Detailed Descriptions:

Datacenters

Use of DC power [top]

LBNL is working with over 20 firms in a demonstration to show that large energy savings are possible by eliminating multiple conversions of alternating and direct current that typically occurs in today's data centers. The objectives of this demonstration were to show that DC powered servers could readily be built from existing components and reliably provide the same level of functionality as their AC powered counterparts. Energy efficiency gains were expected by eliminating the power losses that occur during power conversion. A traditional AC distribution system was set-up along side of the new DC distribution system at a Sun Microsystems facility using components loaned by industry partners (including DC lighting). Intel and Sun Microsystems modified servers to directly accept high voltage (380V) DC power. Both systems were fully instrumented in order to measure electrical power losses at various points in the system and preliminary results are indicating energy savings of 10-15% within the electrical distribution system. This saving is compounded with additional saving through reductions in the cooling that is required — typically about the same amount — so the result is an overall facility level saving. Estimates of data center energy use nationally have been placed at 10-15 TWh so implementing this strategy could be significant. Large numbers of data center professionals are viewing the demonstration in open house tours and have expressed interest in seeing the technology adopted. [more information]

[top]

"Air Management" Improvement [top]

IT equipment housed in data centers today rely on air cooling. In most cases, air distribution in the centers involves mixing of cooled air with air that has been heated by the IT equipment making it difficult to supply the cool air to where it is needed and resulting in inefficient heat transfer to the cooling system. LBNL demonstrated that large energy savings are possible by completely isolating the cold and hot airstreams. The objective of the study was to demonstrate feasibility and quantify savings when certain isolation strategies were implemented. The test configurations were implemented in LBNL's Oakland Scientific Facility housing the National Energy Research Scientific Computing (NERSC) supercomputers. In one arrangement where "cold aisles" were completely isolated using temporary partitions, six fold energy savings in fan energy were measured. In another, where the "hot aisle" was isolated, three fold fan energy savings were observed. In both schemes, additional efficiency gains of the chilled water system were calculated based upon the ability to raise the temperature of the cold air supply since undesired mixing of hot and cold air streams was eliminated. This demonstration is important because it shows the resulting energy savings along with other non-energy benefits of more uniform cooling which has reliability implications.

[top]

Encouraging Use of Outside Air Economizers in Data Centers [top]

In many parts of the country, there are many hours per year where the outside air conditions are such that IT equipment could be cooled by directly supplying outside air thereby minimizing or eliminating the use of chillers. There are a number of data centers that routinely use outside air economizers but there also are a large number of centers where there is a belief that bringing in large amounts of outside air will cause problems. Common concerns are that contamination levels would increase and cause failures of IT equipment or that humidity control would be a problem. LBNL is addressing this by measuring contamination levels in both centers that use outside air and those that don't. Particle counters are used to measure concentrations of various size particles inside and outside of the centers. Simultaneously, humidity is being measured to see if there are issues concerning humidity control. The primary objective of this demonstration is to obtain some evidence that use of outside air is either not a concern, or to develop recommendations to mitigate any problems that are noted to provide CA public utilities with technical information for their customers. A secondary objective is to quantify the potential annual savings a center that currently uses air economizers in order to encourage broader use. Preliminary results are encouraging in that low concentrations of contaminants have been found in both operating scenarios and no humidity control problems have been identified. A total of 6 or 7 centers in various climate zones will be studied. Research is also being planned, in collaboration with ASHRAE, to investigate the types of contaminants that could cause damage to IT equipment in larger concentrations. [more information]

[top]

Cleanrooms

Demand-controlled Filtration in Cleanrooms [top]

LBNL has developed a methodology for controlling cleanroom airflow based upon contamination levels in the room. Using commercially available particle counters, fan speeds are directly controlled by sensing particle counts in real time (rather than full-time, full-speed ventilation based on little more than rules of thumb.) Since fan energy varies with the cube of fan speed, small changes in fan speed will lead to large changes in fan energy. In a pilot study, LBNL implemented the strategy in a 300 ft2 ISO class 5 cleanroom, measuring particle concentrations using multiple particle counting instruments while changing recirculation system fan speeds. The results validated our expectation that DCF can save energy, i.e. higher fan speeds (step curve to left) do not necessarily mean lower particle counts (jagged curve to left). There may be an optimum recirculation fan speed that is unique to each facility and/or processes occurring in each facility. Following the pilot study, the strategy was demonstrated in an industrial cleanroom where again, large energy savings were realized through lower air flow when contamination levels were low. Other control strategies using timers and occupancy sensors were also demonstrated and similarly showed the potential for energy savings. In a previous study we estimated that implementing DCF had a payback time of 1 to 4 years. [more information]

Characterizing Cleanroom Minienvironments [top]

Cleanrooms have extraordinarily high rates of energy use in part because they are traditionally configured to maintain ultra clean conditions over very large areas. However, sensitive processes requiring high cleanliness levels are only carried out in relatively small areas within the larger cleanroom. “Minienvironment” technologies—essentially a cleanroom within a cleanroom—can be used to isolate these sensitive processes, thereby reducing the risk of contamination. LBNL field tests have shown that—if cleanliness conditions are appropriately relaxed in the main cleanroom through the use of minienvironments--significant energy savings can also result. In addition, we found that the efficiency of minienvironments varies widely, depending largely on the quality of the fan filter unit used to provide filtered air to the minienvironment. Simply installing a minienvironment does not guaranty energy savings. [more information]

Laboratories

Automated Fumehood Sash Closure [top]

LBNL identified an emerging technology for reducing fume-hood energy use by up to 75% through installing an automatic sash closure system on a VAV (variable air volume) hood that is controlled by an occupancy sensor. The strategy was tested in two laboratories at UC Davis. The value of energy saved ranged from $3400 to $4600 per year per hood, including about 2kW of peak demand for hood assuming electric space cooling. The extra cost of implementing the strategy would be paid instantly by downsized HVAC in new construction. [more information]