JRAME Vol 3 2015 No 2 Part 2 PP 2 (3)

Transactions of the TSME (2015) Vol. 3, No. 2, 54 63
Journal of Research and Applications in Mechanical Engineering
Copyright © 2015 by TSME
ISSN 2229-2152 print
DOI: XXXXXXXXXXXXXXXXXX
Research Article
NUMERICAL SIMULATION OF THERMO-FLUID DYNAMIC
ENTROPY GENERATION IN CENTRIFUGAL COMPRESSORS FOR
MICRO-TURBINE APPLICATION
K. Wasinarom
D. Boonchauy
J. Charoensuk*
Department of Mechanical
Engineering, Faculty of
Engineering, King Mongkut's
Institute of Technology
Ladkrabang, Bangkok, Thailand
10520
ABSTRACT:
The investigation on quantitative entropy generation in the streamwise
direction of flow passage in centrifugal compressors with different exit beta
angle was carried out under the operating condition of small gas turbine
application. The flow field was obtained by 3D numerical simulation with the
help of commercial CFD code. The analysis coupled both flow structure and
quantitative entropy generated from the inlet through to outlet. The comparison
have been made between 10, 20, 30, 40 and 50 exit beta angle. The simulation
result showed that at the streamwise location of 0.1-0.6, entropy generated
around 60 J/kgK per streamwise location length for all exit beta angles, where
the inflow direction was parallel with the inlet impeller passage. In contrast to
the location of 0.6-1.0, the entropy generated around 480 J/kgK per streamwise
location length, around 8 times of the entropy generated in location 0.1-0.6.
This was correspondent to high deformation rate of the flow field in this area.
The separation and secondary flow can be observed as a result of blade tip
flow leakage. Moreover, strong flow distortion with massive turbulent intensity
took place, and as a consequence, high local eddy viscosity was present.
Increasing the beta angle had alleviated jet-wake shear layer at the exit area of
the compressor as a consequence of less entropy generation in the location of
0.6-1.0 streamwise location.
Keywords: Entropy, flow passage, centrifugal compressor
1. INTRODUCTION
Micro-turbine engine (MT) become more practical means of power generation. This application has several
advantages compared to reciprocating engine. MT has more capability to operate with wider range of low calorific
fuel. Because MT has continuous combustion process. MT has better combustion efficiency and less emission.
Although, MT compression ratio is limited by turbine inlet temperature (TIT) and components efficiency.
Nowadays, compressor efficiency and turbine inlet temperature are increasing with the help of CFD and material
technology. Consequently, higher compression ratio and thermal efficiency of small gas turbine engine can be
realized. Moreover, exhaust temperature of gas turbine is generally higher than reciprocating engine, which
increase the potential to recover waste heat from the exhaust gas.
* Corresponding author: J. Charoensuk
E-mail address: kcjarruw@kmitl.ac.th
/ Volume 3(2), 2015 Transactions of the TSME: JRAME42

Thermal efficiency of small gas turbine depends on the operation of three engine’s sub-components which
comprised of 1.) Compressor 2.) Combustion chamber 3.) Turbine. In order to achieve high system thermal
efficiency, each component have to operate with high efficiency at their operating condition. The operating
condition of individual sub-component can be obtained from thermodynamics analysis. Thermodynamics analysis
is working on the working fluid state at the inlet and outlet of the engine sub-component in order to analyst on the
system operation performance. Thermodynamics analysis is the engineering tool which is able to carry out sub-
component’s desired working fluid state, during initial design phase. It is able to analyst on coupled interaction of
sub-component efficiency and the restriction of firing temperature of turbine. It shows that a small gas turbine
engine, with compressor has an efficiency of 78% (including combustor pressure drop), turbine firing temperature
at 1200K, is able to achieve thermal efficiency at around 28% (Fig 1) which comparable to piston engine. The
recuperative cycle micro-gas turbine application is considered as high potential for commercial application. MT
engine is able to integrate in co-generation or tri-generation system with interesting profitability potential in
Thailand. Many available Micro-turbine manufacturers select recuperative cycle for their designed engine. Most of
them provide engine’s thermal efficiency around 30% with optimum pressure ratio around 4. At this low pressure
ratio condition, the compressor is possible to employ single stage centrifugal compressor which is considered as a
benefit for low manufacturing capital cost of the MT. Moreover, MT integrate with synthesis gas from gasifier
(BIG/GT) reactor, is currently become the most interesting research topic due to their advantage compare with
reciprocating engine. MT is capable to consume hot synthesis gas from the gasifier reactor at around 500 °C. At this
temperature, tar is in vapor stage. So it could directly burn in turbine combustor. In contrast to reciprocating engine
which synthesis gas is necessary to cool down to room temperature while all tar need to remove before deliver syn-
gas to engine. Therefore, Micro-turbine is considered to deliver higher system thermal efficiency with gasification
reactor integration.
Fig. 1. Thermal efficiency of small gas turbine.
Fig. 2. Compressor Energy analysis.
Entropy generation is the quantitative parameter. It indicates the reversibility of process where energy is
transformed from useful pressure and kinetics form to useless thermal internal energy (Fig. 2). The transformed
energy is dissipated in working fluid. Thermodynamics discipline is categorized dissipation process as the internal
irreversibility. Increasing in dissipation relates to increasing in input energy required to obtain target exit pressure
and flow rate. Local entropy generation depends on local internal flow structure. Entropy generation in
compressor and turbine has significant impact on system thermal efficiency [1]. It is generally cast by isentropic
efficiency. In order to design compressor that capable to operate with high isentropic efficiency at design condition,

physical flow field that occur in the flow passage have to be well understood. Flow field characteristic is a result of
various interaction. Despite two third of entropy was generated in the diffuser [2], it has been studied that diffuser
efficiency was largely dependent on impeller exit flow condition. The interaction at the transition area from
impeller to diffuser inlet significantly contribute to the overall a compressor performance [3]. However, flow in
diffuser is beyond the scope of discussion in this paper.
2. GOVERNING EQUATIONS
3-D Reynolds averaged compressible Navier–Stokes equations was used for governing equations. The transport of
momentum, mass, energy are taken into account. Eddy viscosity was computed by k-ε turbulent modeling. Mass
and momentum transport equation can be written as below
( ) 0i
i
u
x
ρ
∂
× =
∂
(1)
( ) 2 2
3 3
j ji l i l
i j i j t ij
j i j j i l j j i l
u uu u u uP
u u k
x x x x x x x x x x
ρ µ δ µ ρ µ δ
     ∂ ∂  ∂ ∂ ∂ ∂∂ ∂ ∂ ∂
× = − + + − + + − +     ÷÷  ÷  ÷∂ ∂ ∂ ∂ ∂ ∂ ∂ ∂ ∂ ∂          
(2)
The local entropy generation depends on mean flow velocity gradient and level of turbulent intensity which can be
computed via the equation below.
TKE transport and TKE dissipation transport can be computed using the transport equation below
( ) ( ) t
i k
i j k j
k
k ku P
t x x x
µ
ρ ρ µ ρε
σ
  ∂ ∂ ∂ ∂
+ = + + −  ÷
∂ ∂ ∂ ∂   
(3)
( ) ( )
2
1 2
t
i k
i j j
u C G C
t x x x k k
ε ε
ε
µ ε ε ε
ρε ρε µ ρ
σ
  ∂ ∂ ∂ ∂
+ = + + −  ÷
∂ ∂ ∂ ∂   
(4)
When kσ and εσ are the turbulent Prandtl numbers assigned to be constants equal to 1.0 and 1.3 respectively, kp
**(5)
is the local turbulent generation rate which is proportional to local mean flow velocity gradient, can be
computed by the equation below
2
3
3
j ji l
k t t ij
j i l i
u uu u
P k
x x x x
µ ρ µ δ
  ∂ ∂ ∂ ∂
= + − +  ÷  ÷ ÷∂ ∂ ∂ ∂    
(5)
Turbulent viscosity ( tµ ) can be calculated using assumption for k-ε turbulent modeling which depends on local
values of TKE and TKE dissipation rate as below
2
t
k
Cµµ ρ
ε
= (5)
Where ε1C , ε 2C and Cµ are assigned to be 1.44, 1.92 and 0.09 respectively.
3. BOUNDARY CONDITIONS AND NUMERICAL METHODOLOGY
Preliminary compressor geometry was obtained corresponding to the designed thermodynamics cycle for 200 kW
rated power gas turbines application (Table 1, [16]). The preliminary design procedure conducted on
thermodynamics analysis that detailed flow was not considered and attention was given on state of working fluid at
Transactions of the TSME: JRAME 2015, Volume 3(2)/ 44

the inlet and outlet of the engine sub-component [4, 5]. After that, CFD was implemented for flow analysis.
Detailed geometry and fine tuning of the flow passage were made in order to achieve high isentropic efficiency and
desired operating condition.
Table 1: Thermodynamics data.
Isentropic Efficiency of Compressor 85
Pressure Ratio 4
Mass flow rate (kg/s) 1.33
Table 2: Geometry dimension from [9-11].
Dimension Size
Inlet impeller diameter (mm) 90
Outlet impeller diameter (mm) 440
Inlet impeller height (mm) 15
Outlet impeller height (mm) 10
Number of impeller / splitter blade 9/9
Impeller speed (rpm) 20,000
Tip clearance (mm) [8] 1
3.1 Numerical methodology
The computational study has been conducted by commercial ANSYS CFX Code [15]. The flow had been modeled
with RANS scheme. k-ε Turbulent model was employed to predict transport of turbulent property. Governing
equation was discretized by Finite volume method. Flow domain was divided into 630,000 cells of structured
hexahedral cell [6, 8, 14]. Grid independent was tested by compressor model of previous research [6, 7]. High grid
density was employed in near wall region to capture high gradient of flow property in that area. The normalized
residual convergence criteria of every transport equation was set at 1.0e-04 [6, 4, 14]. Model validation was
perform and good agreement was achieved [6]. However, it need to be realized, as generally found from the
previous research in this area [13], that the difference between steady and unsteady simulation was expected and the
unsteady simulation gave more accurate result than the steady one. It has revealed that the steady simulation
(RANS) tends to predict less entropy generation than the experiment and unsteady (URANS) simulation [12-14].
However, this has to be trade off with computer resource and modeling complexity.
3.2 Boundary conditions and validations
- Inlet boundary condition; static frame, total pressure was assigned to 1 atm
- Outlet boundary condition; average static pressure was assigned to 4 atm
- Stationary Wall and no-slip condition was assigned to wall
- Inlet temperature was 30°C
The flow was modeled in steady flow regime. Air was treated as compressible.
Fig. 3. Compressor geometry.

Fig. 4. Structure of hexahedral mesh.
Fig. 5. Meridional plane of the computational domain.
Model validation had been done by comparing the simulation result from Rigi test rig at Turbomachinery
Laboratory. ETH Zurich [6, 7] where the impeller model A8C41 (Albert Kammerer) was employed. Good
agreement between testing result and simulation result was obtained.
Fig. 6. Static entropy at different stream-wise location for all exit beta angle.
4. RESULTS AND DISCUSSION
4.1 Quantitative entropy generation
The simulation result showed the same character of entropy generation for all different Beta (β2) ตตตตตตต
ตตตตตตตตตต angle. The entropy generation was linearly increasing to around 60 J/kg-K at the stream-

wise location range of 0-0.6, as shown in Fig. 6. The total entropy generation at the stream-wise location range 0.6-
1.0 was around 480 J/kg-K, which was 8 times more than that in the impeller inlet region. The highest total entropy
generation was observed at the impeller exit region for the case of 10 degree exit beta angle. The lowest total
entropy generation was revealed for the case of 50 degree exit beta angle. Nevertheless, back-flow occurred from
recirculation at the exit computational domain, resulting in a decrease in entropy near the exit boundary cell faces.
This was observed at 0.9 stream-wise location toward to the exit of computation domain (Fig. 6).
4.2 The influence of exit beta angle on secondary flow region
At stream-wise location range of 0-0.6, the flow was parallel to the flow passage (Figs. 11 and 12), because in-flow
angle was coincide with the inlet impeller angle. The flow channel cross area was throated causing the fluid to
accelerate along the flow passage (Fig. 7). It is easy to manage the flow in throat region since the flow momentum
become higher along the passage. Flow separation is unlikely to occur in throat region. However, this was a trade-
off with the flow blockage which limit the delivered mass flow rate of the compressor. Despite the separation was
not observed in the impeller inlet area, the secondary flow was formed due to blade tip leakage that interacted with
flow in the passage (Fig 8). Blade tip leakage was the cross-jet flow that induced secondary vortices within the flow
passage.
It was difficult to manipulate the flow at the outlet area (0.6-1.0 stream-wise location). Since the flow was
decelerated rapidly due to wider area passage, so called “diffuser”. Various physical interaction influenced flow
field in this area. In general, Jet-wake region was formed (Figs. 13 and 14). Low pressure and low velocity zone in
the suction side of the impeller blade was the cause of flow separation. This region tend to decelerated on some
portion of the flow to separate from jet region. At the impeller outlet, jet-wake flow structure interacted with flow
downstream resulting in recirculation flow at the impeller outlet (Figs. 13 and 14). Stronger secondary flow from
blade tip leakage interaction, compared to the inlet region, could be observed.
Fig. 7. Mach number at various stream wise location for all exit beta angle.
Fig. 8. Secondary flow at stream-wise location of 0.25.
Figs. 9 and 10, showed entropy generation along stream-wise location of the impeller blade. At the location range
of 0.1-0.6, entropy generation was very little, resulting from non-deformation flow field at the inlet area as stated

before. The major source of entropy generation came from near-wall region, which shear layer generated from both
secondary flow from blade leakage (Figs. 9 and 10) and boundary layer structure. At the location range of 0.5-0.7,
more entropy was generated in near-wall region due to the development of boundary layer and blade tip leakage
interaction. However, entropy generation was in the same level as the location 0-0.4 because of the flow in this
location was diffused. It was around 400% lower in velocity magnitude compared to the location 0-0.4 (Fig. 7). At
the location of 0.7-1.0, entropy generation was 3-4 times higher than of the location 0-0.6, because of the
interaction of many physical phenomena that leading to high deformation flow field. The separation flow was
formed in the region around 0.6 stream-wise location (Figs. 11 and 12), which was a beginning of the jet-wake flow
structure. The accumulated secondary flow along blade tip and accumulated turbulence intensity that generated
from local high shear flow structure was transported by convection mode from upstream flow to exit region of 0.7-
1.0 location of the impeller.
Fig. 9. Entropy distribution of case 10 degree exit beta angle.
Fig. 10. Entropy distribution of case 50 degree exit beta angle.
Fig. 11. Velocity structure of 10 degree exit beta angle.

Fig. 12. Velocity structure of 50 degree exit beta angle.
Fig. 13. Jet-wake and recirculation structure of 10 degree exit beta angle.
Fig. 14. Jet-wake and recirculation structure of 50 degree exit beta angle.
Beta angle has a significant impact on flow structure. By increasing it the Jet-wake shear layer structure could be
alleviated. The flow was accelerated in a low-momentum wake region by increasing beta angle as a result of
smaller flow cross sectional area (Fig. 17 ไไไไไไไไไไไ 17).

Fig. 15. Pressure distribution of 10 degree exit beta angle.
Fig. 16. Pressure distribution of 50 degree exit beta angle.
Fig. 16. Effect of beta angle that cause smaller area in wake region and wider area in jet region.
5. CONCLUSION
A computational investigation has been undertaken to investigate on the characteristic of entropy generation along
blade passage of the centrifugal compressors. The impeller exit beta angle influenced on the entropy generation and
flow field, is depicted. Four different exit beta angle values provided same entropy generation characteristic. At
stream-wise location of 0.1-0.6, the entropy generation is almost the same for all four cases, which is very little
comparing to impeller exit region. At the impeller exit region (0.6-1.0 stream-wise), massive entropy generation

was formed. Because there are several physical interaction in this area. The exit beta angle has influenced on the
impeller outlet flow pattern resulting in less entropy generated, compared to inlet region. The lowest entropy
generation was observed in the case of beta angle 50 degree. The highest entropy generation occur in the case of
beta angle 10 degree. Flow field velocity showed that exit beta angle effect on the flow structure in exit area. Beta
angle alleviated on the jet-wake shear layer and also impeller exit flow distortion. The separation flow was earlier
found in the case of low Beta angle. Higher deformation rate of flow structure can be observed in the case of low
beta angle.
REFERENCES
[1] Soares, C. Microturbines Application for Distributed Energy System, London, Elsevier, 2007.
[2] Anish, S. and Sitaram, N. Computational investigation of impeller-diffuser interaction in a centrifugal compressor
with different types of diffusers, Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power
and Energy, Vol. 223, 2009, pp. 167-178
[3] Liu, R. and Xu, Z. Numerical investigation of a high-speed centrifugal compressor with hub vane diffusers,
Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power and Energy, Vol. 218(3), 2004,
pp. 155-169, 2004, pp. xx-xx.
[4] Wilson, D.G. The Design of High-Efficiency Turbomachinery and Gas Turbines, Cambridge, MA: The MIT
Press, 1984.
[5] Asuaje, M., Bakir, F., Smaine, K. and Rey, R. Inverse design method for centrifugal impellers and comparison
with numerical simulation tools, International Journal of Computational Fluid Dynamics, Vol. 18(2), 2004, pp.
101-110.
[6] Armin, Z., Albert, K. and Abhari, R.S. Unsteady Computational Fluid Dynamics investigation on inlet distortion
in a centrifugal compressor. Journal of Turbomachinery, Vol. 132, 2010, pp. xx-xx.
[7] Schleer, M. and Abhari, R.S. Clearance effects on the evolution of the flow in the vaneless diffuser of a
centrifugal compressor at part load condition, Journal of Turbomachinery, Vol. 130, 2008, pp. xx-xx.
[8] Tang, J., Turunen-Saaresti, T. and Lorjola, J. Use of partially shround impeller in a small centrifugal compressor.
Journal of Thermal Science, Vol. 17, 2008, pp. 21-27.
[9] Baskharone, E.A. Principle of Turbomachinery in Air-Breathing Engines, Cambridge University, New York,
Cambridge University Press, 2006.
[10] Dixon, S.L. (2005). Fluid mechanics and thermodynanics of turbomachinery, Liverpool, Butterworth-Heinemann.
[11] Florin, L., Trevino, J. and Sommer, S. Numerical Analysis of Blade Geometry Generation Techniques for
Centrifugal Compressors. International Journal of Rotating Machinery, 2007.
[12] Trébinjac, I., Kulisa, P., Bulot, N. and Rochuon, N. Effect of Unsteadiness on the Performance of a Transonic
Centrifugal Compressor Stage. Journal of Turbomachinery, Vol. 131, 2009, pp. xx-xx.
[13] Dickmann, H.P., Wimmel, T.S., Szwedowicz, J., Filsinger, D., and Roduner, C.H. Unsteady flow in a
turbocharger centrifugal compressor: Three-dimensional computational fluid dynamics simulation and numerical
and experimental analysis of impeller blade vibration, Journal of Turbomachinery, Vol. 128, 2006, pp. 455-465.
[14] Pierandreiand, G. and Sciubba, E. Numerical Simulation and Entropy Generation Maps of an Ultra-Micro-
Turbogas Compressor Rotor. ECOS 2010, 14-17 June 2010, Lausanne CH, 2010.
[15] ANSYS CFX-Solver Theory Guide. ANSYS CFX Release 12.1, ANSYS Europe Ltd., 1996-2006.

JRAME Vol 3 2015 No 2 Part 2 PP 2 (3)

More Related Content

What's hot (20)

Similar to JRAME Vol 3 2015 No 2 Part 2 PP 2 (3) (20)

JRAME Vol 3 2015 No 2 Part 2 PP 2 (3)