Print 19 comment(s) - last by imm2114.. on May 23 at 1:54 PM

The Oak Ridge National Laboratory waste heat generator relies on thousands of tiny piezoelectric cantilevers, similar to this one, produced at the École Polytechnique Fédérale de Lausanne (EFPL).  (Source: EFPL)
Tiny cantilever generators could recycle as much as 30 percent of the heat chips put off

The Oak Ridge National Laboratory located about 23 miles south west of Knoxville, Tennessee has a storied history as the home of the Manhattan Project and birthplace of the atomic bomb.  Today it has its sights set on many more tiny objectives, only this time it hopes to provide the world with cleaner energy, rather than destructive power.

The scientists at ORNL estimate that billions of dollars is lost every year in waste heat from electronic devices like central processing units (CPUs).  Server farms spend millions a year to simply cool down their fields of racked computers.  Waste heat is also a major source of energy loss in automobiles.

To that end, a team led by Scott Hunter has developed [press release] tiny piezoelectric generators that use a tiny cantilever, approximately 1 mm^2 in size.  Approximately 1,000 of these tiny generators can be affixed to a 1 inch^2 CPU.  Each tiny generator pumps out 1 to 10 milliwatts -- and together they can produce enough electricity to partially power fans or system sensors.

While that might not sound like much, it's actually significant progress.

Scientists have long hoped to use piezoeletric generators to recycle waste heat, but past designs have struggled with efficiencies, only able to convert between 1 and 5 percent of the waste heat to electricity.  By contrast Professor Hunter's design has an efficiency of between 10 to 30 percent.  The exact efficiency is dependent on the temperature of the waste heat generator (computer chip).

The tiny generator works using widely known physics principles.  The tiny cantilever, anchored to the piezoelectric substrate bends due to the bi-material effect when heated.  This is the same effect that traditional room thermostats rely upon.  

As the cantilever bends it touches the heat sink, cooling it.  It then flops back down on the hot surface.  This flopping creates vibrations in the piezoelectric material that creates an alternating flow of current, which can be converted to usable DC current.

Describes Professor Hunter, "The tip of the hot cantilever comes into contact with a cold surface, the heat sink, where it rapidly loses its heat, causing the cantilever to move back and make contact with the hot surface. The cantilever then cools and cycles back to the cold heat sink. The cantilever continues to oscillate between the heat source and heat sink as long as the temperature difference is maintained between the hot and cold surfaces."

The key to higher efficiencies was to pick the right material and the right physical design.  Professor Hunter comments, "The fast rate of exchange in the temperature across the pyroelectric material is the key to the energy conversion efficiency and high electrical power generation."

This kind of piezoelectric device is part of a growing class of devices known as micro-electrical-mechanical systems (MEMS).  Professor Hunter envisions his new MEMS device being installed inside PC and server coolers, saving businesses and home users money on their power bills.  

By improving the efficiency of electricity usage, the devices would also reduce fossil fuel usage.  Comments Professor Hunter, "In the United States, more than 50 percent of the energy generated annually from all sources is lost as waste heat, so this actually presents us with a great opportunity to save industry money through increased process efficiencies and reduced fuel costs while reducing greenhouse gas emissions."

As the device is still pretty expensive, presumably, Professor Hunter is eyeing high-performance systems as the first target for commercial deployment.  The tiny generator would allow additional cooling not currently available with a traditional block water-cooled system.  By offering greater cooling to the chip, high performance mainframes could be run at higher speeds than ever before, a perfect proving ground for the new MEMS device.

Nickolay Lavrik, Thirumalesh Bannuru, Salwa Mostafa, Slo Rajic and Panos Datskos were among the other researchers who worked on the device for ORNL.  The project was the the Laboratory Directed Research and Development program under the auspice of the U.S. Department of Energy.

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

Anybody Else...
By DougF on 5/17/2011 2:34:22 PM , Rating: 5
Read this as: "National Lab Produces Highly Efficient Waste Heat Generators" and think: "Wait..what? Why would a national lab produce waste heat generators, let alone highly efficient ones?"

Maybe: National Lab Produces Highly Efficient Generators that use Waste Heat...hmmm?

RE: Anybody Else...
By tng on 5/17/2011 4:30:57 PM , Rating: 2
Got me as well, the title is confusing, but it is accurate.

RE: Anybody Else...
By chagrinnin on 5/17/2011 4:43:37 PM , Rating: 3
I thought people were gonna start poopin' into some new Amish heaters in the wintertime. :P

RE: Anybody Else...
By wordsworm on 5/18/2011 1:19:42 AM , Rating: 2
I think you're taking it the wrong way. It would be like saying 'gas generator.' The generator uses gas to make electricity. In the same way, a waste-heat generator would make electricity using waste-heat. Perhaps adding a hyphen might make it a bit easier for some readers.

RE: Anybody Else...
By Fritzr on 5/18/2011 5:50:18 PM , Rating: 2
Not really since a gas generator IS a device that generates specific gases. It is only when it is read in context that you make he selection of generator of gas or gas powered generator. The second is more common so the word powered is simply assumed until context says otherwise

A Waste Heat Generator in the context of the article is a Waste Heat powered generator. Actually it is a waste heat powered electricity generator :P

The article clears up the ambiguity right at the beginning so even the laziest of readers should not be confused by it :)

By Connoisseur on 5/17/2011 12:32:04 PM , Rating: 2
If the piezoelectric motors are meant to be situated directly on top of the CPU, wouldn't it DECREASE the heat exchange rate for the overall system? It's another layer between the heatsink and the CPU. Therefore, CPU's would get much hotter since the exchange rate is lower.

Maybe i'm thinking about this incorrectly...

RE: Confused...
By JasonMick (blog) on 5/17/2011 12:46:49 PM , Rating: 2
If the piezoelectric motors are meant to be situated directly on top of the CPU, wouldn't it DECREASE the heat exchange rate for the overall system? It's another layer between the heatsink and the CPU. Therefore, CPU's would get much hotter since the exchange rate is lower.

I would assume that the idea here is that the devices are small/thin enough that they don't create a significant barrier to standard heat flow.

Still you raise a valid point.

I suppose that if you crafted these into tiny pockets inside the chip casing, you might be able to do this without significant damage to the heat wicking capabilities of the cooling block.

Sounds expensive, which is why they're targeting performance systems first...

RE: Confused...
By nafhan on 5/17/2011 1:41:40 PM , Rating: 2
Is there a reason they have to be between the heatsink and the CPU? It seems like these could be attached to a heatsink, allowing the heatsink size (and number of generators) to be as large as necessary without being tied to the size of the CPU die.

RE: Confused...
By DanNeely on 5/17/2011 2:40:08 PM , Rating: 2
How well they work is dependent on the temperature gradient. The farther from the heat source you get the less efficient they'll be.

RE: Confused...
By DanNeely on 5/17/2011 1:25:14 PM , Rating: 3
If I'm following the writeup correctly (and it's correct) the cantilevers are switching from one position to the next depending on their temperature. This would require them to transfer their heat to the heatsink before swinging back to the CPU.

The question then becomes how efficient they are as thermal conductors vs typical thermal paste. If they're even in the ballpark then using enthusiast class heatsinks on servers instead of oem class models might be able to overcome the difference. Since data centers are increasingly limited not by the number of racks they have space for, but buy the amount of power they can provide and heat they can dissipate; dropping the power consumed per blade by a few watts could pay for an extra $20-30/cpu by letting them cram more computers into the data center without having to expand to a larger building.

Problem solved
By Motoman on 5/17/2011 12:57:42 PM , Rating: 4 all we need to do is hook up one of these puppies to a few dozen Pentium 4s, and the world's energy problems will be solved. Sweet. I love it when a plan comes together.

RE: Problem solved
By BadAcid on 5/17/2011 1:20:43 PM , Rating: 2
I thought Superman gathered all of those up and threw them into the sun.

RE: Problem solved
By Chapbass on 5/17/2011 1:26:35 PM , Rating: 3
Pshhhh, the sun would only cool these suckers down.

RE: Problem solved
By TSS on 5/17/2011 6:08:19 PM , Rating: 2
Did you know when people say the sun runs on fusion they really mean the fusion of 2 pentium 4's? I heard they decided to call it the Pentium D.

solar thermal power?
By kattanna on 5/17/2011 2:05:23 PM , Rating: 2
if they are really able to reach 30% efficiencies i wonder if they would be able to use it for generating power via concentrating solar arrays. the heat produced would be quite intense.

price though im sure is an issue, at least it was alluded to in the article.

RE: solar thermal power?
By tng on 5/17/2011 2:20:29 PM , Rating: 2
Good idea, but this really is not a solid state device. The Bi-metal interface that transfers heat would probably prove to be a major wear item in the system. Metal fatigue will be an issue I think.

home of the Manhattan Project?
By Azethoth on 5/17/2011 10:33:52 PM , Rating: 2
Try Los Alamos, NM. It is situated on a mesa near Santa Fe and Albuquerque and per requirements west of the Mississippi and more than 200 miles from international borders. Oak Ridge is semi famous, but absolutely not for housing the entire centralized scientific research project into nuclear weapons during WWII.

RE: home of the Manhattan Project?
By Iaiken on 5/19/2011 4:03:34 PM , Rating: 2
Really? That's what you were taught? Sounds like a simplified history for imbeciles...

The Manhattan Project was largely disseminated as part of it's security. While most of the fission theory was developed at Los Alamos, it was verified and built upon, in pieces, from places ranging from the University of Southern California to the University of Toronto.

Most of the theory work was completed by 1942 and Los Alamos became more of a center for administration and testing until the end of the war. The majority of the remaining effort was expended at places like the Clinton Engineer Works (now Oak Ridge National Laboratory) and The Hanford Site, which endeavored to enrich uranium and breed plutonium for use in the actual weapons themselves.

By all accounts, the procurement of the fissile material at the Oak Ridge and Hanford sites accounted for roughly 80% of the total science and engineering effort.

Other Technologies
"I mean, if you wanna break down someone's door, why don't you start with AT&T, for God sakes? They make your amazing phone unusable as a phone!" -- Jon Stewart on Apple and the iPhone

Copyright 2015 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki