Things get heated in tight quarters, whether it is people crowded in small spaces or miniaturized electronic components packed in products. It’s a particular problem with all the modern electronics in today’s devices and goods, as it can lead to degraded performance, shorter lifespans or outright failure. It’s also an environmental issue — all those electronics are generating heat that dissipates to the surroundings. Plus, extra heat is produced in the process of cooling these systems.
We’re investigating novel cooling techniques to remove heat from confined spaces so innovations in industry sectors like information technology, cloud computing, avionics, and electric vehicle batteries and motors can continue apace while reliability, efficiency and performance increase. Our Cooling Technologies Research Center (CTRC) focuses on longer-term R&D in the area of high-performance heat removal from compact spaces, guided by input from industry partners.
For example, supercomputers require innovative thermal management. A major performance limitation of these machines is how much heat one can take out of the chips inside of them. We’re working with companies that design liquid cooling systems for supercomputers to improve a system’s effectiveness and get more heat out of the system more efficiently. Using liquid instead of air as a cooling agent can remove hundreds of times the amount of heat per amount of surface area. These liquids even can be evaporated to absorb the heat from the device.
Additive manufacturing, or 3D printing, plays a major role in these advances. When you pump air or liquid over a component, you need material in the pathway of the coolant — in the shape of a fin, say — to extract the heat. We’re using 3D printing to create new geometries outside the capabilities of conventional manufacturing, and we’ve developed groundbreaking design approaches and algorithms to form the twisting and turning shapes of the fluid-filled channels.
Simulation is vital here to assist in a design process called topology optimization. Imagine you’re molding the shape of the cooling channels like a piece of clay, a little here and there, trying to enhance performance. Using simulations, we quickly can optimize shapes to create the geometry we want, then 3D-print and test the components to confirm the performance predictions. We’re developing novel algorithms that extract information from the simulations to make design decisions that account for constraints on fabrication.
Thermal management has become a bottleneck in many systems, and innovations will be held back if we don’t address these limitations. The more we miniaturize electronics, packaging them closer and closer together so they can communicate faster and operate more efficiently, the more intense the heat generation inside shrinking spaces. To put it in perspective, the amount of heat generated in future supercomputer chips could be as high as 1 kilowatt in the volume of a cubic centimeter — 10 times that of your desktop computer in the space of a sugar cube. We have demonstrated that we can liquid-cool the cube with thousands of internal channels to extract that kilowatt. The intensity of heat generated at the surfaces of these electronics can compare to that coming off the surface of the sun!
Being able to remove heat from tiny spaces is the key to advancement in many other industries that depend on electronics, which have almost singlehandedly powered the information revolution and a robust U.S. economy for decades.
Justin A. Weibel, PhD
Research Associate Professor
Director, Cooling Technologies Research Center
School of Mechanical Engineering
College of Engineering, Purdue University