The chillers in large commercial, industrial, and institutional cooling systems account for 13% of the power consumed by America’s buildings and 9% of the nation’s overall demand for electric power. To reduce the demand, Dr. Mark A. Kedzierski seeks to improve chillers’ efficiency by adding nanoparticles of substances, such as copper oxide or diamonds, to the refrigerant inside the chillers.
Kedzierski is a mechanical engineer with the National Institute of Standards and Technology (NIST), a branch of the US Department of Commerce that, until 1988, was known as the National Bureau of Standards.
“If nanofluids improve chiller efficiency by 1%, we could save 320 billion kilowatt-hours of electricity a year—the equivalent of 5.5 million barrels of oil—and that’s just buildings,” he says. “It doesn’t include the military, which uses chillers to meet a variety of cooling loads on its ships.”
Kedzierski works in the Heating, Ventilating, Air Conditioning & Refrigeration Equipment Performance Group, part of the Building Environment Division at NIST’s Building and Fire Research Laboratory in Gaithersburg, MD.
He hopes that his research will improve the operating efficiency of existing chillers, and help manufacturers of refrigeration and air-conditioning equipment improve their products—by making smaller and less-costly, new chillers as efficient as those in use today, or chillers comparable in size to present models that operate more efficiently and economically.
How a Chiller Works
A chiller operates on the same principle as an aerosol can. As you spray out the can’s contents, less liquid remains inside so the pressure of the remaining contents goes down and their temperature drops, cooling the metal of the can. If the can gets really cold, then you also might have evaporative cooling, with the liquid stealing heat from its surroundings and from your hand, and actually boiling inside the can.
A commercial chiller’s basic function is to steal heat from cooling water by boiling refrigerant in a device called an evaporator. A chiller evaporator is a large shell housing hundreds of copper tubes, each about three-quarters of an inch in diameter and arranged in a bundle. The chiller’s compressor maintains a low refrigerant pressure in the tube bundle, just like an aerosol can, to provide chilled cooling water that circulates through the building. As the refrigerant in the evaporator boils, it extracts heat from the cooling water.
Chiller tubes are designed with what Kedzierski describes as “passive enhancements” to create more surface area on which boiling can occur. The tubes in some older chillers have exterior striations that look like threads on bolts. More modern chiller tubes have fins, or treelike protuberances with trunks and limbs. Sometimes these protuberances are sliced and flattened to form a canopy above the fin-tips. In addition to increasing the surface areas of the tubes, the trees and canopies also create cavities for bubbles to grow and shelter for “tube superheat.”
All fluids have an equilibrium—the saturation point—that is determined by a combination of temperature and pressure. “To initiate and sustain boiling requires temperatures elevated above the saturation temperature,” says Kedzierski. “The game is to reduce that elevated temperature—to try to do the same amount of boiling with less superheat.”
Kedzierski describes his goal in terms of entropy, a measure of irreversibility or disorder. “Entropy is the monster,” he says. “It’s what keeps me employed. We talk about reducing energy consumption, but what we really mean is reducing entropy consumption, making the process more efficient.
“The whole trick in efficiency is to have smaller heat exchangers and lower approach temperatures,” he adds. “A chiller that operates so it boils a mixture of refrigerant and lubricant with low-approach temperatures while very close to freezing—at about 3˚C [37.4˚F]—has to be very efficient.”
Experimental Design
In his experiments, Kedzierski uses a test apparatus with flat plates instead of round tubes. “I sacrifice a bit of reality for a lot of certainty,” he explains. “Inside a chiller bundle there is chaos, lots of boiling, but it would be very difficult to measure what’s going on. In my test rig, water flowing across the bottom of a flat plate heats the plate, and the liquid refrigerant boils on a flat surface.
“Sometimes I cut an actual tube along its axis so I can open it up, flatten it out, and put it on a thick piece of copper,” says Kedzierski. “Then, I insert a matrix of 20 thermocouples [temperature-sensing devices that detect the electromotive force—the potential—between dissimilar metals] to measure, non-destructively, the temperature of the plate and the heat flux.”
The purpose of adding copper oxide nanoparticles to the refrigerant/lubricant mixture is to reduce the amount of wall superheat—the difference between the temperature of the plate and the saturated liquid temperature—needed to make enough refrigerant vapor to initiate and sustain the cooling process. “The ultimate goal is to get that to zero,” adds Kedzierski. “It can never be achieved, but approach temperatures of 0.5˚C [32.9˚F] are not unheard of, and it may be even less.”
The Recipe
The copper oxide nanolubricant that Kedzierski uses in his experiments is formulated to his specific requirements by Nanophase Technologies Corp., of Romeoville, IL. It’s a synthetic product containing a gigantic molecule called a polyolester. “It lasts longer, you don’t have to change it as often, and it doesn’t break down,” says Kedzierski. “Also, it’s pure; you know what’s in it because you made it. You didn’t pull it out of the ground and filter it. It contains fewer impurities that can get attached to the compressor parts.”
The nanolubricant doesn’t boil. It accumulates on the tubes, in a “lubricant excess layer” about 40 microns thick.
Kedzierski’s nanolubricant recipe also includes a surfactant to keep the copper oxide particles from clumping together. “Any time you cook and burn something on the stove or weld something, you’re making nanoparticles of carbon that we call soot,” he explains. “The trick is to get these particles very small and introduce them into the fluid one at a time so they stay discrete. If you dump a powder of nanoparticles into the liquid, they will stick together in big clumps and settle, so people have devised ways to add them one at a time.”
Discrete particles in suspension move in a random fashion, a phenomenon known as Brownian motion after the English botanist Robert Brown, who, in 1827, peered through a microscope and observed the movement of pollen grains suspended in water. For the copper oxide particles in the nanofluid to maintain their Brownian motion, they must be small enough to remain dispersed as discrete particles, and they must be suspended in a very viscous fluid.
The surfactant improves the stability of the nanoparticles in the nanolubricant, just as soap improves the stability of dirt in water. “If you try to wash oil off of something with just water, it doesn’t work,” says Kedzierski. “Oil is immiscible; it doesn’t go into solution with water. Soap dissolves the oil because a soap molecule is polar; one end likes the oil and not the water, and the other vice versa. Soap molecules encircle the oil molecules, so the water doesn’t see the oil; it just sees water.”
When Kedzierski received the nanolubricant for one set of experiments from Nanophase Technologies, the copper oxide nanoparticles comprised 40% of its volume. Before testing this preparation, he diluted it and combined it with refrigerant in six different strengths: 2% nanolubricant by volume, mixed with the refrigerant in 0.5%, 1%, and 2% mass fractions; and 4% nanolubricant by volume, also mixed with the refrigerant in 0.5%, 1%, and 2% mass fractions. Thus the percentage of nanoparticles by volume in the final mixtures ranges roughly from 0.01% to 0.08%.
The excess lubricant layer creates a slippery surface, reducing the surface energy of the heat-transfer surface, increasing the number of bubbles, and reducing their size. “The bubbles’ hold-down force is reduced,” says Kedzierski. “With lubricant, they don’t have to get very large before they become buoyant and rise off the boiling surface. Like the lubricant, the nanoparticles don’t boil. They congregate in the excess lubricant layer just above the tube surfaces. If you get enough nanoparticles in the excess lubricant layer, they may further increase the thermal conductivity.”
The refrigerant Kedzierski used in this experiment was R134a, a high-volume, medium-pressure fluid suitable for applications ranging in size from small automotive air-conditioners to large chillers for buildings. “It’s good to test, because it’s applicable to a large percentage of the chillers that are out there and being built,” he says.
The Results
Compared to the boiling heat transfer of the refrigerant/lubricant mixture without nanoparticles, the 4% volume fraction nanolubricant mixed at a 0.5% mass fraction with the refrigerant improved boiling heat transfer by 50% to 275%, while the improvement was more modest for larger quantities of the 4% nanolubricant in the refrigerant: 19% better for the 1% mass fraction mixture with the refrigerant and only 12% better for the 2% mass fraction mixture.
The results of this experiment may seem counter-intuitive, but Kedzierski says they simply follow the laws of physics. “Compared to boiling a pure refrigerant, lubricants make the bubbles smaller but more numerous,” he says. “The increase in the number of bubbles causes an enhancement of heat transfer to the power of one, but the reduction in bubble size causes heat transfer to go down to the squared power. As you add more lubricant, bubble diameter always wins. Because the bubble size effect is squared, if you have an enhancement of heat transfer, its maximum always will happen at the lower concentrations of lubricant. When you add less, you get more improvement.”
This means that in a chiller equipped with oil management, the operator should adjust the oil separator to favor the light end of the mass-fraction spectrum, between 0.5% and 1%. “Some systems aren’t that sophisticated; you get what you get,” says Kedzierski.
By contrast, the 2% nanolubricant mixture at all three mass-fraction levels yielded little or no improvement in boiling heat transfer. “I didn’t put enough nanoparticles in the 2% nanolubricant,” says Kedzierski, suggesting that some initial threshold of nanoparticle content is necessary to achieve significant boiling heat transfer.
Three Heat-Transfer Mechanisms
Kedzierski believes that three distinct mechanisms affect the way in which nanoparticles influence heat transfer:
- Improved thermal conductivity due to highly conductive nanoparticles in the lubricant, especially those in the concentrated nanolubricant excess layer that coats the heated boiling surfaces. This mechanism accounts for about 20% of the boiling heat transfer enhancement, Kedzierski estimates.
- Boiling enhancement due to nanoparticles interacting with bubbles. This accounts for the remaining 80% of the overall increase in boiling heat transfer. It may take the form of increasing “primary nucleation” (bubble formation) at the heated boiling surface, “secondary nucleation” (one bubble growing on top of another), and individual bubbles growing larger as they interact with hot nanoparticles.
- Degradation of boiling heat transfer as nanoparticles in the lubricant fill nano-sized cavities on the heated boiling surface, causing a decrease in the surface area available to support boiling. This would explain what happened with the 2% nanolubricant mixtures, Kedzierski explains. “An overall improvement in the boiling heat transfer will result if the enhancement due to nanoparticle interactions more than compensates for the boiling heat transfer degradation as caused by the filling of boiling cavities with nanoparticles,” he says. “After the nanoparticles go into every little nook and cranny, these cavities are no longer available to make bubbles. If you put in just enough nanoparticles so the surface gets saturated and no particles are left over to interact with the bubbles, you’ll see a heat-transfer degradation. If you put in enough nanoparticles to compensate for the loss of cavities because they’re full of nanoparticles, you’ll have an enhancement. You have to pay for some of the degradation with the particles. Then, what’s left over can be used for enhancement.”
Potential Hazards
The foregoing is just one experiment in a series to discover what kinds of lubricants, nanoparticles, and refrigerants provide the best heat-transfer and boiling combination. Among other possibilities, Kedzierski has tested a refrigerant/lubricant mixture of R123 and naphthenic oil (which didn’t change heat-transfer performance), and a lubricant that contains diamond nanoparticles (which may offer heat-transfer benefit in smaller concentrations than copper oxide).
Other NIST engineers are working on engine nanolubricants that increase lubricity through a submicroscopic ball-bearing effect.
Although the possible mixtures and percentage combinations of refrigerants, lubricants, and nanoparticles are vast, if not infinite, Kedzierski remains undaunted by his task. “I’m confident we can find nanolubricants that are compatible with chillers and compressors,” he says. “The key is to find one that improves heat transfer without creating a hazard. Chiller manufacturers are rightfully conservative. They’re like physicians; their motto is, ‘First, do no harm.’”
Potential hazards from nanoparticles fall into two categories: hazards for the public and hazards for the equipment. “A nanoparticle in a lubricant is encapsulated,” says Kedzierski. “Only when it becomes airborne, can it cause a potential problem, and that depends on what the nanoparticle is made of. If it’s a carbon nanoparticle with a twisty shape that resembles asbestos, that’s not good. There is no data to show that copper oxide or diamond nanoparticles are harmful, especially considering that they are encapsulated in lubricant.
“For the equipment, if nanoparticles are small enough and remain discrete, the equipment won’t even know they are there,” he adds. “If they agglomerate over time and fall out of suspension, they could clog the system. This is a prevention issue. We would have to do enough tests and make sure the particles stay in suspension over time, and then ensure that they would have a minimal effect on the system even if they all agglomerated. These are design issues, which require more data.”
Technology Transfer
Kedzierski has been contacted by venture capitalists who want to sell nanolubricants to chiller operators as aftermarket products, but he would prefer to see the companies that make chillers and chiller lubricants offer such products themselves. “I’ve been contacted by two chiller lubricant manufacturing companies,” he says. “They spend a lot of money tweaking and playing with the chemistry of their lubricants to reduce noise in chillers, improve lubricity so you don’t have to put as much in, and reduce foaming. Maybe nanoparticles would be one of the things they include. This may happen a lot quicker than normal technology transfer does, due to the increased pressure today for energy conservation combined with the improved comfort zone for nanotechnologies.”
As a federal researcher, he emphasizes that his role is to aid in technology transfer by conducting experiments and reporting their results through presentations at scientific meetings and publication of professional papers. “My goal isn’t to do the chiller manufacturers’ job,” says Kedzierski. “I stay in business by staying out of their business. I look at effects, present my research as a tool to the chiller industry, and see if they think it’s interesting enough to apply.
“I’m very proud of the refrigeration and air-conditioning industry,” he adds. “They are the benchmark for reliability. Air conditioning in a large building is always there, but the manufacturers are strapped for research funds because their industry is very competitive. They need my help to bring them along in terms of high technology. They don’t have the resources to get into it.”
Kedzierski has had requests for field tests of his most promising nanolubricants from chiller and power plant operators who seek to reduce the costs of running their equipment, but he eschews field tests in real chillers, where he can’t control their internal operation.
“Field tests are yucky,” he says. “We stay away from them, because it’s very difficult to study effects—what happens when you add nanoparticles or change refrigerants—where you can’t control the conditions. It’s like predicting the stock market. You just don’t know what’s going on.”
Nonetheless, he may undertake a two-year field test soon in a southern office building. “This one fellow wants me to put nanoparticles in one of his end-of-life chillers,” says Kedzierski. “It’s 15 years old and not meeting load. You turn the thermostat down, and you’re still hot; the chiller has gotten old. Also, there’s more demand. His load has increased; he now has twice as many computers and people in the building as he did before. To meet the load, he’s faced with buying new chillers or trying nanoparticles.
“This, for me, would be a good demonstration of whether these particular nanoparticles are compatible with the chiller,” continues Kedzierski. “I’m confident that they are, but anything that comes from this field test would be just a talking point when I go to the manufacturers.”
References
Kedzierski, Mark A., Effect of CuO Nanoparticle Concentration on R134a/Lubricant Pool Boiling Heat Transfer, MNHT2008-52116, Proceedings of MNHT2008, Micro/Nanoscale Heat Transfer International Conference, Tainan, Taiwan, January 6-9, 2008.
- A nanometer is 39.37 quadrillionths (0.000,000,000,000,003,937) of an inch.
- Viewed a different way, one inch is 25.4 million (25,400,000) nanometers.
- A single particle of copper oxide is 30 nanometers in diameter—almost 1,200 quadrillionths of an inch (which equates to 1.2 quintillionths or 0.000,000,000,000,000,000,012 of an inch).
By comparison, the thickness of a piece of paper is about 40,000 nanometers, or 157.48 trillionths (0.000,000,000,157,480) of an inch.
