Energy and water environmental trade-offs of data center cooling technologies.
The last decade has seen much research into the energy consumption of ICT (Koomey 2009) and data centers (Koomey 2011). This research has shown levels of consumption and facilities growing well into the future as more of the world comes online. This consumption comes mostly from IT equipment and the mechanical services used to support them. The data center industry has reacted to this growth, and improvements to energy efficiency have been well documented (Tozer et al. 2015). However, whilst there are sound scientific methods for improving efficiency, such as air management (Tozer et al. 2015), there is still scope for further optimization.
In recent years, the concern with energy consumption has moved beyond the financial implications to its broader impact on the environment. Many metrics have been established to understand these impacts, such as PUE, WUE and CUE (power/water/carbon usage effectiveness), but they consider different subjects in isolation. This means that in improving PUE, an operator could be shifting the environmental burden to WUE (through additional water consumption). It is difficult for non-environmental experts to understand the true impact that this shift has on the environment, because energy and water have different units of measurement. There is therefore a need for a method of assessment that allows data center designers, owners and operators to understand these trade-offs.
THE ENVIRONMENTAL IMPACT OF ENERGY AND WATER CONSUMPTION
Every human activity creates environmental impacts, which vary in intensity, severity and longevity, and may be positive or negative. Until recently the emphasis of the environmental impact assessment of energy was related to emissions from operation, because the majority of energy was from fossil fuels. The most significant emissions from fossil fuel combustion include [CO.sub.2] and other Greenhouse Gases (GHG), sulphur dioxide [SO.sub.2], nitrogen oxides [NO.sub.x], particulates and substances such as mercury. These substances are emitted to the air, soil and water and can create numerous diverse impacts; examples include human health problems (ranging from the short term/seasonal to death), and damage to and depletion of species (resulting from changes in pH levels of rain and sea water). The recent development of renewable technologies means that many operational emissions and impacts have been reduced and/or eliminated. A comparison of the operation alone is misleading, however, because the inputs, outputs and impacts of the generation technologies differ, and it is therefore necessary to assess these 'embodied' impacts. These are far more numerous and diverse and there can be more than 700 associated with any individual type of technology.
While energy is important for human activity, the importance of water cannot be underestimated as it is essential for life and although it is abundant, only 2.5% is fresh and only 1% of that is accessible for direct human use; furthermore the water system is closed, i.e. there is a finite quantity in the ecosphere. Population growth and related land occupation and use, agriculture and irrigation and industrialization mean that demand is continually increasing. Water quality is affected by natural phenomena such as leaching mineral deposits and sedimentation while changes in climate and weather patterns impact on rainfall patterns and levels of surface runoff and absorption. The impacts of human activities also affect water quality and while many supplies have been polluted with pathogens from sewage, toxic chemicals and pesticides from cultivation and industrial processes, other reserves have been depleted and not replenished. These numerous inter-related factors have led to global 'water stress' and cost implications.
LIFE CYCLE ASSESSMENT
Life cycle assessment (LCA) is a systematic tool for assessing iteratively the impact that a product, process or service has on the environment (Baumann et al. 2009). LCA is used to assess the environmental impact from raw material extraction (cradle), transportation, the manufacturing and use of the product, and its eventual disposal (grave). There are four stages to an LCA: the goal and scope (including system boundaries); life cycle inventory (flows entering and leaving the system); impact assessment (inventory results are translated into environmental impacts); and interpretation (the consequences of the above environmental impacts).
Flows describe materials, energy, emissions or products entering or leaving a system. There are three common flows: elementary, product and waste flows. Elementary flows are untreated materials or energy, whilst product and waste flows are as their names suggest. At the inventory stage, flows into (inputs) and out of (outputs) the system are inventoried for a given product (process). Once the inputs and outputs have been quantified, they are classified according to the environmental impacts they contribute to, i.e. [CO.sub.2] contributes to climate change. A characterization method then uses cause-effect chains to quantify the relative contribution the emission/material consumption has on environmental impact. For example the global warming potential (GWP) of GHGs can be expressed as [CO.sub.2e]. Using damage models, the relative impact these environmental phenomena have on areas of protection (AoPs), namely human health, ecosystem quality and resources is then calculated, and a weighting applied to provide a single score.
LCAs can be completed to varying degrees of accuracy, from screening to 'full-blown' process-based (detailed) studies (Rebitzer et al. 2004; Baumann et al. 2009). In a screening LCA process-based life cycle inventory (LCI) data from previous studies is used to approximate the environmental impact. By reducing the precision of the study, users can understand the general pattern of impact in a short period of time, and identify potential areas for improvement.
THE ENVIRONMENTAL TRADE-OFF BETWEEN ENERGY AND WATER CONSUMPTION
Goal and Scope
Using a screening LCA, the goal of this study was to provide equations for assessing the environmental trade-off between different cooling technologies based on their water/energy consumption and location. The work aims to understand the general pattern of impact and provide a basis for more detailed investigations. The functional unit of the study is the provision of cooling for one year. The system boundaries include all inputs and outputs from cradle to grave for the water and electricity, but no impacts embodied in the technology itself. These should be investigated in future work.
The electricity data is country-specific, and includes: the electricity production in the relevant country from material extraction; the transmission network; direct SF6-emissions to air; electricity losses during LV transmission (including the HV transmission from the grid); and the transformation between voltages at switching stations (Dones 2007). The assumptions for the transmission network and emissions are based on Swiss data. A European average tap water process was used to represent the water entering a data center. The process included infrastructure and energy use for water treatment and transportation to the end user, but no emissions from water treatment. The process was compiled with estimated data for a water works in Switzerland and energy use in Germany. The data was then adapted using the Pfister et al. (2009) impact factors for regionalization to give an impact based on the water stress index found in each country. It is assumed that for every [1m.sup.3] of water used, 50% is evaporated and 50% requires end treatment. A generic sewage water treatment has been used. Both water processes are likely to underestimate the true impact and should be the subject of future research. All datasets were secondary, based on average technologies, and had a reference year of early 2000. The Eco-indicator 99 method was used to characterize the results in a freely available LCA software tool.
Water Scarcity and Regionalization in LCA
The environmental impact of water consumption is dependent on its origin (country or regional level) and source (ground/surface water, rivers and lakes). Until recently, differentiation between LCA results at a regional level has been difficult because characterization methods did not include indicators and impact factors for water scarcity. There are now a number of methods that give an impact at the midpoint level (no indication of the environmental phenomena that the impact causes). These include: characterization by water stress index (Pfister et al. 2009); ReCiPe (Goedkoop et al. 2009); human health impacts (Boulay et al. 2011); and water footprinting (Hoekstra et al. 2011). However, Pfister et al. (2009) is the only method that provides an impact at the endpoint in all three AoPs (human health, ecosystem quality and resources) for the Eco-indicator and ReCiPe methods, and was therefore used. Because these impact factors were not included in the OpenLCA software, the impact factors for freshwater consumption were applied manually to the inventory results in order to understand the impact of regionalization on the results.
LCA RESULTS AND INTERPRETATION
The results of the LCA are given in Eco-indicator points (Pt). 1,000 Pt are equivalent to the environmental load of one average European in one year. Figure 1 shows the levels of water and energy consumption required for a given location to yield the same impact of 100,000,000 Pt. A WUE/PUE of 6 is assumed to represent a worst-case legacy facility (see Table 2), and a line has been added to the graph. Assuming WUE and PUE values from Table 2, Table 1 below shows what size facility this WUE/PUE value relates to in each country. In Sweden a 16MW IT load, using 2 million [m.sup.3]/yr (5.28 x [10.sup.8] gal/yr) of water creates the same impact as a 22kW IT load and 3000 [m.sup.3]/yr (7.93 x [10.sup.5]gal/yr) of water in the US. Note, for this example, electricity and water loads are taken for the whole facility.
The most important result is therefore the impact the energy mix has on the environmental impact of electricity consumption. The impact from energy consumption in Sweden is logarithmically three orders of magnitude smaller than that in the US (Figure 1). This is because the Swedish mix is largely reliant on hydropower and nuclear, whilst the majority of the US and UK mixes are from fossil fuels, and the French from nuclear (see table 15.1 Dones (2007)). Although the datasets were based on old grid mixes, the impact from electricity is so large, that the pattern of results is likely to remain the same until grid mixes resemble that found in Sweden.
When completing an LCA, many factors can change the results--the characterization method, emission timescales, boundaries, allocation, weighting, assumptions, and the life cycle inventory (to name but a few). Validation is therefore important. Turconi et al. (2013) reviewed 167 case studies of electricity generation LCAs with respect to GHG, [NO.sub.x] and [SO.sub.x] emissions. It found that the infrastructure provided the highest impact for renewables, and direct emissions for fossil fuels. Comparing life cycle [CO.sub.2]-eq values, coal ranged from 660-1050 kg/[MWh.sub.out] and hydropower from 2-20 kg/[MWh.sub.out]. Additional studies for coal found ranges up to 1200 kg/[MWh.sub.out], therefore assuming a mid-range impact of 1000 kg/[MWh.sub.out] for the coal, and a value at the lower end of the hydropower range of 5 kg/[MWh.sub.out], shows there is three orders of magnitude difference between the results. Although this is based only on GHG emissions, studies of the operation of a UK data center (Whitehead et al. 2015) showed that the next biggest impacts in the operation phase were from carcinogens (PAHs), respiratory inorganics ([NO.sub.x] and [SO.sub.x]) and fossil fuels. These are in abundance during the operation of fossil fuel technologies, but not renewables. It can therefore be concluded that a more detailed study would most likely find the overall life cycle impact of the fossil fuel technology to be even bigger than that of the renewable technology. This therefore supports the pattern of results found above. For more information on the science behind the characterization of these results, and the use of Cultural Theory for the weighting of results, see the Eco-indicator Methodology Report (Goedkoop 2001).
The second thing to note is the bearing this impact from energy mix has on the relationship between water and power impact. When the renewables content of the generation mix is realtively low, the water consumption impact is also relatively low. It would be easy therefore, for these countries to ignore the topic, when in reality solutions should be sought to increase the use of renewables, either on- or off-grid, whilst also limiting water consumption.
APPLICATION OF THE RESULTS
For a given country, the following equations can be used to calculate the environmental impact experienced in one year from operating a cooling system, where Pt are Eco-indicator points for the subscript location, [E.sub.cooling] is the energy used by the cooling system in MWh/year, and [W.sub.cooling] is the water used by the cooling system in [m.sup.3] /year:
[Pt.sub.UK] = 169,866[E.sub.cooling] + 42.52[W.sub.cooling] (1)
[Pt.sub.France] = 25,222[E.sub.cooling] + 42.50[W.sub.cooling] (2)
[Pt.sub.Sweden] = 28.66[E.sub.cooling] + 42.50[W.sub.cooling] (3)
[Pt.sub.USA] = 204,065[E.sub.cooling] + 42.57[W.sub.cooling] (4)
CASE STUDIES--INDIRECT AIR-SIDE FREE COOLING
Values for WUE/PUE from the data centers of a single company were used to apply the equations shown above. It is assumed that the facility has 1MW of IT, with an average annualized IT loading of 75%. It is assumed that 90% of the non-IT energy is used for cooling. It is a UK site, with a WUE of 0.527 L/kWh and a PUE of 1.167.
[E.sub.IT] = 0.75 x 1MW x 8,760hours = 6,570MWh/year
[E.sub.DC] = 6,570 x [10.sup.3] x PUE = 6,570 x [10.sup.3] x 1.167 = 7,667,190kWh/year
WUE/PUE = 0.527/1.167 = [W.sub.DC]/[E.sub.DC] = 0.452
[W.sub.DC] = 0.452 x 7,667,190 = 3,462,390 liters (9.15 x [10.sup.5]gallons) = 3,462[m.sup.3]
Assuming 100% of the onsite water use is for cooling, equation 1 is used to determine the total points in the UK:
[Pt.sub.UK] = 169,866 x (7,667 x 90/100 x (1.167 - 1)) + 42.52 x 3,462 = 1.96 x [10.sup.8] + 1.47 x [10.sup.5] = 1.959 x [10.sup.8]Pt/yr
The impact from electricity in the UK is 3 orders of magnitude greater than from the water consumption. Using equation 3, the same scenario in Sweden, however, finds the greatest impact results from the water consumption:
[Pt.sub.Sweden] = 28.66 x (7,667 x 0.9 x (1.167 - 1)) + 42.50 x 3,462 = 33,031 + 147,156 = 180,186Pt/yr
COMPARISON OF DIFFERENT TECHNOLOGIES
Using the above IT loads and characteristics (1MW IT etc.) PUE values were assumed for different cooling options. For options with chillers, values for COP (coefficient of performance) and [[eta].sub.w] (efficiency of water use) have been assumed. Accounting for the evaporation of water and additional load due to chiller COP, values for WUE (L/kWh) and WUE/PUE have been calculated as follows and are shown in Table 2 (note, for legacy systems using air cooled chillers, ignoring any water used for humidification and in the chilled water closed circuit, WUE = 0, and because chiller COP is worse, PUE is higher).
WUE/PUE = [W.sub.DC]/[E.sub.DC] and WUE = PUE x seconds/hour/specific enthalpy of water x [[eta].sub.w] x (1 + 1/COP)
Table 3 and Figure 2 show the resulting environmental impact for each option in the USA, UK, France and Sweden.
For each option, the impact from electricity provides the greatest contribution to environmental impact in the UK, LTSA and France. In Sweden, the greatest impact is from water consumption, even in the case of airside free cooling where water is used for evaporative cooling (via humidification) and humidification. The over-riding burden in countries with poor access to renewables is therefore from the electricity consumption. The difference in total water impact between each country is small (LTSA is greatest for each option). This is because the impact from the water treatment is far greater than the actual consumption itself (two orders of magnitude greater), and in the developed world, there is little difference in the way that water is extracted and transported, meaning relatively the regionalization has little impact on the results. This conclusion should be investigated for countries experiencing extreme water stress.
CONCLUSIONS AND FUTURE WORK
In countries with a high renewables grid content, water usage should be limited when choosing a cooling solution. In countries with a poor renewables mix, the approach is less straightforward. Globally, governments are looking for ways to move away from fossil fuels towards more renewables. New build data centers that are optimized only for energy rather than water consumption could find themselves having (relatively) more impact from water consumption (than electricity consumption) in the future, than if they had optimized their water consumption as well.
The quality of data also needs to be improved to understand how much the relative impact changes with current grid mix data. For example, UK energy trends for the last quarter of 2015 (DECC 2016) showed that fossil fuels accounted for 81.7% of energy consumption. Production rather than consumption datasets were used because fossil fuels accounted for 74.8% and 71.9% of the mixes respectively. Although there will be seasonal variations in the actual data, it is likely that the electricity impact has still been underestimated.
In the countries assessed, water scarcity had only a small impact on the overall result. In the case of electricity, this is because the contribution of water to the overall impact is relatively minimal. In the case of the water consumption, the majority of the impact comes from the treatment of the water. Because a generic wastewater treatment was used, this data also needs to be improved to reflect the contaminants that would be present--for example biocides, algaecides and scale/corrosion inhibitors. Further work also needs to focus on countries with poor water availability, such as UAE and Australia, to understand the pattern of this impact in more detail. It should also be noted that if the LCA results were interrogated in more depth (beyond the single score results), based on different environmental phenomena there are likely to be more differences in location selection than suggested by this study.
The work in this paper can help clients decide between facility locations and technology types. It should also focus the industry to look into ways to reduce the volume of water used and its additives, for example by designing systems that require less of these additives. It is clear that for data centers to reduce their total environmental impact, effort should be made to include more grid-based renewables, and (where feasible) on-site renewables as suggested by Sharma et al. (2010), as well as reducing their water consumption. To fully understand the most sustainable options, a cost dimension and embodied impacts from the physical technology should also be added to the selection criteria presented here.
NOMENCLATURE [E.sub.cooling] = Energy used by the cooling technology [E.sub.DC] = Energy used by the data center [E.sub.IT] = Energy used by the IT [[eta].sub.w] = Efficiency of water use [Pt.sub.country] = Eco-indicator points for the subscripted country [E.sub.cooling] = Energy used by the cooling technology [W.sub.DC] = Water used by the data center
Baumann, H., Tillman, A.-M. 2009. The Hitch Hiker's Guide to LCA. Sweden: Studentlitteratur AB.
Boulay A.-M., Bulle C., Bayart J.-B., Deschenes L., Margni M. 2011. Regional Characterization of Freshwater Use in LCA: Modeling Direct Impacts on Human Health. Environmental Science & Technology. 45:8948-8957
DECC 2016. Energy Trends March 2016. UK: National Statistics
Dones R., Bauer C., Bolliger R., Burger B., Faist Emmenger M., Frischknecht R., Heck T., Jungbluth N., Roder A., Tuchschmid M. 2007. Life Cycle Inventories of Energy Systems: Results for Current Systems in Switzerland and other UCTE Countries. Ecoinvent Report No. 5. http://www.ecoinvent.org/login-databases.html (09/06/2016)
Goedkoop M., Spriensma R. 2001. The Eco-indicator 99. A Damage Oriented Methodfor Life Cycle Impact Assessment. Methodology Report. Third Edition. http://www.pre-sustainability.com/download/misc/EI99_methodology_v3.pdf (17/06/2016)
Goedkoop M., Heijungs R., Huijbregts M. A. J., De Schryver A., Struijs J., van Zelma R. 2009. ReCiPe 2008--A Life Cycle Impact Assessment Method which Comprises Harmonised Category Indicators at the Midpoint and the Endpoint Level. First edition. Report I: Characterisation. NL
Hoekstra A. Y., Chapagain A. K., Aldaya M. M., Mekonnen M. M. 2011. The Water Footprint Assessment ManualJ: Setting the GhbalStandard. Earthscan: London, Washington.
Koomey J., Belady C., Patterson M. et al 2009. Assessing Trends Over Time in Performance, Costs and Energy Use for Servers. Oakland, CA: Analytics Press.
Koomey J. 2011. Growth in Data Center Electricity Use 2005 to 2010. Oakland, CA: Analytics Press.
Pfister, S., Koehler A., Hellweg S. 2009 Assessing the Environmental Impact of Freshwater Consumption in LCA. Environmental Science & Technology. 43(11):4098-4104
Rebitzer G., Ekvall T., Frischknecht R., Hunkeler D., Norris G., Rydberg T., Schmidt W.-P., Suh S., Weidema B.P., Pennington D.W. 2004. Life Cycle Assessment. Part 1: Framework, Goal and Scope Definition, Inventory Analysis, and Applications, Environment International, 30(5):701-720.
Sharma R., Christian T., Arlitt M., Bash C., Patel C. 2010. ES2010-90219 Deisgn of Farm Waste-Driven Supply Side Infrastructure for Data Centers. Proceedings of ASME 2010 4th International Conference on Energy Sustainability, Vol 1(2010):523-530
Tozer R., Flucker S. 2015. Data Center Energy-Efficiency Improvement Case Study. ASHRAE Transactions 121(1):298-304
Tozer R., Whitehead B., Flucker S. 2015. Data Center Air Segregation Efficiency. ASHRAE Transactions 121(1):454-461
Turconi R., Boldrin A., Astrup T. 2013. Life Cycle Assessment (LCA) of Electricity Generation Technologies: Overview, comparability and limitations. Renewable and Sustainable Energy Reviews. 28(2013):555-565
Whitehead B., Andrews D., Shah A. 2015. The Life Cycle Assessment of a UK Data Centre. The International Journal of Life Cycle Assessment. 20(3):332-349
Beth Whitehead, PhD
Robert Tozer, PhD, CEng
Deborah Andrews, PhD
Sophia Flucker, CEng
Beth Whitehead is an associate sustainability engineer and Sophia Flucker is a director at Operational Intelligence, London, UK. Robert Tozer is a visiting fellow at LSBU and managing director at Operational Intelligence Ltd, London, UK. Deborah Andrews is an associate professor at LSBU, London, UK.
Caption: Figure 1. Water and Electricity Consumption that Yields 100,000,000 Pt of LCA Impact for Different Location
Caption: Figure 2. Electricity and Water Points for Different Cooling Technologies in Each Country
Table 1. IT and Water Loads that Create a 100 Million Pt Impact (WUE=15 and PUE = 2.5) Country Facility Electricity (MWh/yr) Total Electricity (MW) USA 489.4 0.0559 UK 587.9 0.0671 France 3928 0.448 Sweden 352,755 40.2 Country IT Electricity (MW) Facility Water ([m.sup.3]/yr) USA 0.0223 2,936 (7.76 x [10.sup.5] gal/yr) UK 0.0268 3,527 (9.32 x [10.sup.5] gal/yr) France 0.179 23,570 (6.23 x [10.sup.6] gal/yr) Sweden 16.1 2,117,000 (5.59 x [10.sup.8] gal/yr) Table 2. PUE, COP, [[eta].sub.w], WUE and WUE/PUE for Various Cooling Technologies Technology PUE Chiller Efficiency of Water COP [[eta].sub.w] (assumed) Worst case legacy 2.5 3 0.33 Standard legacy 1.6 6 0.50 Water cooled (100%/yr with 1.3 0.50 cooling towers, no chillers) Indirect airside free cooling 1.177 (average energy priority) Indirect airside free cooling 1.187 (average water priority) Direct airside free cooling 1.1 Technology WUE (L/kWh) WUE/ (gallons/kWh) PUE Worst case legacy 15 (3.96) 6 Standard legacy 5.6 (1.48) 3.5 Water cooled (100%/yr with 3.9 (1.03) 3 cooling towers, no chillers) Indirect airside free cooling 0.825 (0.218) 0.701 (average energy priority) Indirect airside free cooling 0.47 (0.124) 0.394 (average water priority) Direct airside free cooling 0.22 (0.058) 0.2 Technology Comments Worst case legacy WUE = PUE x 3600 x (1+1/3)/ (2400 x 0.33) Standard legacy WUE = PUE x 3600 x (1 + 1/6)/ (2400x0.5) Water cooled (100%/yr with WUE = PUE x 3600 cooling towers, no chillers) /(2400 x 0.5) Indirect airside free cooling Site measured data (average energy priority) Indirect airside free cooling Site measured data (average water priority) Direct airside free cooling Values assumed Table 3. Environmental Impact for Cooling Technologies in Each Country Technology Location Electricity Water PUE Impact Impact (Pt/MWh) (Pt/[m.sup.3]) (Pt/gal) Worst Case USA 204065.40 42.57 (0.16113) 2.5 UK 169866.34 42.52 (0.16094) 2.5 France 25221.85 42.50 (0.16087) 2.5 Sweden 28.66 42.50 (0.16087) 2.5 Standard Legacy USA 204065.40 42.57 (0.16113) 1.6 UK 169866.34 42.52 (0.16094) 1.6 France 25221.85 42.50 (0.16087) 1.6 Sweden 28.66 42.50 (0.16087) 1.6 Water Cooled USA 204065.40 42.57 (0.16113) 1.3 UK 169866.34 42.52 (0.16094) 1.3 France 25221.85 42.50 (0.16087) 1.3 Sweden 28.66 42.50 (0.16087) 1.3 Indirect Air-side USA 204065.40 42.57 (0.16113) 1.177 Energy Priority UK 169866.34 42.52 (0.16094) 1.177 France 25221.85 42.50 (0.16087) 1.177 Sweden 28.66 42.50 (0.16087) 1.177 Indirect Air-side USA 204065.40 42.57 (0.16113) 1.187 Water Priority UK 169866.34 42.52 (0.16094) 1.187 France 25221.85 42.50 (0.16087) 1.187 Sweden 28.66 42.50 (0.16087) 1.187 Direct Air-side FC USA 204065.40 42.57 (0.16113) 1.1 Assumed PUE UK 169866.34 42.52 (0.16094) 1.1 & WUE France 25221.85 42.50 (0.16087) 1.1 Sweden 28.66 42.50 (0.16087) 1.1 Technology Location WUE (L/kWh) Electricity Water (gallons/kWh) Pt Pt Worst Case USA 15 (3.96) 4,524,895,1 4,195,1 UK 15 (3.96) 3,766,573,6 4,189,9 France 15 (3.96) 559,262,93 4,188,7 Sweden 15 (3.96) 635,564 4,188,4 Standard Legacy USA 5.6 (1.48) 1,158,373,1 1,566,1 UK 5.6 (1.48) 964,242,85 1,564,2 France 5.6 (1.48) 143,171,31 1,563,8 Sweden 5.6 (1.48) 162,704 1,563,7 Water Cooled USA 3.9 (1.03) 470,589,09 1,090,7 UK 3.9 (1.03) 391,723,66 1,089,3 France 3.9 (1.03) 58,163,345 1,089,0 Sweden 3.9 (1.03) 66,099 1,089,0 Indirect Air-side USA 0.825 (0.218) 251,377,83 230,732 Energy Priority UK 0.825 (0.218) 209,249,74 230,449 France 0.825 (0.218) 31,069,517 230,382 Sweden 0.825 (0.218) 35,308 230,367 Indirect Air-side USA 0.468 (0.124) 267,836,38 130,888 Water Priority UK 0.468 (0.124) 222,950,02 130,728 France 0.468 (0.124) 33,103,743 130,690 Sweden 0.468 (0.124) 37,620 130,681 Direct Air-side FC USA 0.22 (0.058) 132,730,25 61,528 Assumed PUE UK 0.22 (0.058) 110,486,16 61,453 & WUE France 0.22 (0.058) 16,405,046 61,435 Sweden 0.22 (0.058) 18,643 61,431 Technology Location TOTAL Pt Worst Case USA 4,529,090,3 UK 3,770,763,6 France 563,451,70 Sweden 4,824,060 Standard Legacy USA 1,159,939,3 UK 965,857,11 France 144,735,12 Sweden 1,726,410 Water Cooled USA 471,679,83 UK 392,813,05 France 59,252,426 Sweden 1,155,108 Indirect Air-side USA 251,608,56 Energy Priority UK 209,480,18 France 31,299,899 Sweden 265,676 Indirect Air-side USA 267,967,27 Water Priority UK 223,080,75 France 33,234,432 Sweden 168,301 Direct Air-side FC USA 132,791,78 Assumed PUE UK 110,547,61 & WUE France 16,466,481 Sweden 80,074
|Printer friendly Cite/link Email Feedback|
|Author:||Whitehead, Beth; Tozer, Robert; Andrews, Deborah; Flucker, Sophia|
|Date:||Jan 1, 2017|
|Previous Article:||Metering measurement challenges & monitoring of a large scale Ground Source Heat Pump (GSHP) system.|
|Next Article:||Thermosyphon cooler hybrid system for water savings in an energy-efficient HPC data center: modeling and installation.|