Journal of Energy Storage 53 (2022) 105081

Contents lists available at ScienceDirect

Journal of Energy Storage

journal homepage: www.elsevier.com/locate/est

Research papers

Experimental and numerical investigation on off-design performance of a

high-pressure centrifugal compressor in compressed air energy
storage system
Chong Meng a, b, Zhitao Zuo a, b, c, d, Wenbin Guo a, c, Jianting Sun a, c, Qi Liang a, c,
Haisheng Chen a, b, c, d, *
a
Institute of Engineering Thermophysics, Chinese Academy of Science, Beijing 100190, China
b
University of Chinese Academy of Science, Beijing 100049, China
c
National Energy Large Scale Physical Energy Storage Technology R & D Center (Bijie), Bijie 551712, China
d
Nanjing Institute of Future Energy System, Institute of Engineering Thermophysics, Chinese Academy of Science, No.266 Chuangyan Road, Jiangning District, Nanjing
211135, China

A R T I C L E I N F O A B S T R A C T

Keywords: In order to explore the off-design performance of a high-pressure centrifugal compressor (HPCC) applied in the
Compressed air energy storage compressed air energy storage (CAES) system, the author successfully built a high-pressure centrifugal
High-pressure centrifugal compressor compressor test rig for CAES, whose designed inlet pressure can reach 5.5 MPa, and carried out some experi­
Adjustment inlet guide vanes
ments on adjustment of inlet guide vanes (AIGVs), adjustment of rotational speed (ARS) and adjustment of inlet
Adjustment of rotational speed
Adjustment of inlet pressure
pressure (AIP). Firstly, by analyzing the experimental data with the angle of AIGVs varying from − 20◦ to +50◦ , it
can be concluded that the mass flow range is widened by 41.18 %, the maximum efficiency is increased by 3.28
%, and the maximum pressure ratio can realize a slight increase. Then, the comprehensive performance map of
the HPCC with AIGVs is obtained through interpolating the existing experimental data. Subsequently, the
comparison between experimental and CFD results has been analyzed. Since the losses in the inlet expansion
joint and outlet expansion joint, the aerodynamic performance obtained by experiments is slightly lower than
CFD. Afterwards, the experiments on the ARS are carried out by three different speeds, and the change regulation
of the performance by ARS is obtained. Lastly, the experiments on the AIP are carried out by seven different inlet
pressure, when the inlet pressure of the HPCC is >50 % of the designed pressure, the results after simulated
modeling are very close to the experiment results of the designed pressure. But once the inlet pressure is lower
than 50 % of the designed pressure, the further the inlet pressure deviates from the designed pressure 5.5 MPa,
the further the performance curve obtained by simulated modeling deviates from the design working condition.
In this research, the experimental performance of a high-pressure centrifugal compressor applied in CAES under
off-design conditions was obtained for the first time, which is of great significance to the research and devel­
opment for the CAES system in the future.

1. Introduction created by geological location condition, and has been considered as one
of the most promising large scale physical EST [1–3]. As one of the most
Energy storage technology (EST) plays an important role in the large- important equipment in the large-scale and high-efficiency physical
scale application of renewable energy, and it is also regarded as the key energy storage system, the compression subsystem can convert electrical
technology to improve efficiency, safety and economy of conventional energy into internal energy and potential energy of air, which usually
power system. The compressed air energy storage technology (CAES), need to operate under off-design conditions [4,5]. Especially, the high-
which can effectively solve the challenges caused by the instability of pressure centrifugal compressor is the last stage of compression sub­
renewable energy systems and the strong volatility of the state power system, which need to work under high intake-pressure, high rotation
grids. Also, this successful technology can overcome the problems speed and large mass flow condition.

* Corresponding author at: Institute of Engineering Thermophysics, Chinese Academy of Science, Beijing 100190, China.
E-mail address: [email protected] (H. Chen).

https://doi.org/10.1016/j.est.2022.105081
Received 9 March 2022; Received in revised form 15 May 2022; Accepted 10 June 2022
Available online 20 June 2022
2352-152X/© 2022 Elsevier Ltd. All rights reserved.
C. Meng et al. Journal of Energy Storage 53 (2022) 105081

ķ Stabilization tank ļ high-pressure centrifugal compressor

ĸ Heat exchanger Ľ High-speed gearbox
Ĺ Orifice flowmeter ľ Asynchronism motor
Fig. 1. The high-pressure centrifugal compressor test rig. ĺ Damping chamber Ŀ Outlet regulating valve
Ļ Expansion joint ŀ Safety valve
Adjustment of inlet guide vanes (AIGVs) is one of the commonly used
adjustment methods for centrifugal compressors in industry, and many Fig. 2. The schematic diagram of a high-pressure centrifugal compressor
scholars have carried out some researches on AIGVs. Crouse [6] con­ test rig.
ducted an important experimental study on the adjustment of inlet guide
vanes and found that the inlet angle of air flow can be controlled But most of the above studies give few attentions on the high-
accurately to achieve higher efficiency for conventional compressor. C. pressure centrifugal compressor adopted in the CAES, which especially
Rodgers [7] found in the experiment that the stall state of impeller can usually operates at the high pressure and high speed environment. In the
be significantly improved and the surge boundary can be delayed by high pressure environment, the physical parameters of the real gas in­
AIGVs. The effect of negative pre-swirl on the performance and the flow side the HPCC have changed. So, in this paper, the off-design aero­
in a centrifugal compressor stage are investigated through both the dynamic performance of a high-pressure centrifugal compressor applied
experimental and numerical methods by Zhao [8]. Meng [9,10] have in CAES system is obtained by experimental and numerical methods for
done some researches on a high-pressure centrifugal compressor applied the first time, which is very important and meaningful.
in CAES through numerical methods, and got the adjustment strategies
of AIGVs and analyzed the internal flow field and loss distributions in­ 2. Experimental description and methodology
side the compressor. Guo [11,12] had carried out the performance
experiment of a low-pressure centrifugal compressor applied in CAES, 2.1. Experiment facility
and obtained the off-design performance and adjustment strategies of
AIGVs. In order to explore the aerodynamic performance of a high-pressure
Adjustment of rotational speed (ARS) is another common regulation centrifugal compressor in the compressed air energy storage system, the
method in centrifugal compressor. Wang [13] carried a variable speed author built a high-pressure centrifugal compressor test rig applied in
experiment on a small compressor, he found that the relationship be­ the CAES shown in Fig. 1 which is a closed loop arrangement. Fig. 2
tween power, mass flow and speed is not completely linear. Based on the shows the schematic diagram of the high-pressure centrifugal
variable speed characteristics of centrifugal compressor, Tang [14] compressor test rig. And the inlet pressure and temperature of the high-
proposed a two-step fitting method to accurately predict the perfor­ pressure air can be controlled and held constant independently under
mance of other working conditions. Tian [15] analyzed the flow sepa­ the effect of the gas stabilization tank ① and heat exchanger ②. A
ration loss of a centrifugal compressor and quantitatively explains the flowmeter ③ is used to measure the mass flow of the inlet air flow. The
mas flow and flow separation loss under variable speed. Chen [16] ob­ flow straightener inside the damping chamber ④ can make the inlet air
tained the stall and surge characteristics of a compressor at different flow uniform, which will flows into Expansion joint ⑤. Then the inlet air
speeds and throttle openings, and analyzed the dynamic characteristics flow is compressed in the high-pressure centrifugal compressor ⑥. The
of inlet distortion resulted by ARS. A. Cortinovis [17] investigated the opening of the outlet regulating valve ⑨ is reduced until it is close to the
anti-surge control strategies of compressor by ARS and established a surge condition, and the safety valve ⑩ can ensure the safety during the
compressor anti-surge control system based on model predictive control. experiments. The rotor is driven by a 4 MW motor ⑧ with continuous
Adjustment of inlet pressure (AIP) is another off-design operation speed control and a maximum shaft speed of 2200 rpm. A planetary
condition. Zhang [18] studied the influence of different inlet pressure on gearbox ⑦ with a gear ratio of 18.26:1 allows a maximum rotor speed of
the dynamic characteristics of a centrifugal compressor, and established 40,172 rpm. In addition, the high-pressure compressor stage consists of
a new dynamic similarity modeling method. the AIGVs with 11 blades, a shrouded impeller with 13 blades, a non-
By studying the influence of variable inlet pressure condition on the adjustable diffuser with 10 blades and an asymmetric volute.
compression process of one engine, Wei [19] thought that the energy
conversion efficiency can be improved by reducing inlet pressure
appropriately. Wang [20] analyzed the influence of different inlet 2.2. Measurement techniques
pressure on the performance of reciprocating compressors, and obtained
the quantitative relationship between compressor power, mass flow and Four probes are installed on the section S1 by the circumferential
inlet pressure. direction shown in Fig. 3(a), and Four probes are installed on the section
S2 shown in Fig. 3(b). There are three measuring holes on each probe

2
C. Meng et al. Journal of Energy Storage 53 (2022) 105081

(a) (b) (c)

(d) (e)
Fig. 3. The measurement system of pressure and temperature in the experiment.

shown in Fig. 3(c), which can accurately measure the total pressure and ( ) [( (γ− 1) ]
total temperature, and Advantech module is responsible for transferring TS1 PS2 γ
η= * − 1 (2)
the pressure and temperature data to the computer system. As shown in TS2 − TS1 PS1
Fig. 3(d), the pressure measurement system consists of eight probes,
ε represents the total pressure ration. η represents the isentropic ef­
Advantech pressure module and intelligent pressure scanner. As shown
ficiency. γ represents the isentropic exponent. The subscripts S1 and S2
in Fig. 3(e), the temperature measurement system consists of eight
means the section at the inlet and outlet of the compressor stage,
probes and Advantech temperature module. And the air mass flow is
respectively.
measured through Balanced flow meter, which has the advantages of
The error of the measurement parameters y can be computed by
high precision and large measurement range. The high-speed torque
transfer formula:
meter is used to measure the rotation speed and torque of the high-
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
pressure centrifugal compressor. dy
(
x1 ∂f dx1
)2 (
x2 ∂f dx2
)2 (
xn ∂f dxn
)2
The aerodynamic performances of the high-pressure compressor can = + +⋯+ (3)
y y ∂x1 x1 y ∂x2 x2 y ∂xn xn
be calculated by the measured pressure and temperature, the formulas
are as follows: where x represents the direct measurement parameters by equipment. In
P addition, the maximum error of total pressure ratio and isentropic effi­
ε = S2 (1)
PS1 ciency can be calculated by the following formulas, the maximum error
are ±0.09 % and ± 0.85 %, respectively.

3
C. Meng et al. Journal of Energy Storage 53 (2022) 105081

Fig. 4. The numerical model of a high-pressure centrifugal compressor.

√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
( )2 ( )2̅
dε dpS1 dpS2 Table 1
= + (4)
ε pS1 pS2 The grid sensitivity study.

√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ Grid number/(million) Total pressure ratio Isentropic efficiency

( )2 ( )2 ( )2 ( )2̅
dη dT S1 dΔTt dpS1 dpS2 Values Error(%) Values(%) Error(%)
= + +A 2 +A 2 (5)
η TS1 ΔTt pS1 pS2 5.73 1.3470 0.52 87.28 0.34
( )/[ 6.38 1.3540 0.38 87.62 0.21
γ− 1 ] 6.82 1.3591 0.15 87.83 0.08
A= 1 − (pS2 /pS1 )(1− γ)/γ
(6) 7.22 1.3611 – 87.91 –
γ

dΔTt = dT S1 + dT S2 (7)
parameters are as follows: the total inlet pressure is 5.5 MPa, and the
total temperature is 303.15 K.
2.3. Test procedure Adjustment of inlet guide vane, Impeller and Diffuser are meshed by
CFX-Turbogrid with structured hexahedron grids, and the Shroud cavity
The high-pressure centrifugal compressor experiments were carried and Backside cavity are meshed by ICEM with structured hexahedron
out at eight different IGVs (− 20◦ , − 10◦ , 0◦ , +10◦ , +20◦ , +30◦ , +40◦ , grids. The Volute is meshed by ICEM with unstructured tetrahedral
+50◦ ), three different rotation speeds (26,000 rev/min, 23,000 rev/min, mesh. At the regions close to blade surfaces and end walls, the grid is
20,000 rev/min) and seven different inlet pressures (5.5 MPa, 4.95 MPa, refined to obtain a good distribution of y plus value. The first boundary
4.4 MPa, 3.85 MPa, 3.3 MPa, 2.75 MPa, 0.55 MPa). The inlet gas pres­ layer offset is set as 2 × 10− 6 m and 10 layers in all are required to
sure of the high-pressure centrifugal compressor is maintained at 5.5 generate. The Spalart-Allamaras model turbulence is selected to solve
MPa through the gas stabilization tank. The inlet gas temperature is 3D steady Reynolds averaged Navier-Stokes (RANS) equations, which
maintained at 303.15 K through the heat exchanger. At every different
AIGVs, rotation speeds or inlet pressures, the compressor performance
was obtained by reducing the opening of regulating valve, which can
reduce the mass flow from the blockage condition to surge condition.
At the beginning of the experiment, the safety valve was fully closed
and the outlet regulating valve was fully open. When the pressure ratio
of high-pressure centrifugal compressor is close to 1 and the pressure
ratio rises rapidly once the mass flow decreases slightly, we take this
operation condition as the blockage condition. As the outlet regulating
valve gradually closes in the experiment, once hear the sharp stall
sound, we will open the safety valve immediately, and take this opera­
tion condition as the surge condition.

3. Numerical method and verification

3.1. Numerical method

In the present research, one high-pressure compressor adopted in

CASE system is selected as the research object, which is composed of
Adjustment of inlet guide vanes (AIGVs), Impeller, Diffuser, Shroud
cavity, Backside cavity and Volute, as shown in Fig. 4. Each component
used in this stage compressor is designed by in-house codes. The design Fig. 5. The grid sensitivity study.

4
C. Meng et al. Journal of Energy Storage 53 (2022) 105081

(a) Pressure ratio-Mass flow (b) Isentropic efficiency-Mass flow

Fig. 6. The aerodynamic performance of NASA LSCC.

(a) pressure ratio performance (b) efficiency performance

Fig. 7. The performance map with AIGVs tested in the experiment.

has been widely used in the numerical simulation of compressors due to 3.2. Numerical method verification
its good performance in convergence, stability and accuracy [21], and
the turbulence intensity of 5 % (Medium) is set at inlet boundary con­ NASA a low-speed centrifugal compressor (LSCC) impeller is applied
dition for the Spalart variable. The maximum wall y + after numerical to verify the reliability of the numerical method [23]. The inlet total
calculation is within 10, which satisfies the requirements of S-A turbu­ temperature is set at 288.15 K, the inlet total pressure is 0.1 MPa and the
lence model. The interfaces between the rotor and stator are set as ‘Stage mass flow is given at 30 kg/s. In addition, S-A model is selected as the
(mixing plane)’, which performs a circumferential averaging of the turbulence model. As shown in Fig. 6, it can be clearly seen that the
fluxes through bands on the interface. A second-order accurate dis­ numerical results of the high-pressure compressor are in agreement with
cretization (“High Resolution”) scheme is adopted for the advection the experimental results [24,25]. On the design condition, the relative
scheme and turbulence numerical option. The non-slip and adiabatic error between numerical and experimental results are <0.22 % in total
boundary conditions is given at all solid walls. The real gas model is pressure ratio and 0.37 % in isentropic efficiency respectively, which is
adopted in the calculation, and its property data are imported from NIST enough to demonstrate the high reliability and accuracy of the numer­
REFPROP [22] with self-programmed physical property table. The ical method adopted in this study.
uniform total pressure and total temperature with normal flow direction
are given at the inlet boundary condition. The average mass flow is 4. Results and discussion
specified at the outlet boundary. The leakage flow from the backside
cavity to the outer environment is assumed as 0 kg/s. Aiming to save the 4.1. Adjustment of inlet guide vanes
required time cost and ensure the accuracy in the simulation process, the
verification of grid independence is carried out. After completing the 4.1.1. Experimental results
grid sensitivity study, the total grid number of the numerical calculation The aerodynamic performance of the high-pressure centrifugal
model are about 6,820,000, shown in Table 1 and Fig. 5. The CFX is compressor with AIGVs were tested. Fig. 7 can show the high-pressure
chosen to calculate steady turbulent flow and the convergence criterion compressor performance of total pressure ratio and isentropic effi­
of numerical calculation is set as 10− 6. ciency under different pre-whirl angles in detail. The right endpoint of
each curve represents the blockage condition, and the left endpoint of
each curve represents the surge condition. When the angle of the AIGVs

5
C. Meng et al. Journal of Energy Storage 53 (2022) 105081

Table 2
The surge position at the different angles of AIGVs.
Angle/◦ − 20 − 10 0 +10 +20 +30 +40 +50

Mass flow /(kg/s) 17.94 16.33 15.57 13.23 11.54 9.95 9.02 8.93
Pressure ratio 1.396 1.395 1.391 1.393 1.391 1.383 1.379 1.370
isentropic efficiency/% 79.29 79.42 82.36 76.68 77.66 76.20 76.48 75.38

Table 3
The blockage position at the different angles of AIGVs.
Angle/◦ − 20 − 10 0 +10 +20 +30 +40 +50

Mass flow /(kg/s) 39.05 37.88 36.90 36.29 35.18 33.43 30.47 26.71
Pressure ratio 1.045 1.040 1.043 1.038 1.036 1.034 1.029 1.022
isentropic efficiency/% 14.05 13.74 15.23 14.69 14.70 15.31 25.7 13.61

(a) Isentropic efficiency-Mass flow (b) Isentropic efficiency-Pressure ratio

Fig. 8. Performance comparison between AIGVs and fixed inlet guide vanes.

is 0 , the mass flow ranges from 16.33 kg/s to 37.88 kg/s. And when the pressure ratio performance curve gradually shifts almost parallel to the
◦

mass flow is 22.49 kg/s, the maximum efficiency is 88.09 %. And the left. When the angles of AIGVs are further increased to 50◦ , the total
pressure ratio keeps increasing gradually with the decrease of mass flow pressure ratio performance curve of the compressor moves to the lower
until it reaches the surge point. The isentropic efficiency firstly increases left, the blockage flow decreases faster and faster, but the surge flow
to the highest efficiency point, and then decreases gradually. decreases slower and slower. Especially when the AIGVs gradually in­
When the AIGVs are adjusted in positive direction as shown in crease from 30◦ to 50◦ , the surge flow slowly decreases to about 9.00 kg/
Fig. 7a), as the angles of the AIGVs increases from 0◦ to 20◦ , the total s, and basically does not change. At the same time, the blockage flow
decreases significantly, and the total pressure ratio performance de­
teriorates significantly. When the AIGVs is adjusted in negative direction
in Fig. 6(a), as the angles of the AIGVs vary from 0◦ to − 20◦ , the total
pressure ratio performance curve of the compressor moves to the upper
right, the movement of blockage flow boundary is close to that of surge
flow boundary, and the blockage flow is increased to 39.05 kg/s.
From Fig. 7(b), when the angle of the AIGVs gradually increases from
0◦ to 30◦ , the isentropic efficiency performance curve gradually moves
to the upper left, the maximum efficiency increases gradually to 91.37
%, and the isentropic efficiency at small flow conditions are also
improved. As the angle of AIGVs is further increased to 50◦ , the
maximum efficiency decreases gradually to 84.12 %. At the same time,
the efficiency at small flow conditions does not change obviously.
Concerning the negative angle adjustment of the AIGVs, as the angle of
AIGVs ranges from 0◦ to − 20◦ , the isentropic efficiency performance
curve gradually moves to the lower right. On the whole, as for the surge
position shown in Table 2, the efficiency of the surge position ranges
from 75 % to 79 % at the different angles of AIGVs except for the 0◦ . As
for the blockage position shown in Table 3, the efficiency ranges from
13 % to 15 % at the different angles, which is very close to each other.
Fig. 9. Comprehensive performance of the high-pressure centrifugal According to the aerodynamic characteristics of the high-pressure
compressor by AIGVs. centrifugal compressor, there will always be a corresponding

6
C. Meng et al. Journal of Energy Storage 53 (2022) 105081

Table 4 be seen that by changing the angle of AIGVs, the maximum efficiency is
High efficiency operation range of the high-pressure centrifugal compressor by increased by 3.28 % and the mass flow range is widened by 41.18 %.
AIGVs. Moreover, the efficiency of the high-pressure centrifugal compressor has
Isentropic efficiency >90 % >80 % been improved at all mass flow conditions, especially at near surge and
Pressure ratio 1.32–1.37 1.26–1.4
blockage flow conditions. And in Fig. 8(b), the black curve represents
Mass flow/(kg/s) 16.55–22.75 12.15–28.75 the optimal operation curve under different total pressure ratios. It can
be seen that the AIGVs can not only expand the ranges of total pressure
ratio, but also improve the isentropic efficiency under all total pressure
maximum efficiency working condition under each mass flow or total ratios, and can achieve a maximum increase of 7.82 %.
pressure ratio. By connecting the highest efficiency points under each By interpolating the existing experimental data, the comprehensive
mass flows or total pressure ratios, the optimal operation curves of high- performance map of the high-pressure centrifugal compressor with
pressure centrifugal compressor under different mass flow or total AIGVs is obtained, as shown in Fig. 9. The circular curve in the figure
pressure ratio can be obtained. As shown in Fig. 8(a), the black curve represents the constant efficiency line. The red area is the operation area
represents the optimal operation curve under different mass flows. It can with the highest efficiency, where the isentropic efficiency can reach 90

(a) The angle of AIGVs is 0°

᧤b᧥The angle of AIGVs is +20° ᧤c᧥The angle of AIGVs is -20°

Fig. 10. The comparison of the aerodynamic performance by experimental and CFD.

7
C. Meng et al. Journal of Energy Storage 53 (2022) 105081

(a)IGVs=0° (b)IGVs=+20°
Fig. 11. The entropy distribution at small mass flow condition.

% and the orange area is the operation area where the isentropic effi­ When the angle of the AIGVs is +20◦ , the deviation between
ciency can reach 80 %. Table 4 shows the range of mass flows and experimental and CFD results is very small near the blockage conditions.
pressure ratio in the high efficiency operation area of the high-pressure The performance curve of numerical results moves further to the left
centrifugal compressor by AIGVs. than that of experimental results, as the angle of AIGVs gradually in­
creases to +20◦ . As a result, the experimental curve and CFD curve
4.1.2. Comparison between experimental and CFD almost coincide at the large flow conditions. But as the mass flow de­
In order to comprehensively analyze the aerodynamic performance creases, the deviation between experimental and CFD results gradually
of the high-pressure centrifugal compressor, the comparison between increases.
the total pressure ratio and isentropic efficiency obtained by experi­ When the angle of the AIGVs is − 20◦ , the deviation between
mental and CFD is shown in Fig. 10. It can be found that there is a good experimental and CFD results is small at the condition of small mass
agreement between the CFD results and experimental results. Since the flow, while big at the blockage condition. The performance curve of
losses in the leakage flow which leaks to the outer environment from the numerical results moves further to the right than that of experimental
backside cavity are not considered in the numerical simulation process, results, as the angle of AIGVs gradually varies from 0◦ to − 20◦ .
so the aerodynamic performance obtained by CFD is slightly higher than The small flow condition M = 18 kg/s is selected for analysis in
experimental results. And, the mass flow range obtained by CFD is Fig. 11. The high-entropy region mainly appears at the top of the
narrower than that measured by experiment, especially near surge impeller, and gradually expand along the flow channel in Fig. 11(a). And
conditions. the highest entropy region appears at the inlet of impeller shown, which
When the angle of the AIGVs is 0◦ , the CFD results are gradually means that there exist many losses at the impeller inlet. As the angle of
consistent with the experimental results with the decrease of mass flow. AIGVs gradually increases to +20◦ , the highest entropy region at the
Near the surge point of the CFD results, the total pressure ratio and inlet of impeller reduces significantly, and the entropy also decreases at
isentropic efficiency decrease rapidly. If the mass flow continues other sections shown in Fig. 11(b). In order to clearly observe the in­
decreasing to 17.65 kg/s, the CFD results cannot converge. But in the ternal flow inside the impeller, the three-dimensional streamlines of the
experiment, the high-pressure centrifugal compressor can operate at this relative velocity are shown in Fig. 12 when the mass flow is given as 18
mass flow. If the mass flow continues to reduce, the high-pressure cen­ kg/s. From Fig. 12(a), the internal flow field inside the high-pressure
trifugal compressor can still operate continuously until the mass flow impeller deteriorates significantly. Especially on the suction side of
drops to 15.57 kg/s. the impeller, there are many huge backflow vortices, which is also

(a)IGVs=0° (b)IGVs=+20°
Fig. 12. The 3D streamlines at small mass flow condition.

8
C. Meng et al. Journal of Energy Storage 53 (2022) 105081

(a) The performance map with different speeds at 0° tested in the experiment

(b) The performance map with different speeds at +20° tested in the experiment
Fig. 13. Experimental performance of the HPCC by ARS.

consistent with Fig. 12(a). When the angle of AIGVs increases to +20◦ , kg/s to 12.76 kg/s. When the angle of AIGVs is set as 0◦ , the blockage
the huge backflow vortices on the suction surface gradually disappear. flow decreases from 36.9 kg/s to 31.14 kg/s, and the surge flow de­
creases from 15.57 kg/s to 10.61 kg/s. When the angle of AIGVs is set as
+20◦ , the blockage flow decreases from 35.18 kg/s to 29.42 kg/s, and
4.2. Adjustment of rotational speed the surge flow decreases from 11.54 kg/s to 7.58 kg/s. The movement of
blockage flow boundary is close to that of surge flow boundary, which
In order to investigate the performance of the high-pressure cen­ means that the width range of the overall mass flow is approximately
trifugal compressor under different rotation speeds. Under the different unchanged. As the angle of AIGVs varies from +20◦ to +40◦ , the
angles of AIGVs as shown in Fig. 13, the performance experiments are blockage flow decreases from 31.13 kg/s to 24.68 kg/s, however, the
carried out when the rotation speed is set as 26,000 rev/min, 23,000 surge flow decreases very little, only from 7.58 kg/s to 7.1 kg/s. Thus it
rev/min and 20,000 rev/min respectively. As the rotating speeds grad­ can be concluded that when the inlet pressure is 5.5 MPa, the inlet mass
ually decreases from 26,000 rev/min to 20,000 rev/min, the total flow of 7 kg/s is basically close to the limit of the surge boundary of the
pressure ratio curve gradually moves to the lower left, the isentropic high-pressure centrifugal compressor applied in CAES.
efficiency curve gradually moves to the upper left, and there is a slightly Since the variable speed law of the HPCC under different IGVs is
increase in the maximum efficiency. On the whole, the isentropic effi­ basically the same, the internal flow analysis on different rotation speeds
ciency increases significantly at the small flow conditions, and the is carried out when the angle of IGVs is 0◦ . The small flow condition M =
isentropic efficiency deteriorates rapidly at the large flow conditions. 18 kg/s is selected for analysis. In Fig. 14, As the rotation speed de­
When the angle of AIGVs is set as − 20◦ , the blockage flow decreases creases to 23,000 rev/min from 26,000 rev/min, the high-entropy
from 39.05 kg/s to 32.35 kg/s, and the surge flow decreases from 17.94

9
C. Meng et al. Journal of Energy Storage 53 (2022) 105081

(c) The performance map with different speeds at +40° tested in the experiment

(d) The performance map with different speeds at -20° tested in the experiment
Fig. 13. (continued).

(a)N=26000 rev/min (b)N=23000 rev/min (c)N=20000 rev/min

Fig. 14. The entropy distribution inside the high-pressure impeller under different speeds.

10
C. Meng et al. Journal of Energy Storage 53 (2022) 105081

(a)N=26000 rev/min (b)N=23000 rev/min (c)N=20000 rev/min

Fig. 15. The 3D streamlines inside the high-pressure impeller under different speeds.

The experiment results were shown in Fig. 16. As the inlet pressure
Table 5
decreases gradually from 5.5 MPa, the total pressure ratio characteristic
The different inlet pressure conditions.
curve moves to the left, and the blockage flow decreases evenly. And the
Percentage 100 % 90 % 80 % 70 % 60 % 50 % 10 % maximum pressure ratio point increases slightly with the decrease of
Total pressure inlet pressure, but the maximum efficiency almost changes little. In
5.5 4.95 4.40 3.85 3.30 2.75 0.55
/MPa addition, the isentropic efficiency curves and total pressure ratio curves
become steeper with the decrease of the inlet pressure.
Then, based on the design pressure of 5.5 MPa and design temper­
region reduces at the top of the impeller, the losses at the inlet of ature of 303.15 K, the simulated analyze on the experimental data at
impeller significantly reduce shown in Fig. 14(b). When the rotation each intake pressure are carried out, which are shown in Fig. 17. The
speed further decreases to 20,000 rev/min shown in Fig. 14(c), the high- simulated modeling formula is:
entropy zone at the top area inside the impellers further reduced, which √̅̅̅̅̅̅̅̅̅̅̅̅̅̅
means that the losses at the impeller inlet further decreases. Mm = M
5505370 T
(8)
In Fig. 15, as the rotation speed decreases to 23,000 rev/min, the P 303.15
internal flow field inside the impeller is significantly improved, and the
where M represents the mass flow measured by experiments. P repre­
huge backflow vortices on the suction surface gradually disappear.
sents the inlet pressure measured by experiments. T represents the inlet
When the rotation speed further decreases to 20,000 rev/min, the in­
pressure measured by experiments. Mm represents the target mass flow
ternal flow field at the inlet of impeller become more uniform than that
after simulated modeling.
of 23,000 rev/min and 26,000 rev/min.
From Fig. 17(a), as the inlet pressure decreases, the total pressure
ratio curve moves slightly to the left in the large modeled mass flow. But
4.3. Adjustment of inlet pressure in the medium or small modeled mass flow, the total pressure ratio curve
moves up. The further the inlet pressure deviates from the design pres­
In order to explore the off-design performance of HPCC in the CAES sure 5.5 MPa, the further the total pressure ratio characteristic curve
system under different inlet pressure conditions, different inlet pressures deviates from the design condition after simulated modeling. From
of the HPCC are controlled by gas stabilization tank as shown in Table 5. Fig. 17(b), the isentropic efficiency is very close to each other at the

(a) (b)
Fig. 16. The performance map with AIP tested in the experiment

11
C. Meng et al. Journal of Energy Storage 53 (2022) 105081

(a) (b)
Fig. 17. The performance map with AIP by simulated modeling the experiment results.

small or medium modeled mass flow. But in the large modeled mass to the upper left, and there is a slightly increase in the maximum
flow, the isentropic efficiency curve gradually moves to the left with the efficiency. When the HPCC has to work at a small conditions, the
decrease of inlet pressure. Moreover, the further the inlet pressure de­ internal flow state and the compression efficiency can be
viates from the design pressure 5.5 MPa, the further the isentropic ef­ improved by reducing the rotation speed, but cannot improve the
ficiency characteristic curve deviates from the design working total pressure ration.
condition. (c) As the inlet pressure decreases gradually from 5.5 MPa, the total
Overall, when the inlet pressure of the high-pressure centrifugal pressure ratio characteristic curve moves to the left, and the
compressor is >50 % of the designed pressure, the results after simulated blockage flow decreases evenly. And the maximum pressure ratio
modeling are very close to the experiment results of 5.5 MPa. However, point increases slightly with the decrease of inlet pressure, but the
when the inlet pressure is lower than 50 % of the designed pressure, maximum efficiency almost changes little.
especially at the 10 % of the design pressure, there will be more obvious (d) When the inlet pressure of the high-pressure centrifugal
deviation between the simulated results between the experiment results compressor is >50 % of the design pressure, the results after
of the 5.5 MPa, which may be the result of some property parameters of simulated modeling are very close to the experiment results of
the real gas under such high pressure are different from those of low 5.5 MPa. But when the inlet pressure is lower than 50 % of the
pressure environment. designed pressure, the further the inlet pressure deviates from the
design pressure 5.5 MPa, the further the performance curve ob­
5. Conclusion tained by simulated modeling deviates from the design working
condition, which may be the result of some property parameters
In this paper, in order to explore the off-design performance of a of the real gas under such high pressure are different from those
high-pressure centrifugal compressor(HPCC) used in the compressed air of low pressure environment.
energy storage (CAES) system, the author successfully built the first
high-pressure centrifugal compressor test rig for CAES, whose inlet
pressure can reach 5.5 MPa. And the experiments of AIGVs, ARS and AIP Declaration of competing interest
on the HPCC are carried out. The major findings of this study can be
summarized as follows: We declare that we do not have any commercial or associative in­
terest that represents a conflict of interest in connection with the work
(a) As the angel of the AIGVs increases from − 20◦ to +50◦ , the total submitted.
pressure ratio performance curve firstly shifts almost parallel to
the left, then moves to the lower left and the isentropic efficiency Acknowledgements
performance curve firstly moves to the upper left, then moves to
lower left. In addition, the efficiency of the surge position ranges This study is supported by The National Science Fund for Distin­
from 75 % to 79 % and the efficiency of the blockage position guished Young Scholars through grant no. 51925604, The Natural Sci­
ranges from 13 % to 15 % at the different angles, which is very ence Foundation of China through grant no. 52106278, and The Major
close to each other. Through AIGVs, the mass flow range is science and technology projects of Inner Mongolia through grant no.
widened by 41.18 %, the maximum efficiency is increased by 2021ZD0030.
3.28 %. In addition, AIGVs can also improve the isentropic effi­
ciency under all total pressure ratios, and can achieve a maximum References
increase of 7.82 %. By interpolating the existing experimental
data, the comprehensive performance map of HPCC with AIGVs is [1] H. Chen, T.N. Cong, W. Yang, et al., Progress in electrical energy storage system: a
critical review, Prog. Nat. Sci. 19 (2009) 291–312.
obtained, and the high efficiency range under different operation
[2] H. Chen, J. Liu, H. Guo, et al., Technical principle of compressed air energy storage
conditions of HPCC can be clearly seen. system, Energy Storage Sci. Technol. 2 (2013) 146–151.
(b) As the rotating speeds gradually decreases from 26,000 rev/min [3] X. Zhang, H. Chen, J. Liu, et al., Research progress in compressed air energy storage
system: a review, Energy Storage Sci. Technol. 1 (2012) 26–40.
to 20,000 rev/min, the total pressure ratio curve gradually moves
[4] H. Guo, Y. Xu, Y. Zhang, et al., Off-design performance and an optimal operation
to the lower left, the isentropic efficiency curve gradually moves strategy for the multistage compression process in adiabatic compressed air energy
storage systems[J], Appl. Therm. Eng. 149 (2019) 262–274.

12
C. Meng et al. Journal of Energy Storage 53 (2022) 105081

[5] X. Zhang, H. Chen, Y. Xu, et al., Huang, distributed generation with energy storage [16] Chen., Model and Simulation on Technology of Active Stability Control for
systems: a case study[J], Appl. Energy 204 (2017) 1251–1263. Aeroengine Based on Non-Constant Speed Condition[D], Northwestern
[6] J.E. Crouse, R.J. Steinke, Preliminary analysis of the effectiveness of variable- Polytechnical University, Xian, 2019.
geometry guide vanes to control rotor-inlet flow conditions [R], in: NASA TND- [17] A. Cortinovis, D. Pareschi, M. Mercangoez, et al., Model predictive anti-surge
3823, 1967. control of centrifugal compressors with variable-speed Drives[J], IFAC Proc.
[7] C. Rodgers n.d. Centrifugal Compressor Inlet Guide Vanes for Increased Surge Volumes 45 (8) (2012) 251–256.
Margin[R]:ASME Paper No.GT90-158. [18] X. Zhang, D. Ziviani, E.A. Groll, et al., Experimental validation and sensitivity
[8] J. Zhao, Z. Wang, G. Xi, et al., Investigation of performance and flow field of a analysis of a dynamic simulation model for linear Compressors[J], Int. J. Refrig.
centrifugal compressor under negative pre-swirl[C], ASME Paper (2015). No. 117 (2020) 269–380.
GT2015-43141. [19] W. Wei, R. Huang, et al., Analysis of compressed air braking process under variable
[9] C. Meng, Z. Zuo, W. Guo, et al., Research on regulation law of inlet guide vane in inlet pressure[J], J. Mech. Electr. Eng. 34 (6) (2017) 623–628.
high-pressure centrifugal compressor of CAES[J], J. Eng. Thermodyn. 42 (11) [20] W. Wang, W. Sun, Y. Liu, Analysis of energy saving and consumption reduction
(2021) 2834–2840. technology of reciprocating compresso[J] 8 (1) (2014) 049–056.
[10] C. Meng, Z. Zuo, W. Guo, et al., Numerical study on the influence of shroud cavity [21] W. Ma, D. Wang, LU L., Investigation on improving the capability of predicting
in the high-pressure centrifugal compressor for compressed air energy storage separation of Spalart-Allmaras turbulence model, Journal of aerospacePower 23
system[C], IOP Conf. Ser. Earth Environ. Sci. 804 (3) (2021), 032018. (8) (2008) 1474–1480.
[11] W. Guo, Z. Zuo, J. Sun, et al., Experimental investigation on off-design [22] E. Lemmon, M. Huber, Mc Li, MO. NIST reference fluid thermodynamic and
performance and adjustment strategies of the centrifugal compressor in transport properties—REFPROP[EB/OL], in: NIST Standard Reference Database
compressed air energy storage System[J], J. Energy Storage 38 (2021), 102515. 23, Version 9.1, 2013.
[12] W. Guo, Z. Zuo, X. Zhou, et al., Regulation law research on adjustable diffusers of [23] M.D. Hathaway, R.M. Chriss, J.R. Wood, et al., Experimental and computational
centrifugal compressor in CAES[J], J. Eng. Thermodyn. 42 (02) (2021) 335–341. investigation of the NASA low speed centrifugal compressor flow field,
[13] D. Wang, T. Cui, Experimental study on variable speed of a small compressor[J], J. Turbomach. 115 (1992) 527–542.
Fluids Eng. 000 (007) (1985) 15–19. [24] R.M. Chriss, M.D. Hathaway, J.R. Wood, et al., Experimental and computational
[14] Y. Tang, X. Li, X. Chen, et al., Fitting method of variable speed characteristic curve results from the NASA lewis low speed centrifugal compressor impeller at design
of a centrifugal compressor[J], J. Jiamusi Univ. 27 (4) (2009) 5. and part-flow conditions, J. Turbomach. 118 (1996) 55–65.
[15] Y. Tian, Study on separation loss of flow in a centrifugal compressor under variable [25] M.D. Hathaway, R.M. Chriss, J.R. Wood, et al., Laser anemometer measurements of
working conditions[J], Mech. Manag. Dev. 34 (1) (2019). the three dimensional rotor flow field in the NASA low speed centrifugal
compressor, in: NASA Report. No. NASA TP-3527, 1995.

13