Thermal Management of Microscale Electronic Equipment Using Pulsating Heat Pipes

Abstract

Last two decades witnessed a massive growth in the electronics industry. Electronic equipment such as mobile phones and computers are now an integral part of our daily life. The ease of use has driven the electronic industry to develop more powerful yet smaller electronic products. There is an ongoing quest for developing powerful yet miniaturise electronic equipment. While in use, a piece of electronic equipment generates excess heat, and the amount of heat generation increases with the increase in products performance and miniaturisation. The excess heat if not checked, can render the device useless (untimely system crash) or can decrease its reliability. Therefore, the thermal management of an electronic device is necessary to keep the device in its optimal use state without any failure. There are many thermal management techniques in use; this thesis focuses on one such technology called pulsating heat pipes (PHP). PHP is a relatively new technology in electronic cooling space; however, due to its straightforward design, the ability to handle high heat fluxes and simple manufacturing has made it a very promising electronic cooling technology. This research project aims to analyse the performance of PHP as an electronic cooling device. For achieving the aims and objectives of the study, two test rigs were designed and fabricated. The first test rig was a single loop pulsating heat pipe (SLPHP), and the other one was a multi-loop pulsating heat pipe (four loops). A PHP takes some time to reach its operation state, often referred to as quasi-steady state. Before reaching a quasi-steady state, a PHP goes through a start-up process. An efficient thermal management technique must initiate its operation as soon as the electronic device reaches a certain peak heat flux level. Also, the cooling device must keep the operating temperature of the electronic equipment below its maximum permissible temperature. Though there are studies regarding the startup of a PHP, however, none of them are for PHP as a electronic cooling device. For understanding the start-up process of a PHP, a single looped pulsating heat pipe was fabricated from a 670 mm long copper tube with an internal and external diameter 3.25 mm and 4.75mm respectively. Water was used as the working fluid; experiments were conducted for different filling ratios (50 %, 60%, 70%, and 80%) and heat loads (20 W, 33 W, 55 W and 76W). Two types of start-up behaviour were observed, a gradual (soft) start-up and a sudden (hard) start-up, as reported previously. Contrary to previous claims, the two types of start-ups were found to be independent of the heat load. Experiments revealed that the probability of the PHP undergoing a sudden start-up was higher at a higher filling ratio, which was attributed to the higher level of flooding of the evaporator section at higher filling ratios. Experimental results also show that for similar heat load and filling ratio, during sudden start-up time is more as compared to smooth start-up. Besides, after a sudden start-up, the PHP works at a higher operating temperature as compared to a gradual start-up. Therefore, the sudden start-up is not desirable for a PHP based device employed for electronic cooling. It was also shown that, to suitably design a thermal solution (based on PHP) for electronic cooling, the calculation of thermal performance must include start-up temperature and start-up time. The multi-loop PHP experimental setup was designed to evaluate the performance of the PHP under non-uniform heating conditions. In electronic cooling, there are situations where non-uniform heating could occur. For example, a data centre houses a large number of processing units each running at different operating load, generating a different level of heat fluxes. Another example is a high performing laptop or desktop in which, along with CPU, graphic card, and RAM might require cooling. Furthermore, the surface of the microprocessor itself can have different levels of heat fluxes at different locations. Though, each equipment or unit has its cooling device; however, a single device working under different heat load could save much space which could allow further miniaturisation and can reduce the operation cost of the cooling device. The studies about non-uniform heating of a PHP are scarce. Two critical parameters in a PHP operation are heat input and filling ratio (FR), however, the studies on the effect of non-uniform heating on the performance of the PHP at different filling ratio is limited to a narrow range (50% -70%). Furthermore, the mechanism of how non-uniform heating affect PHP performance at different filling ratio is not yet clear. For studying the effect of non-uniform heating, a four-loop water-based CLPHP (closed loop pulsating heat pipe) was fabricated from a copper tube that had an internal diameter (ID) = 3.25mm. The thermal performance of the CLPHP was measured over a range of filling ratios (FR, 20% to 80%), total power input (80 W to 360 W), and two-stage non-uniformity with a power difference (ΔQ̇=Q̇₁− Q̇₂ , 40 W to 140 W). Results show that the thermal resistance of the PHP under non-uniform heating increase on increasing ΔQ̇. The study also revealed that the impact of non-uniform heating is significant at low FR (<20%); however, at high FR (> 50%), the influence of non-uniform heating diminishes. Non-uniform heating was also found to affect the operational ceiling of the PHP. For example, in this study, the optimal filling ratio during uniform heating was 20%, and no dry outs were observed. However, during non-uniform heating, dry outs were recorded for FR 20% at various heating levels, which resulted in very high thermal resistance. Investigation of flow patterns during non-uniform heating revealed the reason for the unusual trend of thermal resistances at different filling ratios. The flow pattern evolution during a non-uniform heating condition was found to be different from the flow transitions during uniform heating. Independent localised flow patterns were observed for low filling ratio whereas at high filling ratio high heating side was found to govern the flow pattern across the PHP. The optimal filling ratio during uniform heating was to be 20%, which was changed to FR 60% for non-uniform heating. A dimensionless non-uniform heating coefficient (δ) is also proposed to quantify the effect of non-uniform heating.