any improvement since there was no solar load.
CFD SIMULATION RESULTS WITH SOLAR LOADING
For cases with a solar heat load, the ∆T values for the middle PCB are almost 20% higher for the sealed black enclosure (see Figure 5), showing that solar loading can be substantial for outdoor enclosures. Similar trends were observed compared to the cases without solar loading. The results again show that just having an internal circulation fan can significantly reduce the internal temperatures. Also, a black or preferably white coated surface can be very effective for cooling compared to a plain aluminum finish. This is because the white coating has a very favorable combination for low solar absorptivity and high radiation emissivity. The temperatures were the least in the case of the air-to-air heat exchanger. It was found that, in this case, nearly 60 W (40 %) were dissipated through the heat exchanger.
To validate the CFD results a mock-up of the system was built and tested. Three types of enclosures were built similar to the CFD modeling configurations: a perfectly sealed enclosure, one with louvered vents, and a sealed enclosure with internal fans. Each configuration was further tested with three types of coatings: white, black, and a plain aluminum finish (Figure 6). Experiments were conducted for all configurations without solar load and for one BC_SL configuration with solar load at Pune (India) in March, in the afternoon (solar load of 600W/m2). Figure 7 represents schematic of the enclosure set-up and the temperature measurement locations.
The enclosures were constructed using 1 mm thick aluminum sheets assembled with nuts and bolts. The PCBs were made using resistor coils sandwiched between two commercially available copper clad boards. The PCB assembly was placed inside the enclosure suspended from two rods. The thermocouples and DC source wiring were taken out through an opening created on the top side of the enclosure. The power supplied to the enclosure was 100 W through an external DC power supply and was kept constant for all the cases. To monitor the temperatures a data acquisition system (DAS) was used. Temperatures were measured on the surfaces of the middle and extreme PCBs, and also the enclosure air temperatures were measured at the bottom and at the top of the PCB assembly using T-type thermocouples. For the test case with internal fans, the fans were placed above the PCBs and oriented in the suction mode pulling air through the PCB slots
It is seen that the tested ∆T values are quite close to the CFD results for all the tested configurations without solar load (see Figure 4). For a configuration with a solar load (BC_SL) the test result for temperature rise of the middle PCB over ambient is 65ºC against a simulation value of 70ºC. It should be realized that it is extremely difficult to get a reliable match between experimental and numerical results for a complex electronic system because it is essentially impossible to solve all boundary layers around all objects of interest to a degree that the physics are captured correctly. Furthermore, many properties that influence the analysis are not sufficiently well known: thermal conductivities of the PCB, emissivities and absorptivities, air resistance coefficients, etc. (see ).
It was found from this study that the effect of solar heat load on an outdoor system can be quite significant and can increase the internal air temperature by 20%. Different cooling approaches for outdoor electronic enclosures were analyzed and compared. The results indicate the relative effectiveness of these different cooling solutions. Relative to the sealed enclosure without any coating it was found that a black coating, or preferably a white coating, on the outside enclosure wall is a very simple and effective cooling solution that can reduce internal temperatures by around 25%. It was also found that having vents did not reduce the temperatures as significantly as having internal circulation fans. In fact, there is around a 50-55% reduction in the ∆T due to the internal fans compared to a sealed enclosure with no fans. Having a radiation shield and a double-walled enclosure with air circulation provided relatively modest improvements of around 25%. The most dramatic improvement was almost 75% in case of the air-to-air heat exchanger.
The authors of this article thank Dr. Sukhvinder S. Kang for his valuable review of the paper and suggestions during the work.
Wankhede, M., Khaire, V., Goswami, A., Mahajan, S.D., “Evaluation of Cooling Solutions for Outdoor Electronics,” Therminic-2007.
Estes, R.,“Thermal Management of Telecommunication Cabinets,” ElectronicsCooling, September 1997, Vol. 3, No. 3.
Nevelsteen, K., et al., “Thermal Modelling and Experimental Evaluation of Outdoor Cabinets,” Thermal Challenges in Next Generation Electronic Systems, Joshi & Garimella (eds), Millpress, Rotterdam, 2002.
Hendrix, M., “Cooling Design of a Sealed Optical Network Unit (ONU) Enclosure”, Fujitsu Network Communications, Richardson, TX, 57082, http://www.efluid.com.cn/soft/soft_detail.aspx?id=4002.
Holman, J. P., Heat Transfer, Mc-Graw Hill Publication, 1990.
Lasance, C., “CFD Simulations in Electronic Systems: A Lot of Pitfalls and a Few Remedies,” ElectronicsCooling, May 2005.