The document summarizes modifications made to a two-stage centrifugal compressor to convert it into a single-stage compressor suitable for laboratory testing. Key modifications included replacing the refrigerant fluid with air, installing an external drive motor instead of the internal hermetic motor, and adding static pressure taps to the vaneless diffuser and volute casing. Experimental results showed the compressor was operating off-design, with the vaneless diffuser and volute being too large for the mass flow rates tested. Pressure maps revealed distortion in the diffuser and volute due to the tongue region, reducing stage performance.

1. 1
MODIFICATIONS OF A TWO STAGE CENTRIFUGAL COMPRESSOR FOR THE
VOLUTE AND VANELESS DIFFUSER STUDY
Hooman Rezaei
Turbomachinery Laboratory
Michigan State University
East Lansing, MI
Abraham Engeda
Turbomachinery Laboratory
Michigan State University
East Lansing, MI
Paul Haley
Turbomachinery Design Group
Trane Company
LaCrosse, WI
ABSTRACT
The objective of this work was to establish an advanced testing
facility for studying the flow structure and loss mechanism in a
single stage centrifugal compressor. A Trane’s CVHF 1280, two
stage centrifugal compressor was modified to a single stage for
laboratory environment. This modification included new fluid
medium, driving motor, inlet and outlet designs. In this study,
experiments were performed in order to evaluate the
performance of the vaneless diffuser and volute. However, the
modifications enable the flow structure investigation in all
components of the compressor as well. In addition the testing
facility would accommodate installation of the rest of the unit
for investigation of the flow in the vaned diffuser and return
channel of the two-stage compressor in the laboratory
environment.
Experiments were performed in three speeds and eight mass flow
rates per speed. Static and total pressures were measured at the
inlet and outlet of the stage. Static pressure distributions were
mapped on the vaneless diffuser and volute casings. These data
evaluates the characteristics of these components after the
modifications.
INTRODUCTION
A spiral-shaped volute is used in many applications of
compressors to collect the rotating flow, which discharges from
a diffuser downstream of the impeller, and to deliver it into a
single discharge pipe. Very few investigation results exist on the
performance and design of volutes in comparison to other
components of centrifugal compressors. This can be attributed
to the fact that the volute is a simple collecting device and all
the necessary diffusion has been achieved in the vaned or
vaneless diffuser upstream of the volute.
The volute is usually designed through the application of one-
dimensional analysis. A design objective is to achieve a uniform
pressure distribution at the volute inlet. This is usually attained
at the design flow rate only; at off-design conditions the volute
is either too small or too large and pressure distortion develops
circumferentially around the volute passage. The static pressure
distortions are transmitted to the diffuser exit, the impeller
discharge, and even through the inducer. Therefore, these
pressure distortions reduce the stage performance and have a
direct impact on diffuser and impeller stability. Thus, it is
necessary to perform more research in this area to improve the
performance of the centrifugal compressors.
NOMENCLATURE
P Static Pressure
P0 Total Pressure
r2 Impeller Tip Radius
φ Flow Coefficient (Q/Utip Dtip
2
)
θ CircumferentialAngle
1 Compressor Inlet
2 Volute Outlet
MODIFICATIONS OF THE COMPRESSOR
Trane Company donated a two-stage centrifugal compressor to
Michigan State University (MSU) for this study. The unit
operates with refrigerant gas R123 and is driven by hermetic
motor. The motor is installed on the casing and shares single
shaft with the compressor. In order to prevent working with
refrigerant and high-pressure ratios in laboratory environment,
air replaced the refrigerant gas and the unit was to be modified
to a single stage compressor.
Changing the working fluid and the number of stages forced the
volute into off design operating conditions, where the losses
were the highest and the non-dimensional parameters were far
from the design condition. In addition, the compressor no
longer operated in a refrigerant closed loop, which was the
design condition. In the new design the compressor was to be
driven with an external motor while the old motor casing was in
place. Therefore, a new extended shaft was designed and

2. 2
manufactured in order to replace the old shaft and to be coupled
with the new motor. The compressor was disassembled
completely and the new shaft with new bearings was installed
initially. A stand was designed and assembled behind the old
motor casing. The new motor was bolted to the stand with
required alignment for coupling with the new shaft. Power
provided to this motor with a frequency controller in the circuit
for variable speed operation. After coupling the motor and
compressor shafts, the motor was driven in a low speed to
check the lubrication and coupling of the shaft.
Fig. 1: Trane modified compressor
At the same time, the volute and vaneless diffuser casings were
equipped with pressure taps as described in the instrumentation
section. After initial test, the second stage impeller and vaneless
diffuser casing were assembled. A conic reducer was designed
and manufactured to reduce the inlet size of the impeller to 12-
inch diameter. Similar conversion was performed at the conic
diffuser outlet. Utilizing PVC pipes, inlet and outlet pipes were
provided for the compressor stage. The outlet pipe was
equipped with a valve for throttling the compressor. The inlet
pipe was equipped with orifice plate for measuring the mass
flow rates. (Fig. 1)
INSTRUMENTATION
1. PERFORMANCE INSTRUMENTATION
Four United Sensors Kiel probes KBC-8 were utilized in order to
measure the total pressure at each side of the stage. These
probes can measure pressures in flows up to Mach 1.0 with
some considerations for pitch and yaw angles. The outstanding
advantage of Kiel probes compared with other total pressure
probes is complete insensitivity to direction of the flow within
certain limits. Their yaw and pitch characteristics are generally
the same although stem interference on some designs will
change one from the other.
The probes were setup in this test rig such that the average
values of pressure were obtained by connecting the probes to
each other by a “T” and reading the value at the outlet of the
“T”. The total pressure probes were installed in the direction of
the mean flow path and immersed into the flow 1/3 of diameter of
the pipe based on ASME standards for flow measurement. Four
steel capillary static pressure taps with an outer diameter of 1/16
inch were inserted at 90-degree intervals into the 12-inch
diameter PVC pipe at the inlet and outlet. Static pressures were
measured by averaging the readings from the four pressure
taps. The surface of the taps were sanded and flushed to the
surface in order to reduce the error in reading the static
pressures.
θ=27
θ=72 θ=117 θ=162
θ=207
θ=252θ=297θ=342
θ=0
θ=60
θ=120
θ=180
θ=240
θ=300
(a)
(b)
Fig.3: Static Pressure Taps Locations (a) Vaneless Diffuser (b)
Volute
1.04
1.06
1.08
1.10
1.12
1.14
1.16
0 0.01 0.02 0.03 0.04 0.05 0.06
φ
P02/P01
2000 3000 3497
Fig. 3: Total pressure ratio versus flow coefficient for
different speeds
2. VANELESS DIFFUSER
Static taps were provided at every 60-degree circumferentially
on the vaneless diffuser shroud wall and different radial
locations from impeller exit to the diffuser bend. The holes were

3. 3
drilled for 3/16-inch I.D. steel pipe inserts and similar to inlet and
outlet static taps, the tips were sanded and inserted flush to the
surface of the diffuser cover.
There were pressure taps on the surface of diffuser cover that
were initially used in Trane’s design experiments. These taps
were distributed every 120 degrees at the inlet and outlet of the
diffuser. Pressure taps were provided in addition to the Trane
taps so that overall distribution included two taps installed at
each 0, 120, 240 degrees and six at each 60, 180, 300 degrees and
different radial locations. (Fig. 2)
3. VOLUTE
In the first step, when the volute inner surface was accessible
during the reassembling process, 3/8-inch holes were drilled at
45-degree intervals circumferentially and in different axial
locations. The holes were first tapped and filled with epoxy and
later the dried epoxy was drilled for 1/16-inch O.D. static
pressure tap inserts. Similar to the taps at the inlet and outlet,
the steel pipe inserts were cut and sanded on the tip. The epoxy
on the volute casing was sanded from inside prior to drilling and
the steel capillary pipes were inserted to be flush with the
surface.
4. PRESSURE SCANNER
Due to the large number of the points on the volute and
vaneless diffuser to be measured, two 16-channel differential
Scannivalve DSA3017 pressure scanners were utilized. The
DSA3000 series pressure acquisition systems represent the next
generation of multi-point electronic pressure scanning. Model
DSA3017 Digital Sensor Array; incorporate 16-temperature
compensated piezoresistive pressure sensors with a pneumatic
calibration valve, RAM, 16-bit A/D converter, and a
microprocessor in a compact self-contained module.
The microprocessor compensates for temperature changes and
performs engineering unit conversion. The microprocessor also
controls the actuation of an internal calibration valve to perform
on-line zero and multipoint calibrations. This on-line calibration
capability virtually eliminates sensor thermal errors with a long-
term system accuracy of ±0.05% full scale (FS). Pressure data
are output in engineering unit via Ethernet using TCP/IP
protocol. With the initial experiments at 3600RPM, the pressure
range of 2.5 PSID was selected for the units. Note that these
units measure the pressures with respect to a common
reference.
EXPERIMENTAL RESULTS
1. VANELESS DIFFUSER
Experiments performed for three speeds of 2000, 3000 and 3497
RPM. Fig. 3 shows the total pressure ratio versus flow
coefficient for these speeds. Fig. 4 shows circumferential
pressure distortion at vaneless diffuser inlet on the vaneless
diffuser shroud wall for 3497 RPM. At one flow coefficient
point, the data points should have collapsed to represent the
balancing point in the operation of the diffuser. Collapse of
these points would have shown zero circumferential static
pressure distortion at that operating condition. The 2000 and
3000 RPM cases had higher circumferential static pressure
distortion in comparison with the 3497 case. Fig. 5 represents
the diffuser circumferential outlet pressures for the same flow
coefficients. It is clear that the tongue region (θ=240) has
significant effect on the performance of the compressor by
creating distortion in the circumferential static pressure
distribution at the diffuser outlet. This sudden drop in the static
pressure at tongue region could have a direct effect on the flow
back to the inducer and impact the bearings with unbalanced
circumferential loads. In order to quantify the impact of the
tongue region, total pressure loss coefficient could have been
calculated if there were total pressure measurements available
locally. The flow quality in this region is discussed in [2] based
on the numerical simulations performed for the same operating
conditions.
100
101
102
103
104
105
106
107
108
109
0.020 0.025 0.030 0.035 0.040 0.045 0.050 0.055
φ
P(KPa)
0 60 120 180 240 300
Fig.4: Vaneless diffuser exit static pressure distortion for
3497 RPM
Fig. 6 shows the diffuser radial pressure distribution for the
same flow coefficients at 180-degree angle, which shows the
radial diffusion inside the vaneless diffuser. For mass flow rates
below the design point, deceleration will occur because the
diffuser is too large. Therefore, with increase in speed and mass
flow rate, the rate of diffusion decreases. In 2000 RPM case the
radial pressure distribution presented the possibility of
separated flow in the diffuser and numerical simulation results
proved this event as well [2].
2. VOLUTE
In order to map the static pressure distribution in the volute,
pressure taps were provided as far as the geometry of the volute
cross sections allowed. These taps were located on the flat
portions of the volute casing, where normal drilling on the
surface was possible. Therefore, depending on the
circumferential locations on the volute casing, one to three taps
were provided. These pressure ports were scanned for similar
mass flow rates and speeds as the diffuser experiments. Results
were averaged axially for every cross section at every operating
`

4. 4
point. Fig.7 and Fig.8 show the volute circumferential static
pressure distribution from the tongue region to conic diffuser
outlet for speeds of 2000 and 3497 RPM. The pressure
difference between the 0 and 360 angles indicates the pressure
gradient between the conic diffuser and tongue inlets, which
forces the flow to reenter the volute in this region. This effect
results in increasing losses in this region.
100
101
102
103
104
105
106
107
108
109
0 50 100 150 200 250 300
θ
P(KPa)
Choke 2nd 3rd 4th
5th 6th 7th Surge
Fig. 5: Static pressure distribution at vaneless diffuser exit and
3497 RPM
95
97
99
101
103
105
107
109
1.1 1.2 1.3 1.4 1.5 1.6
r/r2
P(KPa)
Choke 2nd 3rd 4th
5th 6th 7th Surge
Fig. 6: Vaneless diffuser radial static pressure distribution at
180-degree angle and 3497 RPM
The reader clearly observes that the performance of the volute
is improving with increase in mass flow rate and speed.
Diffusion occurs sharply in the cross sections with smaller areas
and decreases as the flow enters the cone inlet. One primary
conclusion would be that this volute is performing at off design
condition because the pattern of diffusion in the volute shows
that the volute is too large for these mass flow rates.
Experiments show that this is true for all speeds and mass flow
rates.
The ideal volute performance is collecting the flow leaving the
diffuser and providing uniform circumferential pressure
distribution at the vaneless diffuser exit. The uniformity of the
pressure will improve the impeller and vaneless diffuser
performances. The wavy shape of the plots for 2000-RPM speed
indicates the nonuniform circumferential distribution of the
static pressure in addition to the pressure drop across the
tongue region. Increasing the speed and mass flow rate,
improves the pressure distribution and therefore reduces the
loss in the stage operation. However, the pressure drop across
the tongue region is inevitable and depends on the design of
the tongue. Note that the reentrance of the flow to the volute at
the tongue region could help reduce and balance the
circumferential pressure distortion at tongue region.
99.0
99.5
100.0
100.5
101.0
101.5
102.0
102.5
103.0
0 50 100 150 200 250 300 350 400
θ
P(KPa)
Choke 2nd 3rd 4th
5th 6th 7th Surge
Fig. 7: Volute static pressure distribution for 2000 RPM
98
100
102
104
106
108
110
0 50 100 150 200 250 300 350 400
θ
P(KPa)
Choke 2nd 3rd 4th
5th 6th 7 th Surge
Fig. 8: Volute static pressure distribution for 3497 RPM
In the design of the tongue for this model, if the distance
between the tongue knife edge and beginning of the volute is
reduced, the length of the sharp pressure gradient region
decreases and would reduce the nonuniformity of the
circumferential pressure distribution.
CONCLUSION
A two stage centrifugal compressor was modified to single
stage with new design of inlet and outlet. The refrigerant flow
medium and internal driving motor were replaced by air and
external driving motor. The vaneless diffuser and volute were

5. 5
equipped with static pressure taps, which can accommodate the
five-hole and cobra probes for velocity measurements.
Experiments performed for three speeds of 2000, 3000 and 3497
RPM showed that the compressor is running at off-design
condition. Sharp diffusion pattern from static pressure maps in
the vaneless diffuser and volute showed that the compressor is
too large for the operating mass flow rates. The impact of the
tongue region on the vaneless diffuser and volute performances
was presented by the distortion of the circumferential static
pressure distribution at the diffuser exit. With increase in speed
and mass flow rates, the performance of these two components
was improved.
ACKNOWLEDGMENT
The authors would like to acknowledge of the Trane
Company support throughout this study by providing the
financial support and technical information of the compressor.
REFERENCES
[1] Rezaei H
“Investigation of the flow structure and loss mechanism in a
centrifugal compressor volute”
Ph.D. dissertation, Michigan State University, 2001
[2] Rezaei H, Engeda A, Haley P
“Numerical analysis of the flow inside a centrifugal
compressor’s vaneless diffuser and volute”
Submitted to International Mechanical Engineering Congress &
Exposition of ASME, New York, 2001