Journal of

Turbomachi nery
Influence of the Reynolds Number on the Performance
of Centrifugal Compressors
1
L. Sapiro.
2
I would like to reiterate and expand the com-
ments I made to Dr. Casey in Anaheim during his oral presen-
tation of this paper.
There is a contradiction in Fig. 6 of the proposed allowable
range of application of the correction formula for Reynolds
number. The lower limit of tolerance shown on that figure
does not correlate with the upper limit of tolerance.
Consider a compressor to be tested at a lower Reynolds
number than at the operating conditions. If the operating
Reynolds number is 5.0 x 10
5
, the lower curve indicates that
the accuracy of the correction formula is valid down to 10 per-
cent of that value, which is 5.0 X 10
4
.
Assume now that the same compressor is to be tested at a
higher Reynolds number than at the operating conditions. If
the operating Reynolds number is 5.0x 10
4
, the upper curve
indicates that the accuracy of the correction formula is valid
only up to about 3.5 times that value, which is 1.6x 10
5
.
One would expect the obtain a value of 10. This would be
the required value to make the test Reynolds number equal to
the original Reynolds number of 5.0X 10
5
in the previous ex-
ample. Similar discrepancies occur at other values.
For consistency, the upper curve should be much steeper or
the lower curve should reach the 10 percent ratio at a higher
Reynolds number value (more conservative) than 5.0xlO
5
.
The second approach appears to be the obvious one. Figure 7
shows the lower curve that correlates with the upper curve.
It should be noted that the same discrepancy between upper
and lower curves appears on the equivalent figure of Wiesner's
paper ([1], Fig. 8). That figure is more conservative than Fig. 6
in this paper, as shown in the comparison of Fig. 8. On the up-
per curve, the difference is substantial. At a value of 7.7 x 10
5
,
the map in Fig. 6 indicates a limit of 100 times, whereas the
map in Fig. 8 of [1] indicates 8.5. To obtain correlation be-
tween the upper and lower curves of his Fig. 8, Wiesner's
lower curve should be modified as shown on Fig. 9. A com-
parison of Fig. 6 and Wiesner's Fig. 8, as modified, is
presented in Fig. 10.
I would appreciate any comments on the reasons for the up-
per curve of Fig. 6 being so much more optimistic than the up-
per curve in Fig. 8 of [1], and I recommend that equations be
given for these curves of Fig. 6 to allow their use in com-
puterized procedures.
I agree fully with the additional limitation imposed on the
lower curve of Fig. 6. The minimum ratio between the
Reynolds number at test and the specified value should never
be lower than 10 percent, even at the highest specified Re.
CENTRIFUGAL COMPRESSORS
o
ft
cc
1
obay
==^
=
= =
EE
1
; i
;|
!
1
E
A
STRUBt
if
l Uf c s
\ 5TRUQc t-o
~-
;i
EEE
: :
EE
^ r
-. v
EE
: : z =
\ H
:
:
EEEE
|{
!i
=i
i l UPPER CURVE J
Tt JF^
II
EH;
REYNOLDS NO. SPECIFIED Rsp
Fig. 7
CENTRIFUGAL COMPRESSORS
100-

&
0.01
OMh
1
=
)'
=
=

1 3* 10"
/
:
f f f HF
==i
J
s
2Sdi
| [ ] |
ana
10'
v
i >t
, , | | | | wi E Ml B
" i Sl IfUlWiSH
K>' I ?
' By R. A. Strut), L. Bonciani, C. J. Borer, M. V. Casey, S. L. Cole, B. B.
Cook, J. Kotzur, H. Simon, and M. A. Strite, published in the October 1987
issue of the JOURNAL OF TURBOMACHINERY, Vol. 109, No. 4, pp. 541-544.
2
Consulting Design Engineer, Solar Turbines Incorporated, San Diego,
CA 92138; Mem. ASME.
REYNOLDS NO. SPECIFIED Rs p
Fig. 8
That prevents excessive measurement inaccuracy when the
reduction in Reynolds number is accomplished by lower suc-
tion pressure and, thus, lower mass flow and power. My own
experience has been that this inaccuracy tends, in most cases,
to produce overly optimistic efficiency values [10].
This paper's formula for correcting efficiency, regardless of
its accuracy or its comparison with other proposed formulae,
has the inconvenience of being an implicit equation. It would
be desirable to simplify it and convert it to an explicit form.
When considering the formula for correcting the flow coef-
ficient, it is arguable whether a correction for use in the ASME
PTC-10 is justified at all. Although the flow position at peak
efficiency may vary with the Reynolds number, in most cases
280/Vol. 110, APRIL 1988 Transactions of the ASME
Copyright 1988 by ASME
Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 07/21/2014 Terms of Use: http://asme.org/terms
CENTRIFUGAL COMPRESSORS
10* B* 10
s
10 ' K)' 10" 10'
REYNOLDS NO. SPECIFIED Rasp
Fi g. 9
CENTRIFUGAL COMPRESSORS
Fig. 10
the efficiency curve at both sides of the peak value does not
slope fast enough to require an adjustment of meaningful
magnitude. Applications at machine Mach numbers below
0.8, using vaneless diffuser staging, fit into that category.
The same argument about the justification of correcting the
flow coefficient, given the generalization of the correcting for-
mulae, applies to the work factor. After predicting and testing
more than 3000 compressors at ambient air conditions and
then in closed loop or field conditions, I have found no reason
also to correct the flow coefficient or the work factor value. I
simply apply the same correction to the head coefficient as to
the efficiency. However, my approach is to include this effect
of Reynolds number in the performance prediction prior to
the test, to affect the stage matching, rather than as a correc-
tion of results after the test. My correction formulae take into
consideration the specific speed of each stage ([10, 11], and
discussion of Wiesner's paper [1], pp. 394-395).
In all, the authors of this paper have made a commendable
contribution to the still unresolved debate about which for-
mulae and allowable ranges should be adopted to replace the
formulae and allowable ranges in the present ASME PTC-10.
References
10 Sapiro, L., "Review of Factory Aerodynamic Tests for Verification of
Centrifugal Gas Compressor Field Performance Prediction," ASME Paper No.
85-GT-154, 1985.
11 Sapiro, L., "Preliminary Staging Selection for Gas Turbine Driven Cen-
trifugal Gas Compressor," ASME Paper No. 73-GT-31, 1973.
Jour nal of Turbomac hinery
Authors' Closure
The authors are very grateful to Dr. Sapiro for his kind
remarks and very useful comments on several aspects of their
paper.
Sapiro points out an inconsistency in Fig. 6 concerning the
proposed allowable range of application of the correction for-
mulae. This inconsistency is purely academic in nature and has
no influence on the application of the correction method. The
same inconsistency is also present in nearly all previous recom-
mendations on allowable range of application of correction
formulae for Reynolds number effects (see, for example,
Wiesner [1] and Davis [6]). In all practical cases, workshop ac-
ceptance tests are carried out with reduced suction pressure
and inlet density and the test Reynolds number is invariably
lower than the specified Reynolds number. Acceptance tests at
higher inlet density and pressure are not usual because of
mechanical limitations from the casing and pipework design
pressure or the power required from the driver. As a result on-
ly the lower half of Fig. 6 (with Re,/Re
jp
< 1.0) has any prac-
tical significance for centrifugal compressor acceptance tests.
The allowable range given in the lower part of Fig. 6 has
been derived by noting that the ICAAMC correction formula
for efficiency is most accurate when the losses (1 - T;) do not
change by more than 20 percent from those at the test condi-
tions. It can be shown from the equations given in the paper
that the allowable range of Reynolds number variation for a
20 percent change in losses is a function of the relative
roughness (Ra/b
2
) of the stage. Strictly speaking, there is a
whole series of curves for the allowable range, each of which
depends on the relative roughness. In the interests of simplici-
ty a single curve has been selected that corresponds to a typical
relative roughness. The allowable range in the upper part of
Fig. 6 has been chosen as a compromise between that deter-
mined from the accuracy of the correction formula and that
determined from mechanical considerations (available driving
power, casing and pipework design pressure, etc.). Other com-
promises are possible for both the upper and the lower curves
in Fig. 6, and for this reason we do not think it is worthwhile
to give equations for the allowable range. We do feel,
however, that a smaller allowable range, as suggested by the
various limits given in Figs. 7-10 of Sapiro, is not necessary
when using the very accurate correction formulae given in the
paper.
Sapiro mentions that the formula for correcting the efficien-
cy is somewhat inconvenient as it is implicit. In fact, there is
no difficulty in writing an explicit form of the formula for cor-
recting the efficiency (equation (4)), as follows:
It is the Colebrook-White equation for friction factor (equa-
tions (6) and (7)) that is inconvenient, as the value of the fric-
tion factor appears on both the right and left-hand sides of the
equation. This was discussed in our committee meetings, and
we decided to adopt the Colebrook-White formula because it
is generally accepted as accurate by all fluid dynamicists and
no accurate and simple alternative was available at the time.
The inconvenience of an implicit formula is of no concern in
most computerized test procedures, as a straightforward itera-
tion can be incorporated. Since the last meeting of our work-
ing group (Oct. 1982), a suitable explicit formula for the fric-
tion factor of a turbulent pipe flow has been published [12], as
follows:
l/VX=-1.8 1og
e
[(6.9/Re) + (V(3.7
J
D,))
111
]
This gives excellent agreement with the Colebrook-White
equation, the maximum error being less than 1.5 percent,
and is ideally suitable for quick calculations on a pocket
calculator.
APRIL 1988, Vol . 110 /281
Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 07/21/2014 Terms of Use: http://asme.org/terms
Sapiro suggests that a correction of the flow coefficient is
not justified because the efficiency characteristic is always
very flat at the peak efficiency point. We do not agree. The
correction of the flow coefficient applies not only at the peak
efficiency point but also at other points along the
characteristic, for example at the choke point where the effi-
ciency characteristic is steep. The increase in the volume flow
of stage with increasing Reynolds number is most likely caused
by a decrease in the blockage of the boundary layers. A change
in boundary layer blockage can also have an effect on the
work input of the stage. These effects are clearly shown in the
experimental data presented in the most recent publications on
the effects of Reynolds number (see, for example, Fig. 1 of
Wiesner [1], Figs. 4 and 5 of Simon and Bulskamper [7], and
Fig. 1 of Casey [9]). The very accurate measurements available
to the working group showed clearly that corrections for effi-
ciency, head, and volume flow are necessary.
Sapiro comments that he applies his own correction method
for Reynolds number effects to the individual stage
characteristics prior to stage-stacking calculations rather than
as a correction of results after the test. The standard design
procedure for multistage compressors used by most manufac-
turers takes into account the effect of Reynolds number on the
stage matching in this way. Although the equations presented
in the paper could be used for this purpose, the paper does not
concern itself with this design problem. The main objective of
the working group was to propose suitable formulae to correct
the measured performance for the effect of a difference in
Reynolds number between the workshop acceptance test and
the specified conditions. This correction has to be applied to
the global performance data of each casing and not to the in-
dividual stage characteristics. Despite the many simplifica-
tions involved, the very good agreement between
measurements and predictions presented in the paper (Fig. 3)
show that the results of the ICAAMC correction equations are
within the range of accuracy of the measurements.
References
12 Haaland, S. E., "Simple and Explicit Formulas for the Friction Factor in
Turbulent Pipe Flow," ASME Journal of Fluids Engineering, Vol. 105, 1983,
pp. 89-90.
282/ Vo l . 110, APRIL 1988 Trans ac tions of the ASME
Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 07/21/2014 Terms of Use: http://asme.org/terms