How Can This Be? {he '2"°= Voley The performance curve showin in Chart 1 models an advertisement Introduction In two previous Engineers Newsletters, headline proclaiming a particular chiler’s ‘we addressed the issue of chiler aly to consume 0:20 kWVfion at 60 crlormance under the ite "The Search — peteent of ts desig capacity and 0.26 Fer cer Efisency" The intpine of Kon at ul espacty. Clary, Wis is for Gite Effency” The inchpins of rational performance, But rea? physics and thermmodmemics ond how Sulriingy, the answer is yesl those lave offer epocte and defined Performance at 020 kWiton seems so limitations on maximum, real-world ‘much better than anything published chiller efficiency. and ARI-rated today. How can this be? Recent HVAC industry advertisements __Thete's some trickery involved, but ‘appear to contradict some of these there's how it's done. tarber-cited imitations. Were the OWS pack To The Basics of physic and thermodynamics recerty Fat, i's important to bulla foundation repeated? Are they being cf fac Overal chile efciency Imisinterpreted? Oris there someting Leck ONO ene ot only funny going on? so, exactly what is four, element the source of confusion? — ‘The objective ofthis newsletter is Wate-to-efrigerant heat transfer answer each of these questions and) provide substantive, fundamental ‘echnical background for thase systern designers who wish to go beyond the Power conversion and transmission claims and facts for themselves. . Centifugal compressor efficiency Retrigerantcyele thermodynamics Chart Effect of decreased condenser water temperature on chiller efficiency (ork’s claim) sewrton 100% Load Emerg Condenser Water Temperature (ea F) g z‘Table 1: Theoretical isentropic efficiencies for commonly used refrigerants at 40 F saturated suction and 100 F saturated condensing temperatures: Tati These tan Towa ior i eae ats 2 687 usr i 0679 st 723 638 ars Zz 066 oot ‘Table 2: Effect of “economizers” on refrigerant cycle efficiency (see Table 1) ‘Theoretical ton Thebiebical Mon Retigerant |" Simple [econ | 2aeon | Simpio_[ t-econ | Zeon vata [0679 | ose | — [0507 | aa? [0 2306s | ~onno | oso0 | —oa7a | oaea | a0 "No such compressor dosign is commacialy aveidble ‘While the differences from one Refrigeration-Cycle manufacturer's design to another aren't Thermodynamics {great in any of these four elements, they “Simple cycle” efficiencies based on are signicant enough to merit, alternative reftigaramts are a tie lower inspection, The Engineers Newsletter than their CFC predocessors in HVAC. focused on this twice in the past once centrifugal chiller applications, As Table 1 in 1981 and again in 1994. Good news! indicates, R-123 is about two percent The basis are unaltered and the laws less efficent than the F-11 i replaces, ‘of physics and thermodynamics remain while R-134a is about tee percent less ‘unchanged, so the fundamentals we all efficent than F-12. learned in our physical science classes stil hold true. Various alterations can be made to improve “simple-cycle” refrigeration That being the case, let's examine the performance. One of the most cornmon latest chiler performance claims in light compressor adaptations is the adton ‘of each of these elements, (of an economizer, also known as an interstage flash chamber. Since this ‘Heat Transfer device is physicaly placed between ‘Thanks to advanced technology, and the Stages of compression, ts use limited ‘addtional research and development ty mulistage compressors Of eaurse activity prompted by the advent of new multiple economizers can be used if alternative refrigerants, heat transfer more than two stages of compression Cantinues to improve in both ate invoived, Table 2 compares the cycle evaporators and condensers, The fruits Sficiencis for common compressor! Of this labor are evident, as we now economizer arrangements with the bserve refigeranttoawater “approach” “Simp eyele” values presented in temperatures in the 210 4F range. For japles ‘evaporators, this means i's now ‘economically feasibie to obtain retiigerant gas temperatures roughly 3 F lower than the chilled water temperature produced there. Likewise, saturatod condensed liquid in the condenser can aften be within 3 F of the leaving condenser water temperature, Twenty years ago, normal ‘approach temperatures were 7 to 10F.Liquid subeooting benefits some refrigerant cycles. Unfortunately, ts potential Benefits are insignificant in ‘water-cooled chiller applications because of the small temperature alfference between the heat sink (Condenser water) and saturated condensed Bquid. Heat transfer improvements further diminish this temperature differential, Not surprisingly, subcealing cycles are seldom employed with this kind of equipment {All commonly used refrigerants respond to changes in “thermodynamic head. {Ass the case for an ideal retigerant, less theoretical power is required when the pressure differential between the: ‘evaporator and the condenser is reduced, Charts 2 and 3 show the effect of different “heads” on the theoretical power consumption of R-123 and -134a cycies, respectively, Clearly head reduction affects each of these rwitigerants in a very sinar manner. Power Conversion And ‘Transmission Efficiency There is almost nothing new to report con these technotoges. (We say “simost” because well examine the effects of frequency variation later in ths article) Induction motor efficiencies remain in the 93 10 96 percent range for three phase, 60-Hz motors in the size ranges Used by centrifugal compressors. ‘This performance holds true for both hermetic and open motors; apart from the cooling system, ther designs are virtually identical. (The power sacritoed ‘by open motors to dive cooling;air fans ig minimal and unworthy of analysis here, although the effect on equiprment room cooling and ventilation can be significant) Both mator designs, for ‘example, use two journal bearings 19 support the rotor, Motor performance ratings reflect the frictional losses associated with these bearings (Editor's Note: The analysis presented in this issue is based on a motor wth an overal efficiency of 96 percent) Chart 2: Effect of reduced “head” on the refrigerant-cycle efficiency of R123 os. oa oa oz 100 0 20 70 60 ‘Satured Condensed Liquid Temperature (deg F) (Chart 3: Effect of reduced “head” on the refrigerant-cycle efficiency of R-134a os 0 os ‘Theoretical kWrTon 02 os 100 2 80 7 6 Saturated Condensed Liquid Temperature (deg F)Compressors operating at speods higher than the motors synchronous speed [600 rpm for 60-He, two-pole motors) Fequire some kind of "step-up" transmission. Various arrangements are Used, each with additonal gears and bearings. By themselves, the gears absorb about one percent oftheir power transmission output in heat-producing friction. Efficiency losses attributable to ‘ficton typical amount to about 1 hp (2.546 Btu/he) per bearing Consider the performance impact on a ‘common transmission design that employs two gears and five bearings frictional bearing losses are roughly 5 hp (12.726 Buu) or nearly 4 RW which, in turn, amounts to about one percent of, the power output of a B4DKW motor Thus, total gear and bearing losses in the transmission of an efficient 1,000- ‘on chiler are about two percent, leaving a transmission efficiency of 98 percent. Of course, chillers without ‘rensmissions suffer no such loss, Compressor Efficiency Ineremental improvements in compressor effciency continue to be achioved with the ongoing competition between single- and mutiple-stage ‘compressor technologies. Both designs {are highly refined and well-suited to ‘water chiler applications, As evidenced by the typical centritugal ‘compressor "map" in Chart 4, all constant speed centrifugal compressors lose efficiency at partoad condition. Consequently, manufacturers generally Select compressors at their “sweet spot” forthe rated design conditions Normal chiled water system unloading rot only results in lower pressure differential demand, but proportionally less volume as wll: .e, fewer tons of capacity Chart 4: Open inlet vane compressor performance at full (100%) speed 09 os ate o7 os os oa imu) ta 2 g a3 02 ox 09 o 1 2 3 4 ‘anvTon Flow Chart §: Chiller performance, % power vs % load 100 0 80 n 0 0 0 2 10 ° 20 “0 60 cy 100 Full Load, Tons,Pressure differential, on the other hand, basicaly folows saturated refrigerant pressure/temperature values between the condenser and evaporator. These ‘values tack closely with the leaving water temperature for each of these heat exchangers, ‘Therefore, as a system unioads, we can ‘expect 3 lower temperature rise in the condenser water, as well as a potently Tower condenser water supply temperature. These two affects combine to reduce the pressure aifferential significant, leading to a much more efficient refrigerant cycle but atthe expense of reduced ‘compressor efficiency. The net resut is better overal chiller partload efficiency 10 a point, Chart § shows this relationship, Points plotted on ths chert criginate from the data contained in Charts 4 and 2, and incorporate the ARI Lnioading factor for decreasing ‘condenser water terperature Cleary, this performance doesn't approach anything close to 0.20 kWiton at any load condition, regardless of entering condenser water temperature, ‘What did we overtool? ‘The Missing Ingredient Chart 2 contains our first chue. Notice the relative theoretical “simple cycle” efficiencies (0.47 and 0.25 KWon) at thermodynamic heads of 98 F and 73 F, respectively. These two values represent normal saturated condensed liquid temperatures for a fully loaded chile with a design entering condenser water temperature of 85 F and for the same chiller when the condenser is supplied with 60 F water. These ‘extremes are plotted on Chart 1 and ‘correspond to performance of (054 kWWiton and 0:26 kWViton for a fuly loaded chiler. The ratio ofthese two values is 0.261054, or 0.48. In other wards, a chiller operating with this eyele Consumes 48 percent of ts design ower when the entering-condenser water is 60 F. Chart 6: Variable-speed compressor performance at 100% and 70% speeds os or ear) sta 109) oe . Point A 2 os g sma ione) Zo oa Point o2 mu 0%) ox eT anon row Chart 2 shows the relative theoretical efficiency values of 0.45 kWiton and (0.24 KWrton at 98 F and 73 F, respectively. The ratio of these two values is 053 . .. amazingly similar to the 0.48 rato calculated above, ‘Theoretically, then, the power consumed at a 73 F saturated condensing temperature is 63 percent of that consumed at 98 F. This means that all other chiller components must. ‘operate with very similar efficiencies at both conditions. How can this be, if compxessor efciency fails at decreased thermodynamic head (Chart 4? We ‘need to find some way for the ‘compressor to produce less ‘thermodynamic head without losing ‘volumetric capacity or efficiency, Said another way, we must develop a scheme to “kil” compressor head ‘without sacrificing avaible cooling tons ‘oF performance. One possible way to perform this feat isto operate the: Compressor at a lower speed, Since head is proportional to (alocity/*, we ‘can theoretically produce 49 percent head at 70 percent speed. But siower velocities also reduce volume capability, How can we obtain full volume at reduced speed? ‘The answer: Ride the compressor curve to a highor relative volume at reduced speed. Chart 6 shows two ferent speed curves for the ‘compressor represented in Chart 4 While any point on either of these curves a possible operating point, we ‘want to find two points that altfer only in thermodynamic head... not volume. Points A and B quali.Chart 7: Comparison of York and Trane centrifugal chiller performance soo soa 8 m7 Rn @ 7 © 8 Hw Entering Condenser Water Temper (208) [Notice that it's important to select the So far, everything we've seen is based ‘original point (A) away from any sort of on the practcal appicaton of compressor volume limit since we must fundamentals Is there any test data that "nde" the volume curve far enough to reflects the actual performance of a reach design gas volume. Faiing to do centrifugal chiler at these specific off 0 wil result ina capacity shartfal Also design conditions? The answer is yes. Significant is ths simple fact: Operating a Chart 7 lustrates the performance of 4 ‘centrifugal compressor at reduced Trane CVHF-080 operating with an speed “tls” thermodynamic head Adaptive Frequency” cive finverter) and vathout sacrificing efficiency, ‘compares it with the advertised clas ‘modeled in Chart 1, Its evident that the ‘Speed Control addition of an inverter enables Reaching high chiler performance cicapaial perlornancs ot astbed values requres a twofold strategy: cones Variable compressor speed and controls that respond to ‘So why doesn’ everyone use frequency ‘thermodynamic head demand. inverters? For the same reason that ‘Specialized frequency inverters can ‘everyone doesn't own 2 $40,000 implement this strategy. Adsing such an automobile. They can't afford it inverter to the drive tain means itis ‘Inverters are very expensive. As Table 3 possible to reach 0.26 kWiton at ful indeates, large and continuous operating ‘capacty and 020 kWiton at reciced cast savings are requited to amortize (60 percent) capacity conditions. (Bear inthe frst cost of a $40,000 frequency ‘mind that inverters do lose smal inverter purchased to drive the amounts of power in the form of heat. compressor motor. In realy, very few These losses are cominuous unless the systems can generate operating cost inverter is electrical bypassed when savings ofthis magnitude the motor must operate at full speed)6 me ‘able 3: Amortization table for a $40,000 inverter 1 eta ny — Br Pre ae : vw | swings [5 a oH oi 7 Te mma! | Boon | —tarano | azo) | en aea | S00 Ha | kWh? 571429 594,286, 600,000 605714 611,429 617,143 622,857 3] an] sation sotane | saisia | aeiaty | sania | saat | sar fame | amora_| ama | sora | artere | stone | sono | a2asao Sena] Saas] sreate | stanes | sia | —sisz@ | sist | — sist fine | tonare | zonota | zooms | ovary | erm | sare | anes 7] amar} sito] Stir] srize0 |] sites] strane) stzart | —sr2oe7 wane | weer | toraza | torseo | teams | ream | rose | vaae2 Sana] Soot sans] — $5200] — ssa] sor66 | —ston1e] —sraaoe wane | isazee | vamos | ronaee | sess | rena | vaane | wso10 *@7 conts per kWh Let's take a closer look at Table 3 and ‘what it els us. The "0%" column represents simple payback valves, More likely, money has an assacited cost to any owner and is expressed as the “annul percentage rate.” If electricity costs 7 cents per KW, the necessary dolar payback values are translated into KWh savings required 10 pay off the invert, Naturaly, the longer ‘an aver can wait far payback, the smaller the threshold for requited savings becomes, For example, if an ‘owner requires payback within three years and assigns six peroent to the cost of money, an annul energy savings of 213,777 kWh is required Assume a diference of 0.27 KWWiton between 0.48 KWWiton (normal 60 F, 60 percent, partoad operation without an invertet| and 0.21 KW/ton (60 F, 10 percent, part-load operation with an inverter, For an €00-t0n chiler, 0.27 kWiton amounts to a decrease in power ‘consumption of 216 kW at this particular off-design condition. Under these circumstances, t would take 990 hours of operation annually to generate the necessary 213,777 kWh The lkelhood of a combined system load and environmental cimato providing this condition over such an ‘extended number of hours seems remota, fo quolify as a candidate for a variable speed operating strategy, the cles load profie must exhibxt many hours of ‘operation at condenser water temperatures betwen 55 and 70 F Lite savings are generated at temperatures above 70 F, Below 55 F, free cooling isa better aternative, OF course, this narrows the “window of ‘opportunity’ for ecanamic payback. In ‘any setious economic analysis, the relative cost of obtaining 55 to 70 F ‘condenser water must be weighed ‘against the cost of B5 F wator 0.0 KW/Ton Face it Free cooling is difficult to beat if the associated capital costs are reasonable. Several forms of fee cooling make sense; one option is the "sidecar” strategy explained in an earlier Engineers Newsletter (vol. 20, No.3. Ilustrated in Figure 1, the "sidecar” heat ‘exchanger permits a measure of free ‘conling whenever the cooing tower water temperature is lower then the system return water temperature. This scheme greatly expands the number of hours during which some 0.0 KWWiton cooling is obtained, Amortization is often quite 2p. Just Kidding? Definitely not. Seemingly unbotiovable performance headines usualy contain at least a shred of truth In this case, ‘we find that 0.20 KWiton chiller performance is very possible, as 1s (00 kWhton. There's no end to one. upmanship in advertising Under the harsh ight of informed sorutny, we find thatthe simple laws of ‘economics ara the only real obstacte to achieving spectacular performance ‘within a natrow range af operating Conditions, Economics forces us to broaden the scope of study Beyond the headlines. Figure 1: "Sidecar" free coolingHow Can This Be? Volume 25, No. 1, of the Engineers Chart 4: Open inlot vane compressor performance at full (100%) speed Newsletter examined the claims of ‘several recent HVAC industry advertisements that seem 10 ‘contradict the laws of physics and thermodynamics. To clarify the effect, of these laws on real-world chiller efficiency, the newsletter provided ‘echnical background to aid system designers wishing to evaluate such assertions themselves. WL-ON ‘Sz loNa66L ‘Two graphs—Chart 4 and Chart 6—were presented as part of the discussion on centrifugal ‘compressor efficiency as one of the four basic elements of overall chiller efficiency. Regrettably, the legends were inadvertently omitted from both illustrations. Both charts are reproduced here, along with more descriptive axis and curve labels. Explanatory notes follow. Overall Work Cootficient(), Adiabatic Efficiency ( Compressor Performance Charts 4 and 6 at right describe three facets of centrifugal compressor performance: Adiabatic efficiency, or eta (n.), refers to mechanical efficiency, and is the ratio of isentropic work, Wy, to actual work, h~ hy; that is, We (hy hy) (0), is determined by multiplying ‘rovitational constant g by Polytropic work W, and dividing the fesult by the sum of the perstage impoller blade tip speed v, squared— ii. (@% W,)» Eu. The relationship between polytropic head H, and polytropie work's expressed as 4, = Wyo. Flow coefficient is dimensionless, and is calculated by dividing the per-ton volumetrie flow rate, Q, by the product of impeller rotational speed N and impeller diameter 0 cubed—Le,,Q (Nx D3. 1 1820 Fiow Coetcint(0/ND") =] ee =] i) © = Ca) 4 H oe ) 4 my () =