Advanced Computing in the Age of AI | Friday, March 29, 2024

Bursting through Barriers to Better Cooling 

<img style="float: left;" src="http://media2.hpcwire.com/dmr/watercooling.jpg" alt="" width="95" height="71" border="0" />Whether you are cooling manufacturing equipment, your data center, or even a nuclear power plant, the search for better, more efficient ways to keep temperatures down is an ongoing one. But a recent breakthrough could help liquid cooling of all types, across an expansive list of applications.

Whether you are cooling manufacturing equipment, your data center, or even a nuclear power plant, the search for better, more efficient ways to keep temperatures down is an ongoing one. But a recent breakthrough could help liquid cooling of all types, across an expansive list of applications.

Recently, a team of MIT researchers completed the first investigation of the factors that control boiling heat transfer from a surface to a liquid. And if you’re asking what that means for you, you wouldn’t be alone. But as it turns out, the process is essential for efficiency in power plants and cooling high-power electronics, and it could even shed light on how to better design objects to move through water.

The study looked into a transition point called the critical heat flux, or CHF, which is the value of heat transfer per unit time and area, where a surface’s heat-transfer properties suddenly change, typically resulting in damage or destruction of the system in question. Now, researchers believe they have found a key to raising CHF, which means that they can augment the safety margin or operating ranges for equipment in such high-temperature environments.

If that's still a bit abstract, the video below from the research team shows just what happens when a wire reaches the point of critical heat flux.

Previously, the trouble was that while researchers had identified key surface attributes affecting CHF as roughness, “wettability” (how easily water spreads over the surface) and porosity, they had limited data on which of these properties had the greatest significance. But the work of Jacopo Buongiorno, associate professor of nuclear science and engineering, along with his team shows that surface porosity has the greatest effect on heat transfer.

Buongiorno explained that the reason for the previous ambiguity stems from researchers in previous studies manipulating multiple surface parameters at the same time, making it impossible to isolate the properties of each one. But once the MIT team was able to independently manipulate the three attributes, Buongiorno says they gathered “some surprising results.”

The team’s research branched off of one of their previous studies that examined nanofluids, or nanoparticles suspended in water, for use in nuclear-plant cooling systems. What they found was that the nanoparticles significantly raised the CHF, and thus the safety in the plant—but they couldn’t explain why.

Co-author Michael Rubner, the TDK Professor of Polymer Materials Science and Engineering at MIT, says that when Buongiorno “indicated that enhancements in CHF appear to be related to the deposition of nanoparticles onto surfaces, we got excited since we had developed methodologies for systematically depositing nanoparticles onto surfaces with nanoscale control over thickness, wettability and porosity. Using these methodologies, we were able to produce well-defined surface characteristics and structures that made it possible to sort out the important factors at play in the process.”

With these new tests in their arsenal, the team was able to determine that the nanoparticles form a hydrophilic porous coating on the metal’s surface, which attracts the cooling liquid to the object’s hot surface. Buongiorno says that the conventional wisdom from researchers in the past has been that wettability alone was the major player in boosting CHF.

“It was the multidisciplinary team that allowed for this finding,” says Tom McKrell, co-author on the paper that was recently published in the journal Applied Physics Letters. He added that without tapping into the team’s expertise in surface nanoengineering, surface characterization and thermal hydraulics, the team’s CHF findings “would have remained a mystery.”

If you’re compute hardware relies on liquid cooling, this could be big news for performance and efficiency. But not everything would benefit from a higher CHF, and Buongiorno and his team have a solution for those items too. For instance, if you’re looking to reduce drag on an object moving underwater, a low CHF is the way to go.

To make this possible the team also studied how a hydrophobic (instead of hydrophilic) porous coating would lower the CHF. This research represents the process at the opposite end of the spectrum of CHF, called quenching.

This can be observed when a cold liquid is introduced onto the surface of a hot object, such as an overheated fuel assembly in a nuclear plant, or when a red-hot piece of metal is put in cold oil to fine tune its microstructure. In these cases, the liquid becomes a vapor barrier that effectively insulates the object’s surface.

Buongiorno explains that in cases such as these, the metal could be “so hot that when you put water on it, it wouldn’t touch it.” If this is undesirable, a hydrophilic porous coating will prevent the barrier from forming in the first place. But if that barrier is beneficial, a hydrophobic coating could actually add a buffer.

EnterpriseAI