# Data center trigeneration with absorption refrigeration and economizer technologies.

INTRODUCTION TO ABSORPTION REFRIGERATION

Absorption chillers use heat as an energy source to regenerate chemical solutions, which enables them to produce cooling. Absorption cycles have a cooling cycle that operates between the cold heat source (e.g., chilled water) and the outdoor ambient heat sink (e.g., cooling towers), and they are driven internally by their own driving cycle that operates between a high-temperature heat source (e.g., hot water at 100[degrees]C [212[degrees]F]) and the outdoor ambient heat sink. These temperatures are related, which means that heat sources of at least 70[degrees]C (158[degrees]F) are required to drive them, limiting the available waste heat sources. Currently, in data centers, the waste heat quality is relatively low grade. The above cycle is a single-effect cycle. However, when higher heat source temperatures are used, e.g., steam at 800 kPa (116 psi), the heat rejected from the condenser of this high-temperature driving cycle can be used to drive a lower-temperature, single-effect cycle. These are double-effect cycles which have higher efficiencies, though they require higher-grade energy, i.e., at a higher temperature.

The vapor pressure of a solution, as of all liquids, is a statement of an equilibrium state (Gosney 1982). If a solution having a vapor pressure of 12 kPa (1.74 psi) were placed in a closed vessel at a pressure of 10 kPa (1.45 psi), the liquid would evaporate until both pressures were equal. Conversely, if the pressure gradient was inverted, condensation would occur until equilibrium was restored. Water at 10[degrees]C (50[degrees]F) has a vapor pressure of 12 kPa (1.74 psi) while a 27[degrees]C (80.6[degrees]F), 50% solution of LiBr in water has a vapor pressure of only 10 kPa (1.45 psi). The solution will therefore draw the water vapor from the surface. If a fresh source of water is available, the process is a continuous one. The water evaporating will cool bodies at a higher temperature; the evaporator and the vessel where the vapor is dissolved after leaving the evaporator is the absorber (Figure 1).

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

The absorption refrigeration cycle has the following components in common with the vapor compression cycle, evaporator, expansion valve, and condenser (Figure 2).

Absorption chillers are heat-operated machines (heat is supplied to the generator) that require only a small amount of electrical power (internal pumps) compared to mechanical compression systems (Alefeld and Radermacher 1994). The heat sources cover a range of possibilities including direct-fired combustion of natural gas, steam, hot water, and also exhaust gases. The potential to use waste heat makes them particularly interesting to integrate into total-energy systems and cogeneration (generation of both heat and power, also known as combined heat and power [CHP]). Trigeneration is the generation of heat, power and cooling, also known as combined cooling, heat, and power (CCHP).

[FIGURE 3 OMITTED]

Absorption cycles operate on the principle of refrigerant being absorbed by a chemical solution, normally LiBr/water or water/ammonia. A chiller schematic is detailed in Figure 3. The top part of the unit is the driving cycle, where heat is supplied to the generator to boil off refrigerant that is condensed in the condenser. It also concentrates the solution (say, LiBr/water) and supplies it via the solution heat exchanger to the cooling subcycle (evaporator and absorber), enabling it to absorb more refrigerant. The main heat exchangers in the following diagram are as follows:

1. High-temperature heat source: generator Qg (i.e., steam, hot water, combustion)

2. Medium-temperature heat sink: absorber Qa and condenser Qc (i.e., cooling tower condenser water)

3. Low-temperature heat source: evaporator Qe (chilled water)

Absorption cycles can be associated with heat-driven cooling cycles, such as the case of a heat-driven power cycle that drives a mechanical compression refrigeration system (Herold et al. 1996). Heat is supplied from a high-temperature heat source (generator or boiler) and from a low-temperature heat source (evaporator), and all this heat is released to a medium-temperature heat sink (absorber and condenser).

Figure 4 represents a general heat-driven cooling cycle schematically, and indicates the heat flows, thermal sources, and sinks involved.

The driving cycle consists of the generator and condenser. Its purpose is to produce refrigerant and prepare the solution so that it can absorb refrigerant in the cooling cycle. The cooling cycle, which is constituted by the absorber and evaporator, absorbs the refrigerant producing the cooling effect and dilutes the solution before returning it to the driving cycle. The product of the driving cycle is the heat of solution, which is used by the cooling cycle. The heat of solution is also referred as exergy of solution (chemical exergy). Carnot cycles (using only the concepts of heat and work) were originally conceived prior to the concept of exergy. Exergy is a more generalized concept of work, which therefore makes these ideal driving and cooling subcycles found in absorption systems equivalent to Carnot cycles (Kotas 1995).

The ideal absorption cycle is constituted by two ideal Carnot cycles, one driving and the other a cooling cycle (Nesselmann 1932; Stierlin 1964; Eber 1968). From the analysis of the differences of entropy of both cycles, the following temperature relationship was derived (Tozer and James 1994):

[T.sub.g] [T.sub.e] = [T.sub.a] [T.sub.c] (1)

The coefficient of performance (COP) is the ratio of the cooling energy to the heat source and is defined by the following:

COP = [Q.sub.e]/[Q.sub.g] (2)

The ideal COP can be expressed in terms of the refrigerant evaporation heat (r) and the heat of solution (l) (Niebergall 1959). The COP could approach unity for low heats of solution, but it can never reach unity.

COP = r/r + l (3)

By derivation of driving and cooling cycles considering the temperature relationship above, the following COP is derived (Tozer and James 1994, 1997; ASHRAE 2013):

COP = [T.sub.e]/[T.sub.a] = [T.sub.c]/[T.sub.g] = [alpha] (4)

[FIGURE 4 OMITTED]

The basic double-effect cycle consists of two absorption cycles where the heat rejected from the condenser (or absorber) of the high-temperature effect is utilized to supply heat for the generator of the low-temperature effect. This means that the low-temperature generator is also the high-temperature condenser. The temperature of the high-temperature condenser has to be higher than the temperature of the low-temperature generator to guarantee heat flow. In real equipment, the evaporators of the high- and low-temperature cycles are merged into one unit, as are the two absorbers.

The ideal COP and temperature relationships for any type of double-effect cycle are the following:

[T.sub.g][T.sup.2.sub.e] = [T.sub.c][T.sup.2.sub.a] (5)

COP = [alpha] + [[alpha].sup.2] (6)

Real-cycle COPs are 0.7 for single-effect and 1.2 for double-effect systems (Wilkinson 1987). By using the COP equations for ideal cycles in terms of two temperatures and comparing to real COPs of LiBr/water chillers, Table 1 is derived. The real COPs are typical for standardAir Conditioning and Refrigeration Institute conditions, with chilled-water temperatures of 12.2[degrees]C/6.7[degrees]C (54[degrees]F/44[degrees]F), evaporator temperature 5[degrees]C (41[degrees]F), condenser water at 29.4[degrees]C/35[degrees]C (84.9[degrees]F/95[degrees]F), and absorber temperature 45[degrees]C (113[degrees]F). The table indicates real to ideal cycles of 70% to 80%. The range of absorption chillers offered are summarized in Table 2.

If the same thermodynamic analysis is extended for E effects with heat sources at higher temperatures, the following is derived:

COP = [alpha] + [[alpha].sup.2] + [[alpha].sup.3] + ... + [[alpha].sup.E] = [alpha](1 - [[alpha].sup.E])/1 - [alpha] (7)

For infinite effects, the generator temperature needs to be infinitely high, which means that the heat source is equivalent to mechanical energy, as a confirmation of this theory. The COP becomes equal to the Carnot COP of a mechanical compression cycle:

COP = [alpha]/1 - [alpha] = [T.sub.e]/[T.sub.a] - [T.sub.e] (8)

This means that the ideal vapor compression and infinite effect absorption cycles have equivalent COPs and require the highest grade of energy (work or infinitely high temperature heat source) as driving energy.

Heat Dissipation Ratio

The heat dissipation ratio (HDR) is the ratio of heat dissipated from the cooling tower (condenser and absorber heat) with respect to the chilled water (evaporator heat) load. This relation is necessary to analyze the cooling water require ments for different types of absorption and mechanical compression (centrifugal/screw/reciprocating) chillers.

HDR = [Q.sub.c] + [Q.sub.a]/[Q.sub.e] (9)

Substituting the previous equations, the absorption and mechanical compression cycle HDR is

HDR = 1 + 1/COP (10)

The cooling tower required to dissipate the heat of the absorber and condenser in cooling cycles is larger for absorption chillers than for mechanical compression chillers (e.g., reciprocating, centrifugal, screw, scroll) because of their lower COPs. For ideal cycles, the ratio of heat dissipation ratios is given in Table 3.

In practice, the size of single and double effect absorption chiller cooling towers are 2 and 1.5 times the size of their mechanical compression counterpart cooling towers, respectively. Therefore, Table 3 can be used for initial assessments of systems. From this simple relationship it can be seen that the high COPs associated with mechanical compression systems have the benefit of lower HDRs and, therefore, a reduced cooling water requirement. Both the capital costs and operating costs of the HDRs need to be taken into account when considering the economics of absorption systems.

COLD-GENERATION CYCLES

Cold-generation systems references (Tozer and James 1995a) are those in which a cooling process is generated from a heat source and no energy is wasted (dumped). They are based on the availability of a high temperature heat source (e.g., combustion process), a low- grade temperature sink (e.g., cooling tower water), and a low-temperature heat source (e.g., chilled water). They consist of a combination of absorption and mechanical compression refrigeration, engines, and turbines, etc.; many combinations are possible.

Two cold generation cycles are indicated in Figure 5, where both systems consume high temperature thermal energy ([Q.sub.1]) from a combustion process and provide mechanical and absorption cooling at temperature [T.sub.E]. All the heat introduced at temperatures [T.sub.1] and [T.sub.E] is dissipated to the environment at temperature [T.sub.AC].

Given a certain high temperature heat source, either of the following systems could prove to be the most efficient:

a. To drive a medium temperature driving cycle (e.g., steam turbine) using heat from the high temperature driving cycle, then to use the heat from the medium tem perature cycle as a supply for a single effect absorption cycle.

b. Using the heat from the high-temperature driving cycle to drive a double-effect absorption cycle.

[FIGURE 5 OMITTED]

If all the heat is utilized within the cycle:

COP = [T.sub.e]/[T.sub.c] - [T.sub.e] [T.sub.l] - [T.sub.c]/[T.sub.l] (11)

The first multiplier of the above equation is the equivalent cooling-cycle COP and the second is the equivalent driving-cycle efficiency:

[COP.sub.cc] = [T.sub.e]/[T.sub.c] - [T.sub.e] (12)

[[eta].sub.dc] = [T.sub.l] - [T.sub.c]/[T.sub.l] (13)

The overall COP is the product of both of these:

COP = [COP.sub.cc][[eta].sub.dc] (14)

If this concept is extended to higher numbers of driving and/or cooling effects, any of these combinations of cycles are equivalent and can be represented by one driving cycle and by one cooling cycle (Tozer and James 1995b). The overall COP is equal to the product of the efficiency of the equivalent driving cycle and the COP of the equivalent cooling cycles as indicated in Figure 6.

The maximum COP will be achieved for infinite effects and is equal to the reverse Carnot COP. This implies that as long as no heat is wasted (in- or outside the cycle) and ideal cycles are achieved for the same source and sink temperatures, there is no thermodynamic difference between different configurations. This can easily be checked by derivation of the exergy equation of the high-temperature heat source Q1 and low-temperature heat source (T1). The exergy of the heat dissipated to the outdoor ambient is considered to be at reference state, therefore, zero exergy.

[FIGURE 6 OMITTED]

TRIGENERATION

Cogeneration is the use of fuel- (normally natural gas) fired turbines or engines that generate electricity (via an alternator), whereby the waste heat (such as exhaust gases and jacket water) is used for heating purposes.

When absorption chillers are combined with cogeneration, which can normally produce high enough temperature heat sources, these systems are referred to as trigeneration because they produce electricity, heat, and cooling energy. From a theoretical perspective of ideal cycles, while absorption and mechanical compression cycles are equivalent, as absorption chillers have lower COPs, their source energy quality (exergy) is less (Carnot factor).

Gas engines have two sources of waste heat: the exhaust gases and the jacket cooling water. Both of these are used to drive a single-effect absorption chiller (Hufford 1991). The exhaust gas heat of the turbines is used to fire absorption chillers either directly or by means of an exhaust gas to steam or hot-water heat exchangers. However, absorption chillers are mostly used with cogeneration/trigeneration systems in the following way:

* Single-effect chillers: hot-water driven (80[degrees]C-140[degrees]C) (176[degrees]F-284[degrees]F)

* Double-effect chillers: steam driven (300-900 kPa) (43.5-130 psi)

Although single-effect absorption chillers can be supplied with low-pressure steam, the most common method is to use the engine jacket cooling water.

The main components are as follows (Figure 7):

* Gas engine/electric generator

* Heat exchanger (exhaust gases to jacket hot water)

* Single-effect hot-water-driven absorption chiller

[FIGURE 7 OMITTED]

Hot-water temperatures are available between 70[degrees]C to 140[degrees]C (158[degrees]F to 284[degrees]F), but, although the heat energy available will depend on the flow and temperature rise of hot water, the supply and return temperatures themselves have a significant influence on the size of the absorption chiller. Low hot-water temperatures require the use of very large chillers when compared to selections with higher hot-water temperatures. Nevertheless, gas engines with single-effect absorption chillers systems normally offer the best payback results due to its low initial cost.

Many cogeneration systems use steam at approximately 800-1500 kPa (116-218 psi), e.g., hospitals and industrial applications; these are suitable for double-effect absorption units that are designed for steam at 800 kPa (116 psi) (Miguez Gomez 1993).

The main components are as follows (Figure 8):

* Gas turbine/electric generator

* Double-effect steam-driven absorption chiller

[FIGURE 8 OMITTED]

The unit has the best COP of double-effect systems and lowest temperature of exhaust gases at 140[degrees]C (284[degrees]F). This is because the exhaust gases heat the steam condensate from 90[degrees]C (194[degrees]F), which, in turn, is used in the absorption chiller.

Current options make the choice of single-effect with gas engines or double-effect absorption chillers with higher temperature heat sources (e.g., from gas turbines) optional with respect to their ideal theoretical COPs; like cold-generation systems, these are identical. The choice will depend on other constraints such as initial costs and requirements to minimize irreversibilities. A key point in the design and operation of any cogeneration/CHP system is that all the waste heat is utilized to maximize the efficiency at all times (DOE UK 2000).

In practical terms, the use of absorption cycles with cogeneration is as efficient as standard mechanical compression cycles using grid electricity. Nevertheless, trigeneration systems can be quite attractive when natural gas is cheap or the national grid electric production is insufficient/inefficient and or expensive. Trigeneration systems require a higher capital cost investment and specialist knowledge of designers, installers, maintainers, and operators. Trigeneration systems have to compete with both the national grid production efficiency and a mature mechanical refrigeration industry (Table 4).

Operational Considerations of Absorption Chillers

With respect to operational considerations, the most important issues are the following (IEA 1990):

* Vacuum requirements: The vacuum levels associated with water/LiBr absorption are high and the sensitivity of the technology to leaks is therefore also very high. Absorption machines are sensitive to leaks both because air in the machine reduces performance and because of corrosion considerations.

* Crystallization: Conventional controls sense chilled-water temperature rising and call for increased heat input to the generator. This, in turn, tends to concentrate the salt solution, and in extremes can cause crystallization. An option to reduce this risk is to provide uninterruptible power supply to the internal pumps to avoid problems of high concentration in the event of power failure.

* Corrosion: In the presence of dissolved oxygen, water/ LiBr is highly aggressive to many metals, including carbon, steel, and copper. Over the operating life of a machine, significant corrosion can occur, and care must be taken to minimize its effects. The primary measurements available are pH control and use of corrosion inhibitors.

* Maintenance: Normal maintenance for water/LiBr technology includes periodic purging of non-absorbable gases, addition of octyl alcohol, addition of corrosion inhibitor, addition of pH buffer, and monitoring of the water/LiBr solution.

* Health and safety: Some of the chemicals used may be toxic, carcinogenic, or in some other way harmful to health.

APPLICATIONS TO DATA CENTERS

On face value, data centers would appear to be ideal candidates for trigeneration systems due to their high power and cooling demands. The IT equipment is the largest component in the data center energy consumption. Until around 10 years ago, energy for cooling systems typically exceeded one-third of the total data center energy. These data centers have a ratio of data center energy to IT equipment energy, i.e. power usage effectiveness (PUE) of 2 or higher (Figure 9).

If an absorption refrigeration system were to be employed, this should be used for the annual base load with mechanical compression systems for the variable cooling load. Similar to other applications, the market penetration of trigeneration for data centers has been low (Figure 10).

[FIGURE 9 OMITTED]

[FIGURE 10 OMITTED]

The data center industry has been through a paradigm shift with regard to energy efficiency best practice (European Commission 2013). Central to this was extending the environmental ranges of IT equipment (ASHRAE 2012). Air management has had a big role to play as it has enabled higher temperature setpoints with no detriment to IT equipment conditions (Tozer et al 2009). Current good practice data center designs are much more efficient, with PUEs of around 1.2-1.3 (Figure 11). This is achieved with the extensive use of free cooling/economizer cycles, among other things (data hall air management, variable-speed fans, and higher cooling setpoints) (Flucker and Tozer 2013). Furthermore, there are many possibilities of worldwide free cooling with zero refrigeration, and there is virtually nowhere in the world where, through implementing best practice, a large amount of free cooling can be achieved (Tozer and Flucker 2012) (Figure 12).

[FIGURE 11 OMITTED]

[FIGURE 12 OMITTED]

This approach is relatively inexpensive and simpler compared to use of trigeneration systems. Even in very warm countries, using the right free-cooling technology (direct or indirect, air- or water-based/adiabatic cooling), there are a considerable number of hours during the year whereby total free cooling with no mechanical refrigeration (zero refrigeration) can be achieved.

This means that there is no longer a cooling base load and thus no requirement for absorption cooling used in conjunction with trigeneration systems.

CONCLUSION

If good practice is adhered to by not wasting heat in trigeneration systems, there are both--in theory and practice--no significant differences of efficiencies between the different types and combination of cycles using gas engines/ gas turbines with single and double-effect absorption chillers. Similar efficiencies and environmental impacts are also found when absorption refrigeration as part of a trigeneration system is compared with standard mechanical compression systems using grid-sourced electricity. Trigeneration has many advantages, such as cost incentives, however, disadvantages include higher investment costs and design and operational complexity, which are undesirable in data center environments. In recent years, the efficiency of data centers has seen a dramatic improvement due to the use of extended free cooling/economizer cycles, enabled through the use of data center air management/containment systems, higher air supply temperatures, and extensive use of variable frequency drives in the HVAC plant. Whilst there may have been a case for trigeneration systems with traditional designs using absorption chillers for the base cooling load, extensive use of free cooling/ economizer cycles means there is no longer a base refrigeration load. These recent changes to best practice should be taken into account and may make trigeneration unviable for data center applications.

ACKNOWLEDGMENTS

The author appreciates the encouragement and support from his colleagues Sophia Flucker, Beth Whitehead, and Alejandra Romano.
NOMENCLATURE

E        =    number of effects
l        =    heat of solution, kJ/kg (Btu/lb)
Q        =    heat, kW (Btu/h)
s        =    entropy, kJ/kg (Btu/lb)
r        =    heat of evaporation, kJ/kg (Btu/lb)
T        =    absolute temperature, K (R)
[eta]    =    efficiency

Subscripts

A     =     absorber
c     =     condenser
cc    =     cooling cycle
dc    =     driving cycle
e     =     evaporator
g     =     generator

REFERENCES

Alefeld, G. and R. Radermacher. 1994. Heat conversion systems. London, UK: CRC Press.

ASHRAE. 2012. Thermal guidelines for data processing environments--Expanded data center classes and usage guidance. Atlanta: ASHRAE.

ASHRAE. 2013. Chapter 2.15, ASHRAE Handbook--Fundamentals. Atlanta: ASHRAE.

DOE UK. 2000. CHPQA: A quality assurance for combined heat and power. London, UK: Department of Energy, Trade and Regions, UK.

Eber, N. 1968. Die darstellung und analyse von absorption-skreisprozessen mit hilfe des temperatur-entropie-diagrammes. Doctoral thesis, Federal Technical High School, Zurich, Switzerland.

European Commission. 2013. Best practices for the EU code of conduct on data centers. Version 4.0.5. Brussels, Belgium: European Commission.

Flucker, S. and R. Tozer. 2013. Data center energy efficiency analysis to minimize total cost of ownership. Building Services Engineering Research Technology. 34(1):103-17.

Gosney, W.B. 1982. Principles of refrigeration. Cambridge, UK: Cambridge University Press.

Herold, K.E., R. Radermacher, and S. Klein. 1996. Absorption chillers and heat pumps. London, UK: CRC Press. Hufford, P.E. 1991. Absorption chillers maximise cogeneration value. ASHRAE Transactions 97(1):428-33.

IEA. 1990. Ab-sorption machines for heating and cooling in future energy systems. Final Report, IEA Heat Pump Program, Annex 24. Paris, France: International Energy Agency.

Kotas, T. 1995. The exergy method of thermal plant analysis. Malabar, FL: Krieger Publishing Co.

Miguez Gomez, C. 1993. Utilizacion de turbinas de vapor para accionamiento de equipos de climatizacion en ciclo combinado para cogeneracion. Presented at CIAR II, Ibero-American Conference of Air Conditioning and Refrigeration, Madrid, Spain. March.

Nesselmann, K. 1932. Zur Theorie der Warmetransformation, Wissenschaftliche. Siemens Konzern 12(2):89-109.

Niebergall, W. 1959. "Sorptionskaltemaschinen.", in Handbuch de Kaltetechnik, Vol. 8. Ed. R. Plank. Berlin, Germany: Springer Verlag.

Stierlin, H., 1964. Beitrag zur theorie der absorptionskaltemaschine. Kaltetechnik 16:213-18.

Tozer, R., and R. James. 1994. Thermodynamics of absorption refrigeration (ideal cycles). Proceedings of the International Absorption Heat Pump Conference, New Orleans, LA, 393-400. New York: ASME.

Tozer, R., and R. James. 1995a. Absorption refrigeration applied to CHP systems. Building Services Research and Technology 16(4). New York: Sage Publications.

Tozer, R., and R. James. 1995b. Theory and application of absorption refrigeration systems, The Institute of Refrigeration Proceedings, 92. Carshalton, UK: Institute of Refrigeration.

Tozer R. and R.W. James. 1997. Fundamental thermodynamics of absorption refrigeration, International Journal of Refrigeration 20(2):120-35. Amsterdam, Neatherlands: Elsevier.

Tozer R., M. Salim, and C. Kurkjian. 2009. Air management metrics in data centers. ASHRAE Transactions 115(1). Atlanta: ASHRAE.

Tozer, R. and S. Flucker. 2012. Zero refrigeration for data centers in the U.S. Presented at ASHRAE Annual Meeting, San Antonio, TX.

Wilkinson, W.H. 1987. What are the performance limits for double-effect absorption cycles, ASHRAE Transactions 93(2):2429.

Robert Tozer, PhD, CEng

Member ASHRAE

Robert Tozer is a visiting fellow at London South Bank University, London, UK, and managing director at Operational Intelligence Ltd., Kingston upon Thames, UK.
Table 1. COPs of Real and Ideal Systems

Number of Effects   COP Real   COP Ideal

1                     0.68       0.87
2                     1.2        1.64

Table 2. Absorption Chiller Range

Effects          Heat Source          Refrigerant   Condensed   COP

1         Direct fired natural gas        NH3          Air      0.5
Hot water 80[degrees]C/
140[degrees]C

1              (176[degrees]F/        [H.sub.2]O      Water     0.7
284[degrees]F), steam
200/100 kPa (29.0/14.5
psi)

2         Direct fired natural gas    [H.sub.2]O      Water      1

2             Steam 300-900 kPa       [H.sub.2]O      Water     1.2
(43.5-130 psi)

2               Exhaust gases         [H.sub.2]O      Water     1.1
280[degrees]C-
800[degrees]C
(536[degrees]F-
1472[degrees]F)

Table 3. Ratio of Ideal HDRs (Heat Dissipation Ratios)

Number of Effects   Ratio of HDR Absorption/
Absorption Cycle     Mechanical Compression

1                             2/1
2                             3/2
3                             4/3

Table 4. Features of CHP Absorption Chillers

Chiller                                       Features

Single effect hot water driven   * Low chiller cost
(optimum selections)

* Low COP (0.6 to 0.75)

* High electric consumption (hot
water temperature <100[degrees]C
[212[degrees]F])

Double effect steam driven       * Highest COP (1.1 to 1.2)

Double effect hot gas driven     * Highest initial chiller cost

* High COP (1.05 to 1.15)

* No steam installation required
COPYRIGHT 2015 American Society of Heating, Refrigerating, and Air-Conditioning Engineers, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.