An aerodynamic investigation of a centrifugal compressor for HCFC123

An aerodynamic investigation of a centrifugal compressor for HCFC123*

Koji Nakagawa and Sadashi Tanaka Mechanical Engineering Research Laboratory, Hitachi Ltd, Tsuchiura, 300 Japan

Junichi Kaneko Tsuchiura Works, Hitachi Ltd, Tsuchiura, 300 Japan Rece ived 8 February 1991; revised 10 N o v e m b e r 1991

A centrifugal compressor was investigated experimentally to improve HCFC 123 compression performance. A CFC11 compressor working with HCFC123 shows lower adiabatic head rise, efficiency and volume flow. Another CFC II compressor working with HCFC 123 and a modified impeller shows a result comparable to that when CFC11 is compressed. A tandem impeller is designed to compensate for the volume flow increase for HCFC123. This impeller sufficiently absorbs the volume flow increase and also improves efficiency. (Keywords: centrifugal compressor; HCFC123; impeller)

Recherche arrodynamique sur un compresseur centrifuge utilisant du R123

On a btudib expbrimentalement un compresseur centrifuge en vue d'ambliorer la performance de compression du R123. On a chargb avec du R123 un compresseur utilisant habituellement du RI1; l'blbvation adiabatique de pression a btb moins importante ainsi que le rendement et le dbbit-volume. On a renouvelb l'expbrience avec un autre compresseur au R l l que l'on a charg~ avec du R123 et dont la roue a ~t~ modifi~e; on a obtenu des r~sultats comparables ~ ceux observbs avec le R11. On a con¢u une roue en tandem pour obtenir l'augmentation de dbbit-.volume requis avec le R123. La performance de cette roue suffit ~ satisfaire cette augmentation, et on obtient par ailleurs une amelioration du rendement. (Mots clrs: compresseur centrifuge; R123; roue)

Nomenclature

Had M as

w

Adiabatic head rise (m) Mach number Suction volume flow-rate (evaluated by stagnation density) (m 3 min-1) Relative velocity (m s-~)

Greek letters

r/c Total-to-total stage efficiency r/t Stage efficiency including mechanical losses

Subscripts

d Design point 1 s Impeller inlet tip 2 Impeller exit

Although CFC11 is currently widely used for centrifugal chillers, its production will be stopped because of its high ozone-depleting potential. Because HCFC 123 is the best possible substitute for CFC11, an HCFC123 centrifugal chiller development programme was planned and imple- mented. During the programme, testing was made with the goal of obtaining the characteristics of a CFC11 centrifugal chiller filled with HCFC123 by using a 300 USRT unit. The evaporating temperature and condens- ing temperature were selected to be near the design-point values for C F C 11 condensation. Figure 1 shows the unit under test. When the unit was filled with HCFC123, there was a decrease in capacity of about 10% and the

*This paper was first presented at the UNSC/IIR-Purdue Refrigeration Conference held at Purdue University, 17-20 July 1990.

Figure 1 Figure 1

300 USRT machine under test La machine 300 USRT utilis$e pour les essais

0140-7007/92/040199~7 © 1992 Butterworth-Heinemann Ltd and IIR Rev. Int. Froid 1 992 Vol 1 5 No 4 199

An aerodynamic investigation of a centrifugal compressor for HCFC123: K. Nakagawa et al.

2 5 0 0

-r. 2 0 0 0

< 1500

CFC II

- - - o - - - HCFC 123

Head rise

Ef f ic iency

0---%

I I I (3 5 0 IOO 150

Volume f low- ra te , (2 s ( m 3 min -I )

f

o

I

0.9

Figure 2 Comparison of head rise and efficiency for 300 USRT centrifugal chiller (same impeller): r/td, efficiency at design flow-rate Figure 2 Comparaison de l'blbvation de pression et de l'efficacitb pour un compresseur centrifuge 300 U S R T (mOme roue)." rbe, efficacitd pour ddbit-volume nominal

energy consumption increased between 5 and 10% per unit refrigerating capacity at the design temperature. The direct cause of this performance degradation is presumed to be the difference in the thermodynamic properties of CFCI1 and HCFC123. The theoretical compression work and volume flow-rate of HCFC123 per unit refrigerating capacity are higher than those of CFC11. The compressor performance degradation caused by the heavier molecular weight of HCFCI23 led to an overall lower chiller performance. Figure 2 shows the performance of the 300 USRT unit compressor. The impeller and the rotational speed were the same for both refriger- ants. The adiabatic head rise and efficiency dropped for HCFC123 by ~ 6%. This means that a large part of the total energy consumption increase was caused by the compressor efficiency drop.

Akitani and Nakaiwa 1 experimented with CFCI1 and HCFC123 in the same compressor. They concluded that an HCFC123 compressor with a CFC11 compressor performance level is possible with an appropriate impeller design. With this background, an aerodynamic investigation was made to improve the HCFCI23 compressor performance. The aim of this investigation is not only to improve energy consumption, but also to develop a compact machine. HCFC123 compression performances with a CFC11 impeller and a modified impeller are presented. The performance with a tandem impeller for higher flow-rate is also presented.

Experimental procedure Experimental apparatus

The experiments were made by using a centrifugal compressor test system that consists of a single-stage cam-

Figure 3 Test stand Figure 3 Bane d'essais

Ld

Refrigeration cycle r - cooled

Pressure

Figure 4 Test-stand cycle Figure 4 Cycle du bane d'essais

pressor, a torque meter, a d.c. motor, speed-up gears, a system of heat exchangers and flow piping. The appear- ance and cycle of the system are illustrated in Figures 3 and 4, respectively. The refrigeration is not condensed, but chilled to the compressor inlet temperature by a brine-cooled heat exchanger. This gas cycle test system has the following merits. The heat-control equipment is simplified and the energy consumption is lowered. In addition, only a small amount of refrigerant is necessary. A cross-sectional view of the centrifugal compressor is shown in Figure 5. Two types of impeller, described in the following section, were used in this compressor. The compressor casing was covered with insulating materials in order to eliminate the effect of heat leakage on the total temperature measurement.

Measuring method

Both the total pressure and total temperature at the suction and discharge sections were measured with Kiel probes and total temperature probes consisting of copper-constantan thermocouples. The performance of the compressor stage was evaluated on the basis of the total-to-total rating. The flow-rate was controlled by a valve between the heat exchangers and was measured with an orifice-type flowmeter in the suction pipe. The

200 Int. J. Refrig. 1992 Vo115 No 4


\ \ \ N

\ \ \ \ Collector x \ \ \

\ Diffuser

o

¢M

Figure 6 Fully shrouded model impeller Figure 6 Mod~le de roue compl~tement fermbe

I.I

Figure 5 Cross-sectional view of a model compressor rig Figure 5 Coupe d'un mod~le de compresseur

total error of compressor performance measurement was within 4-0.5%. Although the HCFC123 used was delivered without any special purification, the thermodynamic properties of pure HCFC123 were used in cal- culating compressor performances. Usually HCFC123 product contains about 2-3% HCFC123a. This, however, did not increase the uncertainty of the calculated result, because the gas constants of HCFC123 and HCFC123a are the same, and their specific heats are almost the same. The thermodynamic properties of HCFC123 were calculated by using equations of state 2 and values given in Reference 3.

Results and discussions

HCFC123 compression by a CFCl l compressor

The first set of experiments was completed to check the general operation of, and results from, the 300 USRT unit. The combination of a fully shrouded model impeller and a parallel-wall vaneless diffuser was used. Figure 6 shows the fully shrouded impeller that was used in this set of experiments. The impeller had 14 blades, with angles backswept 50* from the radial direction at the exit, and a diameter of 265 mm (10.43 in.). The diameter of the vaneless diffuser was 376 mm (14.8 in.) and the width of the diffuser was the same as the impeller exit width. The design pressure ratio of this combination is 3.5 for C F C l l . The compressor inlet presure for CFC11 ranged between 20 and 30 kPa (2.9 to 4.4 psia),

3000

"E

2ooo

I000

_ -- i . . . ~ 1 ~ ficiency

Head r'

I

---o-- CFCII

---o----o-- HcFC 123

i i 0.9

I I 1.0 I.I

Volume flow-rote, Qs/Qsd .2

0 .9 "-

G

0.8 ®

0.7

Figure 7 Comparison of head rise and efficiency (same impeller): Q~, design flowrate for CFC11; r/~, effÉciency for CFCI 1 at Q~ Figure 7 Comparaison de I'blbvation de pression et de I'e~cacitb (m6me roue) : Qsd, dbbit-volume nominal du R l l ; qcd, efficacitb du R l l au d~bit nominal

and that for HCFC123 ranged between 18 and 28 kPa (2.6 to 4.1 psia). The compressor inlet temperature was between 276.2 and 279.2 K (37.4 and 42.8°F) and the rotational speed was 15 700 rev. min- t for both refriger- ants.

Figure 7 summarizes the experimental results. The general tendency is the same as that for the 300 USRT unit. The difference in performance is summarized in Table 1, based on the levels of CFC11. The choke flow- rate for HCFC123 decreased by 8%. The width of flow-

Rev. Int. Froid 1992 Vo115 No 4 201

An aerodynamic investigation of a centrifugal compressor for HCFC123." K. Nakagawa et al.

Table 1 Performance of HCFC123 Tableau 1 Performance du R123

Qs (choke) Qs (surge) Qs (ch.)/Q, (su.) Haa (surge) r/~ (max.)

0.92 0.95 0.97 0.94 0.93

Equivalent CFC11 values are taken as 1.0

Table 2 Mach number and diffusion ratio Tableau 2 Nombre de Mach et taux de diffusion (m~me roue et m~me vitesse de rotation)

Gas MI~ w~/wl (CFCI 1 based)

CFC11 1.08 1.0 HCFC123 1.16 1.15

Same impeller and same rotational speed

9.2 mm

8.1 mm

For HCFC 125

For CFCll

.J

I

I I I

Diffuser I

range, expressed by choke flow-rate divided by surge flow-rate, decreased by 3%. The head rise at the surge point dropped by 6%, and the maximum efficiency dropped by 7% for HCFCI23 . The molecular weight of HCFC123 is 11% heavier than that of CFCI 1, making it easier to compress HCFC123. This difference seems to have caused the following problems in the compressor performance:

1. a decrease in the choke flow-rate caused by HCFC123 's lower sonic velocity; 2. a drop in the impeller efficiency caused by an increased diffusion ratio (W~s/W2); 3. a diffuser pressure recovery drop caused by severe velocity distortion at the impeller exit.

M

O

g ro

The flow-rate at the surge point for HCFC123 was lower than that for CFC 11, and the ratio of the choke flow-rate to the surge flow-rate decreased by only 3%. This is the only qualitative difference between the results of the 300 USRT unit and the model compressor. The unexpectedly wide flow-range suggests that increase of the impeller diffusion ratio and the deteriorated impeller exit flow- distribution did not directly lead to an earlier surge onset of the impeller and diffuser combination.

The three problems given above are expected to be solved by an impeller especially designed for HCFC123, as pointed out by Akitani and Nakaiwa ~. To examine this point quantitatively, a comparison of the flow situa- tion in the impeller is made in Table 2. The diffusion ratios were calculated at Q,/Qsd = 0.95 by one dimen- sional analysis 4. The diffusion ratio for HCFC123 is 15% larger than that for C F C l l and the inlet tip Mach number is 8% higher. The inlet tip Mach number for HCFC123 is 1.16, which is high for a radial impeller. This high Mach number could cause an HCFC123 performance drop from shock-wave generation added to the effects o f an increased impeller diffusion ratio. Another experiment was made to clarify the dominant effect and its results are reported in the following subsection.

Improvement o f efficiency by impeller diffusion ratio adjustment

A trial compression was made to improve the lower H C F C 123 compression efficiency. The te~t impeller was

Figure 8 Impeller modification Figure 8 Modification de la roue

Table 3 Mach number and diffusion ratio (modified impeller) Tableau 3 Nombre de Mach et taux de diffusion (roue modifike)

Gas M,~ w,/w2 (CFCI I based)

CFCI 1 1.07 1.0 HCFCI23 1.15 1.0

a half-shrouded type with 18 blades and was originally designed for CFC11. The design pressure ratio was three when a vaneless diffuser was used. Its diameter was 300 m m (11.81 in.), and the exit backswept angle was 30 ° from the radial direction. The impeller diffusion ratio was controlled by decreasing the original exit blade height. The meridional shape is shown in Figure 8. Table 3 compares the calculated diffusion ratios and the Mach numbers at the design conditions. The suction volume flow-rate at the design point and the rotational speed (12600 rev min -~) were the same for C F C l l and HCFC123. The inlet-tip relative Mach-number level was the same as that of the fully shrouded impeller presented in the previous subsection. In this experiment, a channel diffuser was combined to aim for a higher efficiency. The installed impeller and diffuser are shown in Figure 9. The diffuser's outer diameter was 465 mm (18.31 in.), and its



Figure 9 Impeller and diffuser Figure 9 Roue et diffuseur

inner diameter was 1.1 times the impeller diameter. The compressor's inlet temperature and pressure were the same as those of the fully shrouded impeller presented in the previous subsection.

Figures 10 and 11 show the results of experiments, in which the impeller exit blade height was reduced as shown in Figure 8 and Table 3. It can be seen in Figure 10 that the head rise and stage efficiency of HCFC123 and CFC11 were almost the same near the design flow-rate. Figure 11 shows the comparison of pressure ratios. The pressure ratio of HCFC123 is about 10% higher than CFC11 near the design flow-rate. This change of pressure ratio is the same as the change of impeller outlet width. This suggests that the head rise and efficiency recovery of HCFC 123 are the result of the impeller diffusion-ratio adjustment. The surge flow-rate of HCFC123 is 15% higher than that of CFCl l and the head rise of HCFC123 is 4% higher than that of CFC11. The choke flow-rate of HCFC123 is a little higher than that of CFC11.

Why is the surge flow-rate of HCFC123 higher than that of CFCl l? (The possibility of impeller stall from shock-wave generation is small.) Because the surge flow- rate of HCFC123 is lower than CFC11 in the previous section, where the impeller inlet-tip relative Mach- number is equal to the level of the impeller stated in this section. On the contrary, it is possible that diffuser stall caused this difference. Two reasons are considered for this difference. One is that the diffuser was matched unintentionally to a slightly higher flow-rate for HCFC123 than for CFC11. The other is the difference in Mach numbers.

The impeller exit absolute Mach number was around one for CFCl l , and 3% higher for HCFC123. This higher Mach number of HCFC123 might have some effect on the diffuser-stall incidence-angle at lower flow- rate. To verify the possibility of diffuser stall, the diffuser inlet volume flow-rate was increased a little by reducing

2800

2000

Z

=;

I000

%

Heod rise " x \ Heod rise \~ \~o

i

I I

CFCII 100% design speed

HCFC 123 100%

HCFC123 97.5%

I I I I 0.7 0.8 0.9 1.0

Volume flow-rote Os/Osa

I.I

0.9

0.8

-- 0.7

I I.I

1.0 .~

~o

Figure 10 Comparison of head rise and efficiency (modified impeller) Figure 10 Comparaison de l'blbvation de pression et de l'eff~cacit$ (roue modifi$e )

the pressure ratio slightly. The rotational speed was lowered by 2.5% for this experiment. This slight change reduced the surge flow-rate by 12%, as expected, and the choke flow-rate by 5%. The pressure ratio at the surge point dropped by 5%, as shown in Figure 11. In general, when the choke occurs in the diffuser, the choke flow- rate reduction is larger than the rotational-speed reduction. With this background, it can be said that the flow choked in the diffuser, or was at least very near to the choke condition, at design rotational speed, so that the amount of diffuser mis-matching is very small, if any. The reduction of surge flow-rate is larger than the reduction of surge pressure ratio. This means that the surge volume flow-rate at the diffuser inlet became smaller after rotational-speed reduction. This is the main reason for the greater tendency for the diffuser to stall for HCFC123. The mis-matching or the difference in Mach numbers, however, is not clear within the scope of this experiment.

Although combining the impeller and channel diffuser caused some complexities, basically the same level of flow-rate, head rise and efficiency can be achieved by controlling the impeller diffusion ratio. As stated above, the Mach number at the diffuser inlet may have some effect, but it was not as apparent as that of the diffusion ratio within the scope of this experiment.

Rev. Int. Froid 1992 Vo115 No 4 2 0 3


o 4 -

E ==

#.

4.0

3.5

3 .0

2.5

\

\ \

\ \

CFClII 100% Design speed

I I I I I I

,, I I I

- - O n HCFC 123 100%

HCFC 123 97.5%

I I I t 0.7 0.8 0 .9 1.0

Volume f low- ra te , O s / Osd

I I.I

Tandem impeller

Radial impeller

/ /

Axial impeller / / / / /

't] _ J V / J \ \ i I

ffJ

, ¢

Figure 12 Comparison of tandem and radial impellers Figure 12 Comparaison entre la roue radiale et la roue en tandem

:6

0

Figure 11 Comparison of pressure ratio Figure 11 Comparaison du taux de compression

Table 4 Mach number Tableau 4 Nombre de Mach

Evaluation o f a tandem impeller

The thermodynamic properties of HCFC123 require about a 20% larger compressor suction volume flow-rate per unit of refrigerating capacity. This is not a difficult problem when there are no design constraints on the impeller size. In general, when the volume flow-rate is increased without impeller diameter enlargement, the relative velocities increase too. According to the results in the previous section, an efficiency drop caused by a Mach number increase is not significant. Therefore, a H CF C 123 compressor can be developed with no enlargement or efficiency drop.

Higher efficiency, however, is always desirable. Thus an experiment using a tandem impeller that can absorb the volume flow increase without causing diameter enlargement was made. A tandem impeller is a combination of an axial impeller and a radial impeller. There are two aerodynamic advantages to this type of impeller, as reported by Klassen et al. 5. One is that transonic axial compressor technology can be applied to diffuse the inlet transonic flow into a subsonic one with high efficiency. The second is that the boundary layer on the axial impeller scatters into the main flow and a fresh boundary layer develops on the radial impeller. This can reduce the losses caused by boundary layer development.

A tandem impeller and a radial impeller were com- pared in a centrifugal chiller with CFC11. The blade exit backswept angles and peripheral velocities of the impellers were the same. The diameter of the tandem impeller was 460 mm (18.11 in.) and the diameter of the radial impeller was 510 mm (20.08 in.). Meridional shapes of these impellers are shown in Figure 12. The tandem impeller's relative inlet-tip Mach-number is

Impeller M~

Radial impeller 1.01 Tandem impeller 1.17

Figure 13 Tandem impeller Figure 13 Roue en tandem

shown in Table 4. The Mach number level is the same as those in Tables 2 and 3. These impellers were combined with vaneless diffusers. The tandem impeller is shown in Figure 13 and the experimental results are shown in



2500

= 2000

o

1500

t o n ° . o \1

Efficiency

I I I 180 200 220

Volume flow-rate, Qs (m3 rain-I)

.o

o

0

- I.I

- I

- 0 . 9

Figure 14 Performance of tandem impeller: r/,d, efficiency at design flow-rate with radial impeller Figure 14 Performance de la roue en tandem: ~,d, e~cacit~ au d~bit- volume nominal avec la roue radiale

increase. Further investigation of tandem impeller technology may provide a possible and effective solution to this problem.

Conclusions

When a CFCl l compressor was used to compress HCFC123, head rise, efficiency and volume flow-rate decreased.

When the diffusion ratio of an impeller was adjusted to match that for CFC11, HCFC123 compression performance recovered to the level of that for CFC11.

A tandem impeller with a transonic inducer is shown to have promise to compensate for most of the increase in volume flow-rate and theoretical compression work resulting from the difference in the thermodynamic properties of CFC11 and HCFC123.

Acknowledgements

The authors express their appreciation to Mr Yutaka Ito of Hitachi Research Laboratory, Hitachi Ltd, and Mr Sigeo Sugimoto and Mr Masatoshi Terasaki of Tsuchi- ura Works, Hitachi Ltd, for their suggestions and helpful discussions at various stages of this investigation.

Figure 14. The tandem impeller achieved the same flow- range, a slightly higher head rise and an efficiency increase of about 2% above that of the radial impeller. The tandem impeller was more compact in diameter than the radial impeller by 10%. This means that if the comparison is made with the same diameter, the tandem impeller can compress 20% higher flow-rate than the radial impeller.

When CFC11 is replaced by HCFC123, it is impos- sible to avoid an increased flow-rate per unit refrigerating capacity and a theoretical compression-work

References

1 Akitani, T., Nakaiwa, S. Development and application of CFC 11 substitution refrigerant (in Japanese) Chem Indus Econ (March 1989) 71-74

2 Soave, G. Equilibrium constants from a modified Redlich- Kwong equation of state Chem Engng Sci (1972) 27 1197

3 Asahi Glass Co Ltd A F Tech Rep AF123, AA-TIO0I-1 (1989) 4 Mishina, H., Gyobu, I. Performance investigation of large capa-

city centrifugal compressors A S M E Paper No 78-GT-3 (1978) 5 Klassen, H. A., Wood, J. R., Schumann, L. F. Experimental

performance of a 13.65-centimeter-tip-diameter tandem-bladed sweptback centrifugal compressor designed for a pressure ratio of 6 N A S A Tech Paper 1091 (1977)

Rev. Int. Froid 1992 Vo115 No 4 205

Documents

An aerodynamic investigation of a centrifugal compressor for HCFC123