In today’s fast paced society, technology changes before consumers have time to ask what is coming next. Within days of purchasing a new phone, the next generation will be announced as being twice as fast, with a better, sleeker appearance. This constant introduction of new products gives the illusion that designing the next revolutionary device is just that easy; it seems that as soon as an engineer has an original idea, it is moments away from being implemented and marketed. In reality, the research and design phase for technology usually takes years and includes a process that is more intensive than coming up with a clever idea.
At UW-Madison, the work done by Timothy Shedd, a professor of mechanical engineering, to create a liquid cooling system for electronics that significantly reduces energy consumption is an example of this research and design process. The results from this liquid cooling system research led Shedd and two former UW-Madison mechanical engineering graduates to found Ebullient LLC, an engineering design company that will commercialize these systems. Reaching this point took 11 years of design and testing by Shedd and his team, following a process that could never have been predicted from the onset of the project.
Shedd, originally from Springfield, Missouri, received his B.S. in electrical engineering from Purdue University and his M.S. and Ph.D. in mechanical engineering from the University of Illinois at Urbana-Champaign. After defending his Ph.D. thesis in 2001, he became an assistant professor of mechanical engineering at UW-Madison. It was here that his project received its first stroke of luck. After class one day, Shedd was confronted by mechanical engineering undergraduate Adam Pautsch about a project he had worked on as an intern for Cray, Inc. Pautsch’s project involved developing a liquid cooling technique for computers that would allow the computers to run faster while possibly reducing the energy consumption of the system. “As a professor, finding a motivated, knowledgeable student like this is more valuable than a pile of gold,” said Shedd. He had an interest in Pautsch’s project since it related to the topics of multi-phase flow and heat transfer, which he had previously researched, so he decided to meet with Cray, Inc. to propose the continuation of this research. The partnership was meant to be; Shedd received a one-year IEDR grant to begin work on this project and soon the research and design process was underway.
For the next five years, Shedd continued to research liquid cooling systems to cool super computers and other electronics that require less energy. During this time, he submitted two patent applications related to the spray and drainage system he and his students had designed for the coolant system. The success of this research prompted Shedd to begin researching how his heat removal system impacts the energy efficiency of the computer system as a whole, not just the performance of the chips. He created a numerical model for a large computer system data center, such as those used by Google or Facebook, to find the impact of this efficiency on a larger scale. This step ensured that Shedd’s research was economical; due to the eliminations of some of the air conditioners and/or water chillers the current systems use, his cooling method would save up to 30% of the energy consumed by these computer systems.
At this point, Shedd knew it was time to understand the intricacies of his system. He began to research the liquid heat transfer process and discovered that by controlling the temperature and pressure of a specific region of a hot surface, a process similar to boiling could be induced by pointing a jet stream of refrigerant at this region. It is important to note that the bubbles did not form on the hot surface directly, as normally occurs when liquids boil. Shedd formed a hypothesis that very small bubbles (less than ten microns in size) are created by forcing liquid through a jet nozzle at the liquid’s saturation temperature (the temperature at which the liquid is in equilibrium with its vapor). As these bubbles approach the hot surface, they absorb the thermal energy coming from the surface and grow until they become vapor. This evaporation heat transfer process increases the heat transfer performance of the jet by a factor of two or three. At the same time, the liquid temperature does not rise very much because most of the heat is used to create vapor rather than to increase the liquid temperature. Shedd realized that these behaviors could be used to effectively remove a significant amount of heat from electronic devices with only a small amount of energy input required to drive the pump. Since this has definite commercial possibilities, Shedd submitted invention disclosure reports to WARF and began the patent process for these concepts.
This discovery prompted Shedd to add another researcher to his team. Bio-Chemical Engineering undergraduate Ian Tonner joined the project and began to design a prototype of this cooling system to function in a compact computer. They packed the pump, tubes and heat exchanger into a small box and connected this to an Intel Core i7 CPU. “The computer ran this way for two months with no problems,” said Shedd, “At this point we knew we were onto something.”
Simultaneous to this research project, Shedd submitted an IEDR proposal with Cray to research a way to lower the energy consumption and improve the chip-level heat transfer efficiency of Cray’s computer system. Once the proposal was accepted, Masters students Brett Lindeman and Robert Buchanan joined the project during the summer of 2011. Buchanan worked with Lindeman to develop an experimental study that tested over 2,700 combinations of flow, liquid temperatures, jet arrangements and cooling fluids.
During this testing, one of the models was successful in cooling a series of four heaters with a single inlet and outlet. Lindeman worked to create a design that could be manufactured using the University’s rapid prototyping facilities and after “trial and error, computational fluid dynamics and a lot of patience,” says Shedd, they “arrived at an extremely elegant and compact design for the jet array.”
In 2012, the research team received a SIRE grant (Sustainability Innovation in Research and Education) to apply their research to a typical computer system. The team turned to Professor Susan Hagness, who is currently researching the electromagnetic imaging and treatment of cancer using computer simulations of electromagnetic waves interacting with human tissues. Her system of about 20 computers generates between 15,000 W and 20,000 W of heat when all of the computers are running. “Imagine being in a room with 12 hair driers operating on high heat. That’s what this room is like,” says Shedd. To keep these computers cool, Shedd estimates that the College of Engineering pays at least 5,000 dollars per year to air-condition this room; Shedd’s new cooling method could reduce this to 790 dollars.
Currently, Shedd is on a year long sabbatical at the University of São Paulo in Brazil. He hopes to continue his research on fluid cooling systems while in São Paulo to utilize their “phenomenal research labs,” and is also working to form an exchange program between the University of São Paulo and UW-Madison. His plans for future applications include the cooling of electronics in hybrid vehicles, wind turbines and eliminating the heat sinks in variable speed drives. Shedd’s research has now evolved beyond UW-Madison labs and will continue to inspire engineers around the world to create the most efficient electronic liquid cooling system that can be designed.