See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/301312306

Computational Fluid Dynamics in Turbomachinery: A Review of State of the

Art

Article  in  Archives of Computational Methods in Engineering · April 2016

DOI: 10.1007/s11831-016-9175-2

CITATIONS READS

36 3,741

5 authors, including:

Runa Nivea Pinto Asif Afzal

2 PUBLICATIONS   42 CITATIONS   
P. A. College of Engineering, Mangalore
46 PUBLICATIONS   220 CITATIONS   
SEE PROFILE
SEE PROFILE

Zahid Ansari
P.A. College of Engineering
29 PUBLICATIONS   188 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A numerical parametric study of mechanical behavior of dry contacts slipping on the disc-pads interface View project

Turbomachinery View project

All content following this page was uploaded by Asif Afzal on 13 March 2017.

The user has requested enhancement of the downloaded file.

Noname manuscript No.
(will be inserted by the editor)

Computational Fluid Dynamics in Turbomachinery: A Review

of State of the Art
Runa Nivea Pinto · Asif Afzal · Loyan Vinson D’Souza · Zahid Ansari ·
Mohammed Samee A. D

Received: date / Accepted: date

Abstract Computational Fluid Dynamics (CFD) plays to CFD investigation in turbomachinery and CFD code
an essential role to analyze fluid flows and heat trans- parallelization are also described.
fer situations by using numerical methods. Turboma-
Keywords CFD · Turbomachinery · CFD in tur-
chines involve internal and external fluid flow problems
bomachinery · Parallelization of CFD software ·
in compressors and turbines. CFD at present is one of
Computational time issues
the most important tools to design and analyze all types
of turbomachinery. The main purpose of this paper is
to review the state of the art work carried out in the 1 Introduction
field of turbomachinery using CFD. Literature review of
research work pertaining to CFD analysis in turbines, Computational fluid dynamics is a discipline of fluid
compressors and centrifugal pumps are described. Vari- mechanics which utilizes algorithm and numerical anal-
ous issues of CFD codes used in turbomachinery and its ysis to analyze and solve the given problems which in-
parallelization strategy adopted are highlighted. Fur- volve fluid flows. Computers are utilized to execute the
thermore, the prevailing merits and demerits of CFD calculations which are required to simulate the interac-
in turbomachinery are provided. Open areas pertinent tion of gases and liquid with surfaces defined by bound-
ary conditions [96]. Ongoing research provides software
that enhances the speed and accuracy of complex sim-
Runa Nivea Pinto ulation scenarios i.e. in turbulent or transonic flows
Department of Mechanical Engineering, P. A. College of En- [84]. CFD is employed in wide spread areas of engineer-
gineering, Mangaluru, India ing like aerospace, architecture, automotive, biomed-
ical, electronics and computers, climate calculations,
Asif Afzal oceanic flows, cardiovascular flows (heart, major ves-
Department of Mechanical Engineering, P. A. College of En-
gineering, Mangaluru, India
sels) etc [71, 31, 95, 37, 36].
Turbomachinery describes machines that transfer
Loyan Vinson D’Souza
energy between a fluid and a rotor, which includes both
Department of Mechanical Engineering, P. A. College of En- compressors and turbines. Energy is conveyed from a
gineering, Mangaluru, India fluid to a rotor in turbine, whereas a compressor con-
veys energy from a rotor to a fluid [26, 71]. CFD plays
Zahid Ansari a major role in the aerodynamic design of turboma-
Department of Computer Science Engineering, P. A. College chinery than in other engineering application [65]. Over
of Engineering, Mangaluru, India
Tel.: +91-7899267361
the years, design of modern compressor or turbine is
E-mail: zahid [email protected] unimaginable without the aid of CFD and this reliance
Mohammed Samee A. D
has magnified as the flow becomes susceptible to nu-
Department of Mechanical Engineering, P. A. College of En- merical prediction [73]. The advantage of CFD ranges
gineering, Mangaluru, India from reduced design cycles to enhanced performance,
reduced weight and cost in turbomachinery [71, 64].
2 Runa et al

In this article, we have performed a comprehensive of turbo tools, formulation of stage interfaces numeri-
survey of literature on the state of the art work carried cally, software layout in OpenFOAM and their position
out in investigation of flows in turbomachinery using in complete code design. For the simulations that are
CFD software. We provide the major issues associated transient with topological changes, a sliding mesh tech-
with CFD software used in turbomachinery related to nique could be utilized to handle the interface. Inter-
its time consumption. Parallelization attempted for the face handling techniques which allow (but are not con-
CFD software used in turbomachinery is provided along fined to) steady-state are precise to turbomachinery
with the parallel computing tool adopted to achieve CFD i.e. frozen rotor technique, General Grid Interface
high performance computing. The remainder of the ar- (GGI), cyclic GGI interface and partial overlap inter-
ticle is organized as follows. In section 2, the literature face [60].
review of numerical methods adopted and analysis car- R. Aghaei tog et al., have demonstrated the more
ried out using CFD in turbines, compressors and pumps advanced turbulence model in the analysis of radial tur-
are described. Section 3 provides the comprehensive bine using FLUENT and suggested a better way to
list of computational time related issues of CFD soft- select turbulence parameters which employs this soft-
ware pertinent to turbomachinery and parallelization ware as a complementary design tool for the turbo-
achieved for this CFD software. In section 4 and sec- machinery components of high-speed. Three turbulence
tion 5, the various advantages and disadvantages of models renormalisation-group (RNG), RSM i.e. Reynolds
employing CFD in turbomachinery are mentioned. We stress model and k-ε model and the results of these
conclude with our final remarks and provide few very models were juxtaposed with the 1D design and ex-
important future directions in the field of turbomachin- perimental results [4]. The main objective of Thomas
ery in section 6. Biesinger et al., was to contribute in understanding
of the unsteady flow phenomena that could lead to
the future generation turbine blading design. They uti-
2 Literature Review of CFD Applied to
lized the shape correction or phase shift and the time-
Turbomachinery
inclining methods in their investigation [12]. J D Den-
ton and W N Dawes have given a report of the main
In this section, literature review related to investigation
CFD method that are in use, discusses their pros and
carried out using CFD in the field of turbomachinery
cons. Through flow calculation is the backbone of turbo-
is provided. CFD analysis pertinent to turbines, com-
machinery and blade-to-blade calculations is the ba-
pressors and pumps are selected, as major research is
sis for describing the blade shape in detail and the
related with them [4]-[41]. Various CFD software em-
other methods three-dimensional single blade row cal-
ployed in turbomachinery analyis, issues, mesh size and
culations, multistage 3d calculations [35].
test cases selcted are mentioned in various tables of this
section. Shahrokh Shahpar and Leigh Lapworth integrated
the analysis codes along with design tools making pro-
vision to implement automatic optimization. A para-
2.1 CFD analysis in turbines metric definition of the geometry is required in auto-
matic optimization strategies and relocation of geom-
CFD analysis of flow over turbines is generally car- etry from CAD or a database, i.e. blade definition file
ried out using commercial software like FLUENT and to be continuous without user intervention. PADRAM
ANSYS-CFX. The various parameters considered, test utilizes both elliptic grid generators and transfinite in-
region of turbomachinery, number of grids used and the terpolation for generating hybrid C-O-H meshes [91].
issues related to CFD software is presented below and Nicholas J. Hills et al., have considered and advanced
is mentioned in table 1. in modeling and understanding of main annulus gas in-
Kai Becker et al., have given an overview of the re- gestion through turbine rim. They presented unsteady
cent breakthrough in the advancement of the hybrid three-dimensional CFD calculations and results from
CFD solver TRACE regarding the transition model- a elementary model and compared with experimental
ing of a Turbine T160 and T106C turbine cascade. data formerly published by Hills et al. Particular at-
A correlation-based transition models making use of tention is concentrated on the flow interaction that are
local variables has been combined in both structure complex in the rim seal region at which the sealing
and unstructured solver [11]. Hrvoje Jasak and Mar- and mainstream flows meet [57]. Wang et al., investi-
tin Beaudoin elucidate the performance of turbine in gated utilizing Computational Fluid Dynamics (CFD)
OpenFOAM, in the spirit of C++ and object orienta- a novel design of composite material axial water tur-
tion. Prominence is given to the primary functionality bine. On the basis of three dimensional numerical flow
Computational Fluid Dynamics in Turbomachinery: A Review 3

analysis, the flow characteristics through the nozzle, dif- 2.2 CFD analysis in centrifugal pump’s
fuser and wheel of the water turbine are predicted. The
exact torque and power for a specific flow speed were CFD analysis of flow over centrifugal pumps is generally
calculated and analyzed for a composite water turbine carried out using commercial software like FLUENT
at rotating speeds which are different [97]. Hellstrom and ANSYS-CFX. The various parameters considered,
et al.,scrutinized the parallel performance of commer- test region of turbomachinery, number of grids used and
cial CFD software on analogous computer networks, as the issues related to CFD software is presented below
recurrent solutions encountered in the industry. The and is mentioned in table 2.
efficiency enhancements procured in the previous ex- R Spence and J Amaral-Teixeira investigated struc-
periments by revamping the draft tube shape will be tured grid multi- block CFD software TASCflows time
considered, to extrapolate if the amelioration can be variation of pressure in an entire double volute, double
captured with the assistance of CFD. In the two draft entry centrifugal pump covering four parameters (geo-
tube geometries the three-dimensional flow field are re- metric) i.e vane arrangement, cutwater gap, side wall
solved using the commercial CFD code CFX-5.7.1, with clearance and snubber gap. This investigation of three
turbulent, incompressible, unsteady and steady flow as- flow rates was carried out by extracting the pulsations
sumption [56]. at fifteen various locations which cover important re-
gions of pump. To rank the relative significance of the
In order to enhance the capabilities in turbo-machinery four parameters (geometric) at each point for each flow
problems, R V Chima and M S Liou proposed two rate the tool used was Taguchi post-processing analy-
codes and these codes where modified using AUSM+ sis tools [93]. Hrvoje Jasak and Martin Beaudoin also
and H-CUSP schemes were used to analyze three turbo- elucidate the performance of centrifugal pump in Open-
machinery blades with the help of isentropic light pis- FOAM using the same method [60]. Richard B. Medvitz
ton compression tube and the results, oscillations of et all., developed a multiphase CFD method to ana-
total pressure are in good agreement with the AUSM+ lyze the performance of centrifugal pump under devel-
scheme and H-CUSP scheme requires less time to com- oped condition of cavitation. Fully-three-dimensional
pute [29]. Jingchun Wu and et al., attempted and in- and Quasi-three-dimensional analyses are carried out
vestigated to optimize the design of the runners with for two impeller configurations [76].
modern design techniques by using inviscid codes by Martin Beaudoin and Hrvoje Jasak have described
virtue of finite difference and finite element methods implementation of quick and robust algorithms for eval-
were added to the software at the primary stage and uating the GGI weighting factors. The current state
various turbulence models are employed for final op- of evolution of the Generalized Grid Interface (GGI)
timization which resulted in no cavitation on the suc- for simulations of turbo-machinery and steps towards
tion side and with improved efficiencies [101].Turbu- its parallelization and generalization is described here
lence treatment by Martin frank and et al., explored [10]. Suthep Kaewnai et al., objective was to utilize the
the potential of turbulence on turbine flow with a non technique of CFD in analyzing and predicting the per-
linear model to investigate four test cases and found formance of centrifugal pump having a radial-flow-type
that the predictions are better by capturing the damp- impeller. The first stage includes the generation of mesh
ing on convex curvature which in turn results in bet- and its refinement on domain of the fabricated impeller.
ter accuracy and non linear model is able to calcu- The next stage deals with the discovering of initial and
late the correct level of flow deceleration [45]. Sim- boundary conditions of the mesh-equipped module. In
mendinger et al., used CFD code TRACE for inves- the final stage, different results are computed and an-
tigation of physics of turbomachinery. The solver in- alyzed for factors affecting impeller performance [63].
teracts with hybrid conservative grid interfacing algo- To explore the performance variables using CFD tools
rithm which permits a mismatching boundary among by analyzing different test pumps which are end suc-
structure and unstructured grids [92]. In order to im- tion volute type was suggested by Erik dick in which,
prove the efficiency of turbo machines Bo chen and Xin first the impellers are generated then impeller channel
Yaun developed advance aerodynamic system to op- is meshed and analyzed with two techniques which are
timize blade design where three parametric modeling frozen rotor technique and mixing plane method and
modules are used and NURBS technique was utilized resulted that both techniques cannot be used with con-
to shape the blades, was integrated with the system fidence because of the erroneous behavior of the steady
to optimize design and found better performance and methods as inability to note fluid discharge [38]. Yang et
reasonably high efficiency when compared with the pre- al., investigated volute main parameter by implement-
vious design [24]. ing developed design utilizing finite volume method to
4 Runa et al

Table 1 CFD analysis in Turbine’s

Test Case or Software Reference

Parameter used Equations and Models used Grid size Issues of software
Test Region Used
3D RANS equations, Wilcox
Mach number (M = k- specific dissipation rate two-
Takes lot of time
0.6) and Reynolds equation, turbulence model, >0.1 Mil-
Turbine T160 TRACE to simulate the fluid [11]
number (Re = intermittency-momentum- lion(M)
flow situation.
70.000) thickness Reynolds number
model.
Huge computational
Francis turbine Francis turbine OpenFOAM RANS, turbulence models >2.2 M time is required to [60]
obtain the results.
As the points per
Last stage of a N-S equations, BaldwinLomax grid increases the
Mach number, flow,
steam turbine, —- turbulence mixing length model, 300000 CPU time required [35]
pressure ratio.
turbine blade k-ε model for the analysis also
increases.
Velocity, static pres- k-ε standard, Reynolds stress
sure, total pressure, model (RSM) and renormaliza- Simulation takes a
Radial turbine FLUENT —– [4]
Mach number, den- tion group (RNG), frozen rotor lot of time.
sity. model
Quasi-Three-
Dimensional Computational time
Pitch ratio, Mach ANSYS- URANS k-ω turbulence model,
blade row i.e. 480k required for simula- [12]
number. CFX N-S equations, k-ε models.
the T106D tion is more.
turbine
Large amount of
Mach number, Ve-
Turbine FLUENT — 1.37M time is wasted in [91]
locity, density
simulation.
Mainstream
Reynolds num-
ber, rotational
Time required to ob-
reynolds number, Turbulence model used is k-ε
Turbine rim seal FLUENT 1M tain converged re- [57]
coolant flow rate, model, RANS equations
sults is intolerable.
viscosity, mass flow
rate, seal gap, cavity
width
Composite ma- Results require a
Static pressure and
terials water FLUENT NS equation —- lot of computational [97]
velocity distribution
turbine time.
Computations take
Draft tube of Flow, pressure dis-
ANSYS- kε turbulence model and RANS a lot of time for un-
a ERCOFTAC tribution, computa- 1.4 M [56]
CFX equations steady CFD simula-
turbine tional time.
tion.
Computations re-
SST k-ω turbulence model and
2D and 3D tur- Reynolds number quired are huge
—- Baldwin lomax model, N-S equa- 869011 [29]
bine vanes and mach number leading to increased
tion
memory.
Obtaining simula-
Blades of the Flow characteristics STAR-CD, k-ε turbulence model and LES
1.52 M tion is a complicated [101]
turbine runner and parameters FLUENT model
computational task
For steady EARSM k-ω model.
Modeling co-
For unsteady RANS type eddy Complexity in pro-
Turbine cascade efficient, Reynolds TRACE 2.8 M [45]
viscosity model and k-ω trans- gramming
stresses
port equations
Pressure, velocities, Simulation results
k-ω turbulence model, Favre-
Turbine volume and temper- TRACE 19.11 M obtained are quite [92]
and RANS equations.
ature long computations
Subsonic blade Mach number, The time for op-
N-S equation. Low-Reynolds
and transonic pressure distribu- timization for
iSIGHT number q-ω two-equation turbu- 0.7 M [24]
blade of turbine tion, pressure load subsonic turbine is
lence model
cascade. distribution about 150 hours.

discretize transport equation and analysis is done for 2.3 CFD analysis in compressor’s
different volute parameter and found that higher ef-
ficiency couldn’t be obtained by design optimization
[104]. CFD analysis of flow in compressor’s is generally car-
ried out using commercial software like FLUENT and
ANSYS-CFX. The various parameters considered, test
region of turbomachinery, number of grids used and the
issues related to CFD software is presented below and
is mentioned in table 3.
Computational Fluid Dynamics in Turbomachinery: A Review 5

Table 2 CFD analysis in centrifugal pump’s

Test Case or Software Reference

Parameter used Equations and Models used Grid size Issues of software
Test Region Used
In centrifugal
pump, the cut- Continuous analysis
water clearance time of 45000 hours
Flow rates of
gaps, sidewall and consisting of
1.00Qn, 0.50Qn and TASCflow N-S equation and k-ω model >5 M [93]
leakages, blade analysis and result
0.25Qn
stagger or clock- data of more than
ing and snubber 550 gigabytes
clearance
ERCOFTAC Velocity field mag-
centrifugal nitude, flow condi- Simulation time is
OpenFOAM RANS, turbulence models >2.2 M [60]
pump with a tions, pressure dis- more
vaned diffuser tribution.
Backswept
Flow coefficient,
7-blade im- Time required and
cavitation numbers,
peller and three UNCLE-M RANS equation 55,195 maintenance cost for [76]
efficiency, head
dimensional the software is more.
coefficient
impeller
Velocity, turbulence Computational run
ERCOFTAC
kinetic energy, mesh OpenFOAM Sutherland-Hodgman algorithm —– time required is [10]
conical diffuser
resolution huge
Software running
Turbulence intensity k -ε, k -ω and RNG k -ε turbu-
Impeller CFX 50000 cost adds to total [63]
of 1% , 5% , 10% lence model.
value
Inlet tube , impeller,
The maintenance
Pumps of end- angular space, outer
cost of huge CFD
suction volute volute ,blade angle FLUENT k-ε model and RANS equations 550000 [38]
software developed
type ,head and rotational
is huge
speed
Throat area, cross
Centrifugal ANSYS- RNG k-ε turbulence model, 3D Huge computational
section shape,radial 0.91 M [104]
pump CFX Navier-Stoke equation. run time
gap, spiral area.

Gallimore et al., have illustrated the incorporation the effect of level of distortion on the performance of
of three dimensional design of blades in the core com- the compressor, the second one focuses on the effect
pressors of a Rolls-Royce Trent engine with special im- of distortion type and extent, the third studies the con-
portance on the use of dihedral and sweep in the de- trast arising from different frequencies of distortion [23].
signs of the rotor. They followed the idea of a univer- Wang et al objectives were to present a framework for
sity research project, from the testing of a multistage multi-objective optimization of rotor37 based on CFD
low-speed model to its implementation in high pres- simulations, approximation model, multi-objective op-
sure engine compressors. The authors combined three timization algorithm and by combining the multi-objective
dimensional CFD design methods and high-speed rig genetic algorithm i.e. back propagation neural network
testing. It concluded with the efficient incorporation (BPNN) and NSGA-II [98]. Dirk et al.,put forth an au-
of the Trent 500 engine with the 3D stator and rotor tomated, multi-disciplinary optimization procedure for
designs within their core compressors [47]. Shahrokh sub-sonic gas turbine compressor blades. A particular
and Leigh also analyzed compressors using the same focus is on a 3D- blade parametrization, which by de-
method [91]. Biesinger et al., also analyzed the com- fault leads to smooth blades, and on the assessment
pressor blades using the same method [12]. Denton and of the off-design behavior. To analyze the shape of the
Dawes also analyzed the compressor utilizing the same blade three steps are required i.e. (1) a blade/profile
method [35]. R. Aghaei tog et al., analyzed a centrifugal generator with an appropriate parameterization, (2) a
compressor using the same methodology [4]. mechanical integrity analysis and (3) a flow analysis
Nikolaos Charalambous et al., studied the effects of [16].
inlet flow distortion in a axial flow compressor attempt- Pierret presents the MAX optimization software evolved
ing to reduce the use of computational time and com- at CENAERO which is utilized to execute the optimiza-
puter resources. They aimed at developing a CFD sim- tion. The objective is to obtain the optimal geometry
ulation of a transonic fan of first stage under distorted for same rotational speed at three different operating
and clean conditions with a computer with a RAM conditions: one at choked mass flow, one at a near peak
memory of 1GB within suitable amount of time i.e. less efficiency mass flow and one near the stall flow. The
than 10 hours for the simulation of distorted flow. Three method implemented combines the use of a genetic al-
different studies are completed: the first one contrasts gorithm utilizing real coded variables very importantly
6 Runa et al

Table 3 CFD analysis in compressor

Test Case or Software Reference

Parameter used Equations and Models used Grid size Issues of software
Test Region Used
As the points per
N-S equations, BaldwinLo- grid increases the
Axial compres- Mach number, flow,
—- max turbulence mixing length 3M CPU time required [35]
sor pressure ratio.
model,kε model for the analysis also
increases.
Mass flow, change
Core compres- in tangential ve- Intense computa-
sors of Trent locity, temperature —- RANS equation >0.1 M tions leading to huge [47]
500 engine. through stator time consumption
shrouds, pressure
CPU time ratio
Three dimen-
Pitch ratio, Mach ANSYS- URANS k-ω turbulence model, is more because
sional compres- >0.4 M [12]
number CFX N-S equations, k-ε models. of slower overall
sor stage
convergence.
More need of com-
Mass flow, tangen-
Multistage com- putational resource
tial velocity, temper- FLUENT k-ε model and rans equations 1.37 M [91]
pressor to produce simula-
ature and pressure
tions
Axial compres-
Axial velocity at the
sor with three CFX- Heavy computa-
compressor inlet, RANS equation 0.2 M [23]
stages transonic TASCflow tional load
mass flow
fan
Choked mass flow
rate at 98% and
Huge simulation
NASA rotor37 100% with the outlet NUMECA RANS equation 0.6 M [98]
cost
pressure 114.2 kPa
and 90 kPa.
The compressor tangential velocity,
Even high end com-
blades of a sub- temperature, mach
—- RANS equation nearly 0.1 M puters cannot gener- [16]
sonic gas tur- number and pres-
ate results fast
bine sure
Analysis time by the
NASA rotor 67
Mass flow, perfor- software is increased
is used with 22 TRAF RANS equation 0.7 M [86]
mance, pressure. as number of nodes
blades
increase
NASA stage
Pressure ratio, tem-
35transonic Unsteady cases take
perature ratio,axial SWIFT RANS equation 0.5 M [28]
inlet stage for a a longer CPU time.
velocity.
core compressor
SST k-ω turbulence model and
3D supersonic Reynolds number High cost for com-
Baldwin lomax model, N-S equa- 0.9 M [30]
compressor and mach number putation’s
tion
Increased complex-
Modeling co-
RANS type eddy viscosity model ity of programming
Compressor efficient, Reynolds TRACE 3.8 M [45]
and k-ω transport equations the code and compu-
stresses
tational time
Massive time con-
Pressure, velocities, sumption by the
k-ω turbulence model, Favre-
Compressor volume and temper- TRACE 19.11 M CFD code, huge [92]
RANS equation
ature memory require-
ment
High computational
Pressure, tempera-
Spalart-Allmaras or k-ε model requirements of de-
NASA rotor37 ture, mass flow rate, elsA 2M [22]
and RANS equations. sign and prediction
mach number
of flows
Results of the tran-
Pressure ratio, tem- ANSYS-
Compressor SST turbulence model. 3.2 M sient simulations are [104]
perature ratio CFX
slower
Tip clearance
Axial compres- k-ω SST TNT model. 3D RANS High computational
height, mach num- —- 0.47 M [103]
sor equation cost
ber.
Momentum, pres-
sure ratio, vane
count, lean angle,
Difficulty in produc-
thickness, thickness Baldwin lomax model 3D RANS
Compressor ADPAC 0.5 M ing results in short [9]
distribution, radial equations
time
and meridional
extent, and camber
angle

accelerated utilizing an approximate method. High per- Linux cluster along with 170 3.06 GHz Xeon processors
formance computers with an infrastructure containing a are utilized [86]. Rodrick V. Chima illustrated a three-
Computational Fluid Dynamics in Turbomachinery: A Review 7

dimensional unsteady CFD model of compressor stabil- consuming and costly computational tasks [88, 25,
ity which utilizes a new technique for extracting infor- 41].
mation regarding compressor performance directly from – For accurate numerical prediction of turbulent, un-
steady CFD calculations executed separately. CSTALL steady and non uniform heat generation leads to
was run in a two-dimensional through flow mode for augmented simulation time [87].
estimation of stall pointsandquick calculations of oper- – Memory required by CFD codes for iterations of re-
ating maps. Chima and Liou used the same method to sults to reduce the truncation error is more [21].
analyze the compressor [28]. Martin Franke et al., an- – Programming for complex flow situations is a real
alyzed the compressor utilizing the same method [45]. test of computer skills [74, 39].
Christian Simmendinger and Edmund Kugeler analyzed – The maintenance cost of huge CFD programs de-
the compressor using the same methodology [92]. Castil- veloped for investigation in multistage turbines and
lon et al., analyzed the technical effects on numerical compressors is also high [99, 2].
simulations which are executed on turbo-machinery ap- – The recent trend in CFD is to shift to unsteady and
plications using chimera technique where the RANS multistage predictions, which need amplified com-
equation is solved for every grid and performed on over- puter power [19].
lapping boundaries and blanked mesh cells and the re- – Unsteady computations are yet too much time con-
sults of total to total pressure and isentropic efficiencies suming that is used for routine design of turboma-
are in good agreement with the experiment [22]. chinery.
To analyze and investigate different transient CFD – Utilizing RSM method, additional seven transport
methods that have been used by Rob Bluementhal and equations in 3D flow simulations of radial turbo-
et al., by utilizing four transient simulation methods machinery must be solved and thus it needs massive
to make comparison between steady and transient with setup time for computations [82] .
fluid modeled as ideal gas and explored that the results – In convergence of steady state solutions, reduction
on various factors are in better agreement with the com- in the magnitude of the order 4-5 is generally ob-
pared results [13]. Haixinet et al., tried to simulate the tained after nearly 5000 iterations. These calcula-
flow in a transonic compressor by implementing stall tions take nearly 12 hours CPU time on a collection
mechanism by self assigned CGCT technique where 7 of present multi-core architectures. With staunch
circumferential grooves were installed on the blade and use of 8 processors of the PC cluster utilized it is ap-
validation is done by varying tip clearance height [103]. proximated that the convergence calculations may
To investigate the margin of operability of the centrifu- take nearly 3 months if run uninterrupted [26].
gal compressor which is integrated with shroud using – A smooth Pareto front cannot be provided by a
CFD by Barton et al., by employing mixing plane ap- small size of population which could enhance the
proach to simulate the compressor and vane shroud and peril of less convergence in turbo-machinery. Even
also automatic grid formation. It concluded that time though evolution generations and small size of popu-
was reduced by automatic mesh generation and the re- lation are utilized, greater than thousand CFD com-
sults are in good agreement with the numerical tech- putation is still required to obtain the converged
nique [9]. Pareto front, that is too expensive computationally
[98, 8].
– The genetic algorithms employed in CFD in general
need CPU time that is not practical with the present
3 Major limitations and paralleization of CFD computing power for real industrial problems which
software applied in turbomachinery engage costly function evaluations [Pierret 2005].
– For unsteady CFD simulation the total execution
In present day, CFD software truly needs tremendous time even using multi-processors and multi-CPUs is
computations to obtain proper results. The CFD soft- nearly intolerable. [69].
ware is run for thousands to millions of iteration to get – To analyze and improve the design of turbines and
converged results and the numbers of iterations still compressors, engineers need the complete computa-
increase if more accuracy is demanded [67, 25]. In the tions over night which is difficult [72, 52].
following, we have listed the major limitations of CFD
software used in turbomachinery, which are most gen-
eral.

– Heavy computational demands of design and predic-

tion of flows in turbines leads to tremendously time
8 Runa et al

Table 4 Parallel computing tools adopted for parallelization of CFD code used for turbomachinery

CFD code used Parallel computing tool adopted for parallelization Related Refer-
for Turboma- ences
chinery
OpenMP MPI CUDA Any other
OpenMP+MPI
TRACE × × × and [92]
MPI+Pthreads
OpenMP+MPI
TAU × × × and [49]
MPI+Pthreads
√
OpenFOAM × × OpenMP+MPI [34, 33]
√
FLUENT × × × [85]
√
ANSYS × × × [70]
√
iSIGHT × × × [67]
√
NUMECA × × × [102, 67]
√
STAR-CD × × × [78]
√
FLOW3D × × × [80]
√
UNCLE-M × × × [50]
√ √
TRAF × OpenMP+MPI [7]
√
elsA × × × [48]
√
ADPAC × × PVM [51]
√
RVCQ3D × × × [94]
√
SWIFT × × × [27]
√ √
CFX × PVM [14]
TASCflow × × × × [75]
√
Turbomachinery × × × [1]
CFD
√
TURBO3D × × × [62]
√
TACOMA × × × [44]
√
SU2 × × × [83]

3.1 Parallelization of CFD software used in 4 Advantages of CFD in turbomachinery

turbomachinery
The various advantages of applying CFD in the field of
turbomachinery are mentioned in this section. CFD is
Parallelization is the computation of tasks of any soft-
of great benefit, especially for close analysis of complex
ware among multi-cores or multi-CPUs at the same
design and flows.
time to reduce the time [99, 6]. In CFD, parallelization
of various software are attempted used for turboma-
chinery design and analysis. General parallel computing – CFD analysis has the ability to predict correctly the
tools adopted for parallelization of CFD software used change in any direction of flow and is reliable [93,
in turbomachinery are: OpenMP, MPI and combination 5].
of these [21, 3]. CFD codes paralleized employing vari- – CFD has the potential to utilize the 3-dimensional
ous parallel computing tools is provided in the table 4 nature of the flow to restrict undesirable features
. like strong secondary flows in turbines or corner sep-
arations in compressors [58, 15] .
In table 4, the CFD software parallelized using a – Using CFD for studying distortion can save time
√
parallel computing tool are either marked with mark and cost, during the early stage of the design process
or its name is mentioned. Parallel computing tools with of the compressor[20, 23].
× mark shows that the corresponding CFD software is – The outcome from both the design of hydraulic tur-
not parallelized using that tool. Hence in future, adopt- bine draft tube CFD simulations i.e. steady and un-
ing the parallel computing tools with × mark, paral- steady unveil that full scalability is almost procured
lelization of the corresponding CFD software can be with commercial CFD software [32].
achieved, and its computational performance can be – Transition model can be used in any general CFD
evaluated. investigation without any advance provision for ge-
Computational Fluid Dynamics in Turbomachinery: A Review 9

ometry and grid topology. The model is flexible for sure. The approximations impact could be least to
engineering applications [40]. estimate the trends of global parameters variations,
– CFD is an extensively used method for the design but, will have a tough impact on the calculation of
and the analysis of gas turbines. The accuracy of local values of vital parameters like heat transfer
CFD is swiftly rising, thanks to the accessible com- and flow temperature [17].
putational resources that permit simulating high- – The CFD models and the real element in the tur-
speed flows employing hi-fidelity methodologies [42]. bine are not the same as all geometrical features
– To control unwanted features like strong secondary in the computational domain are not taken into ac-
flows in turbines or corner separations in compres- count like steps and fillets. Secondary Air Systems is
sors, CFD has the ability to predict the 3D nature most likely ignored in case of integrated simulation
of flow [43]. of compressor/ turbine/ combustor and only their
– Modern turbomachinery design relies almost com- effect will be incorporated using constant mass-flow
pletely on CFD to develop three-dimensional blade source terms or correlations. Even if considered, the
sections. Simple methods with empirical input are true dimension of gaps in service operations is not
still needed for the mean-line design and for through- known [53].
flow calculations [46]. – The accurate understanding of boundary conditions
– The design of modern turbomachinery depends en- is perhaps one of the most difficult tasks in CFD
tirely on CFD to develop 3D blade sections. Only which is very crucial in turbomachinery [68].
simple methods area required with empirical input – Grid dependence analysis is a fundamental task in
for the design and flow calculations [59]. CFD for numerical simulation. But grid spacing ef-
fects are responsible for the meager calculation of
both flow structures and integral parameters. There-
5 Issues of CFD in turbomachinery fore the effects of computational grid must be ac-
counted while performing hi-fidelity CFD [90].
In this section various disadvantages of applying CFD – One of the major challenging topics in CFD is the
in the field of turbomachinery are mentioned. One of simulation of a cooled, transonic high-pressure vane.
the major issue related to CFD in turbomachinery is to – In CFD, if 1D design is not proper, for example,
validate the results with experimental values. stage loading and blade diffusion factors then CFD
cannot turn out a good design [55].
– Three dimensional CFD calculations in compressor
– CFD cannot accurately estimate loss predictions and
blade are still not good enough to accurately calcu-
understanding of its results need great skill and ex-
late the total pressure loss [47, 77].
perience [100].
– A set of kind of stringent rules need to be formed for
– Interpretation of CFD results for unsteady flow is as
the CFD calculations so as to decrease the effect of
difficult as experimental results and need improved
grid in centrifugal pump and calculation parameters
post processing techniques like getting entropy gen-
on the results [54].
eration rates are required [58, 66].
– By CFD design optimization, the peak efficiency has
– The prediction of highly turbulent flows depends on
increased by 2.2% but still lower by 1% with respect
turbulent modeling in the CFD codes and gives lim-
to consumer requirements. The complexity is high
ited accuracy [79, 81].
because of the constraints of the existing turbine
runner [89, 101].
– Numerical errors are commonly obtained utilizing
6 Conclusion
CFD due to finite difference approximations [105].
– Errors are involved during modeling where the true In this article we performed the state of the art review
physics is not well-known and is as well complex to of CFD analysis for turbomachineries. The main con-
model, like turbulence modeling [61]. tributions of this article are listed below:
– Unknown geometry and boundary conditions, such
as leading edge shapes/ tip clearances and temper- – Provided a critical review of CFD analysis for tur-
ature profiles/ inlet pressure respectively [18]. bines, compressors and centrifugal pumps.
– CFD uses several approximations and models hence – Various issues related to the CFD software used in
errors derive from this process. During the discretiza- turbomachinery are identified.
tion of the NS equations and from turbulence mod- – Parallel computing tools adopted for parallelization
els truncation errors arise. Example: mesh resolu- of CFD software used in turbomachinery are ear-
tion, steady flow assumption and turbulence clo- marked.
10 Runa et al

– Issues associated with employing the CFD in the mentioned above. These tools have been proved to be
field of turbomachinery are spotlighted. of great computational performance enhancers for the
Form the literature survey it is found that, the fu- CFD software used in other fields. More specifically
ture of turbomachinery designs will depend even more OpenMP+MPI hybrid parallel computing approach has
extremely on CFD, than they do currently, as the abil- shown highly improved computational performance, which
ity of CFD to predict the behavior of fluid flow and can also be utilized in parallelization of turbomachinery
heat transfer is continuously improving. The conclu- CFD software.
sions drawn from the survey are as follows:
– At present, the trend is to move from single stage References
and steady flow to multistage and unsteady predic-
tions, both of which demand intense computational 1. http://www.cfdsupport.com/turbomachinery-cfd.html
power. accessed on march 5, 2016
2. Adamczyk, J.J.: Aerodynamic analysis of multistage
– The results obtained from unsteady flow can be em-
turbomachinery flows in support of aerodynamic design.
ployed to generate the unsteady blade loading which In: ASME 1999 International Gas Turbine and Aero-
indeed helps in prediction of mechanical issues of engine Congress and Exhibition, pp. V004T06A001–
blading. V004T06A001. American Society of Mechanical Engi-
neers (1999)
– The unsteady flow predicts the loss generation which
3. Afzal, A., Ansari, Z., Faizabadi, A.R., Ramis, M.: Par-
is currently in use and need to be explored further. allelization strategies for computational fluid dynamics
– With CFD, the prediction of generation of entropy software: State of the art review. Archives of Compu-
when vortices and wakes mix in unsteady flow has tational Methods in Engineering pp. 1–27 (2016). DOI
DOI 10.1007/s11831-016-9165-4
become definitely very easy than the experiment but
4. Aghaei tog, R., Tousi, A., Tourani, A.: Comparison
for validation of CFD results, high class experimen- of turbulence methods in cfd analysis of compressible
tal results are required and to interpret the results flows in radial turbomachines. Aircraft Engineering and
from CFD, great talent is needed. Aerospace Technology 80(6), 657–665 (2008)
– Numerical and modeling errors need to be reduced 5. Anagnostopoulos, J.S.: A fast numerical method for flow
analysis and blade design in centrifugal pump impellers.
further. Computers & Fluids 38(2), 284–289 (2009)
– The accuracy of CFD results obtained for turbu- 6. Ansari, Z., Afzal, A., Nayak, S., Muhiuddeen, M.: Liter-
lence flow need to be increased. ature survey for the comparative study of various high
– The assumptions and approximations made in CFD performance computing techniques. International Jour-
nal of Computer Trends and Technology (IJCTT) 27(2),
analysis of turbomachinery has to be developed for 80–86 (2015)
every flow and geometric conditions. 7. Arnone, M.G.M.M.A., Dominguez, A.: A hybrid paral-
lelization strategy of a cfd code for turbomachinery ap-
Another important aspect understood from this sur-
plications. 11th European Turbomachinery Conference,
vey is that the unsteady flow computations need huge March 23-26, Madrid, Spain (2015)
time to obtain results. This is due to the huge compu- 8. Asuaje, M., Bakir, F., Kouidri, S., Kenyery, F., Rey, R.:
tational time required by the CFD software to gener- Numerical modelization of the flow in centrifugal pump:
ate converged results. In order to reduce the compu- volute influence in velocity and pressure fields. Interna-
tional journal of rotating machinery 2005(3), 244–255
tational time of the codes are parallelized to run on (2005)
various multi-core/ multi-processor architectures. The 9. Barton, M., Mansour, M., Liu, J., Palmer, D.: Numeri-
major parallel computing tool popularly used for the cal optimization of a vaned shroud design for increased
parallelization of CFD codes in turbomachinery is MPI. operability margin in modern centrifugal compressors.
Journal of turbomachinery 128(4), 627–631 (2006)
Various parallel computing tools that can be employed 10. Beaudoin, M., Jasak, H.: Development of a general-
further to attempt parallelization of these codes are pro- ized grid interface for turbomachinery simulations with
vided below: openfoam. In: Open Source CFD International Confer-
ence, Berlin, Germany, vol. 468 (2008)
– OpenMP (Open Multiprocessing) 11. Becker, K., Heitkamp, K., Kügeler, E.: Recent progress
– CUDA (Compute Unified Device Architecture) in a hybrid-grid cfd solver for turbomachinery flows.
– PVM (Parallel Virtual Machine) In: Proceedings Fifth European Conference on Com-
putational Fluid Dynamics ECCOMAS CFD, vol. 2010
– Hybrid OpenMP+MPI
(2010)
– Tri-level hybrid OpenMP+MPI+CUDA 12. Biesinger, T., Cornelius, C., Rube, C., Braune, A., Cam-
– OpenCL (Open Computing Language) pregher, R., Godin, P.G., Schmid, G., Zori, L.: Unsteady
cfd methods in a commercial solver for turbomachinery
So far, very less effort have been reported in lit-
applications. In: ASME Turbo Expo 2010: Power for
erature regarding parallelization of turbomachinery re- Land, Sea, and Air, pp. 2441–2452. American Society of
lated CFD codes employing parallel computing tools Mechanical Engineers (2010)
Computational Fluid Dynamics in Turbomachinery: A Review 11

13. Blumenthal, R., Hutchinson, B., Zori, L.: Investigation Power for Land, Sea, and Air, pp. 1157–1168. American
of transient cfd methods applied to a transonic com- Society of Mechanical Engineers (2006)
pressor stage. In: ASME 2011 Turbo Expo: Turbine 29. Chima, R.V., Liou, M.S.: Comparison of the ausm+ and
Technical Conference and Exposition, pp. 1423–1430. h-cusp schemes for turbomachinery applications. AIAA
American Society of Mechanical Engineers (2011) Paper AIAA-2003-4120. Also NASA TM-2003-212457
14. Bochette, N.J.: Computational analysis of flow through (2003)
a transonic compressor rotor. Ph.D. thesis, Monterey, 30. Chima, R.V., Liou, M.S.: Comparison of the ausm+ and
California. Naval Postgraduate School (2005) h-cusp schemes for turbomachinery applications. AIAA
15. Bonaiuti, D., Zangeneh, M., Aartojarvi, R., Eriksson, J.: Paper AIAA-2003-4120. Also NASA TM-2003-212457
Parametric design of a waterjet pump by means of in- (2003)
verse design, cfd calculations and experimental analyses. 31. Chung, T.: Computational fluid dynamics. Cambridge
Journal of Fluids Engineering 132(3), 031,104 (2010) university press (2010)
16. Buche, D., Guidati, G., Stoll, P.: Automated design op- 32. Constantinescu, G., Patel, V.: Numerical model for sim-
timization of compressor blades for stationary, large- ulation of pump-intake flow and vortices. Journal of
scale turbomachinery. In: ASME Turbo Expo 2003, col- Hydraulic Engineering 124(2), 123–134 (1998)
located with the 2003 International Joint Power Gener- 33. Cosden, I.A., Lukes, J.R.: A hybrid atomistic–
ation Conference, pp. 1249–1257. American Society of continuum model for fluid flow using lammps and
Mechanical Engineers (2003) openfoam. Computer Physics Communications 184(8),
17. Bunker, R.S.: A review of turbine blade tip heat trans- 1958–1965 (2013)
fer. Annals of the New York Academy of Sciences 34. Dagnaa, P., Hertzerb, J.: Evaluation of multi-threaded
934(1), 64–79 (2001) openfoam hybridization for massively parallel archi-
18. Campanari, S., Iora, P.: Definition and sensitivity anal- tectures. Tech. rep., Tech. Rep.[Online]. Available:
ysis of a finite volume sofc model for a tubular cell geom- http://www. prace-ri. eu/IMG/pdf/wp98. pdf (2013)
etry. Journal of Power Sources 132(1), 113–126 (2004) 35. Denton, J., Dawes, W.: Computational fluid dynamics
19. Cao, S., Peng, G., Yu, Z.: Hydrodynamic design of ro- for turbomachinery design. Proceedings of the Institu-
todynamic pump impeller for multiphase pumping by tion of Mechanical Engineers, Part C: Journal of Me-
combined approach of inverse design and cfd analysis. chanical Engineering Science 213(2), 107–124 (1998)
Journal of fluids engineering 127(2), 330–338 (2005) 36. Deur, J., Jonnavithula, S., Meeks, E.: The combination
20. Caridad, J., Kenyery, F.: Cfd analysis of electric sub- of detailed kinetics and cfd in automotive applications.
mersible pumps (esp) handling two-phase mixtures. In: Eleventh International Engine Combustion Multi-
Journal of energy resources technology 126(2), 99–104 Dimensional Modeling Conference, Detroit (2001)
(2004) 37. Dhaubhadel, M.: Review: Cfd applications in the auto-
21. Casartelli, E., Mangani, L.: Object-oriented open- motive industry. Journal of Fluids Engineering 118(4),
source cfd for turbomachinery applications: A re- 647–653 (1996)
view and recent advances. In: ASME Turbo Expo 38. Dick, E., Vierendeels, J., Serbrugyns, A., Voorde, J.V.:
2013: Turbine Technical Conference and Exposition, Performance prediction of centrifugal pumps with cfd-
pp. V06BT37A036–V06BT37A036. American Society of tools. Task quarterly 5, 579–594 (2001)
Mechanical Engineers (2013) 39. Dickmann, H.P., Wimmel, T.S., Szwedowicz, J.,
22. Castillon, L., Péron, S., Benoit, C., Billonnet, G.: Nu- Filsinger, D., Roduner, C.H.: Unsteady flow in a tur-
merical simulations of technological effects encountered bocharger centrifugal compressor: Three-dimensional
on turbomachinery configurations with the chimera computational fluid dynamics simulation and numerical
technique. In: 27th International Congress of the Aero- and experimental analysis of impeller blade vibration.
nautical Sciences, Nice, France (2010) Journal of turbomachinery 128(3), 455–465 (2006)
23. Charalambous, N., Ghisu, T., Iurisci, G., Pachidis, V., 40. Ding, H., Visser, F., Jiang, Y., Furmanczyk, M.: Demon-
Pilidis, P.: Axial compressor response to inlet flow dis- stration and validation of a 3d cfd simulation tool pre-
tortions by a cfd analysis. In: ASME Turbo Expo 2004: dicting pump performance and cavitation for industrial
Power for Land, Sea, and Air, pp. 1637–1649. American applications. Journal of Fluids Engineering 133(1),
Society of Mechanical Engineers (2004) 011,101 (2011)
24. Chen, B., Yuan, X.: Advanced aerodynamic optimiza- 41. Ekici, K., Hall, K.C.: Nonlinear analysis of unsteady
tion system for turbomachinery. Journal of turboma- flows in multistage turbomachines using harmonic bal-
chinery 130(2), 021,005 (2008) ance. AIAA journal 45(5), 1047–1057 (2007)
25. Chen, J., Briley, W.: A parallel flow solver for unsteady 42. Fan, J., Eves, J., Thompson, H., Toropov, V., Kapur,
multiple blade row turbomachinery simulations. In: N., Copley, D., Mincher, A.: Computational fluid dy-
ASME Turbo Expo 2001: Power for Land, Sea, and Air, namic analysis and design optimization of jet pumps.
pp. V001T03A044–V001T03A044. American Society of Computers & Fluids 46(1), 212–217 (2011)
Mechanical Engineers (2001) 43. Fecarotta, O., Carravetta, A., Ramos, H.: Cfd and com-
26. Chew, J.W., Hills, N.J.: Computational fluid dynamics parisons for a pump as turbine: mesh reliability and per-
for turbomachinery internal air systems. Philosophical formance concerns. Int J Energy Environ 2(1), 39–48
Transactions of the Royal Society of London A: Mathe- (2011)
matical, Physical and Engineering Sciences 365(1859), 44. Felten, F.N., Kapetanovic, S., Holmes, D.G., Ostrowski,
2587–2611 (2007) M.: Gas turbine temperature prediction using unsteady
27. Chima, R.: Computational modeling of vortex genera- cfd and realistic non-uniform 2d combustor exit proper-
tors for turbomachinery. In: ASME Turbo Expo 2002: ties. In: ASME Turbo Expo 2008: Power for Land, Sea,
Power for Land, Sea, and Air, pp. 1229–1238. American and Air, pp. 1725–1734. American Society of Mechanical
Society of Mechanical Engineers (2002) Engineers (2008)
28. Chima, R.V.: A three-dimensional unsteady cfd model 45. Franke, M., Röber, T., Kügeler, E., Ashcroft, G.: Turbu-
of compressor stability. In: ASME Turbo Expo 2006: lence treatment in steady and unsteady turbomachinery
12 Runa et al

flows. In: V European Conference on Computational predictions. Chemical Engineering Research and Design
Fluid Dynamics, ECCOMAS CFD, pp. 14–17 (2010) 79(8), 887–894 (2001)
46. Fujun, W.: Application of cfd to turbulent flow analy- 62. Joubert, H., Quiniou, H.: Turbomachinery design used
sis and performance prediction in hydraulic machinery. intensive cfd. In: 22nd Congress of International Council
JOURNAL-CHINA AGRICULTURAL UNIVERSITY of the Aeronautical Sciences, vol. 28 (2000)
10(4), 75 (2005) 63. Kaewnai, S., Chamaoot, M., Wongwises, S.: Predicting
47. Gallimore, S.J., Bolger, J.J., Cumpsty, N.A., Taylor, performance of radial flow type impeller of centrifugal
M.J., Wright, P.I., Place, J.M.: The use of sweep and pump using cfd. Journal of mechanical science and tech-
dihedral in multistage axial flow compressor blading: nology 23(6), 1620–1627 (2009)
Part iuniversity research and methods development. In: 64. Keck, H., Sick, M.: Thirty years of numerical flow sim-
ASME Turbo Expo 2002: Power for Land, Sea, and Air, ulation in hydraulic turbomachines. Acta Mechanica
pp. 33–47. American Society of Mechanical Engineers 201(1-4), 211–229 (2008)
(2002) 65. Kielb, R.E.: Cfd for turbomachinery unsteady flowsan
48. Gazaix, M., Mazet, S., Montagnac, M.: Large scale mas- aeroelastic design perspective. AIAA paper 429 (2001)
sively parallel computations with the block-structured 66. Kirtley, K., Graziosi, P., Wood, P., Beacher, B., Shin,
elsa cfd software. In: Parallel Computational Fluid Dy- H.W.: Design and test of an ultralow solidity flow-
namics 2008, pp. 111–117. Springer (2010) controlled compressor stator. Journal of turbomachin-
49. Gerhold, T.: Overview of the hybrid rans code tau. In: ery 127(4), 689–698 (2005)
MEGAFLOW-Numerical Flow Simulation for Aircraft 67. Kwon, J., Ecer, A., Periaux, J., Satofuka, W., Fox, P.:
Design, pp. 81–92. Springer (2005) Aerodynamic optimization design system for turboma-
50. Gupta, S.: Performance evaluation and optimization of chinery based on parallelized 3d viscous numerical anal-
the unstructured cfd code uncle. University of Kentucky ysis. Parallel Computational Fluid Dynamics 2006: Par-
Master’s Theses (2006) allel Computing and its Applications p. 293 (2007)
51. Hall, E.J., Heidegger, N.J., Delaney, R.A.: Performance
68. Langtry, R.B., Menter, F.R.: Correlation-based tran-
prediction of endwall treated fan rotors with inflow dis-
sition modeling for unstructured parallelized compu-
tortion. AIAA Paper pp. 96–0244 (1996)
tational fluid dynamics codes. AIAA journal 47(12),
52. Hall, K.C., Thomas, J.P., Clark, W.S.: Computation of 2894–2906 (2009)
unsteady nonlinear flows in cascades using a harmonic
69. Larjola, J., Backman, J., Esa, H., Pitkänen, H., Salli-
balance technique. AIAA journal 40(5), 879–886 (2002)
nen, P., Honkatukia, J.: Centrifugal compressor design
53. Hamada, K., Smith, T., Durrani, N., Qin, N., How-
and testing in finnish high speed technology. In: ASME
ell, R.: Unsteady flow simulation and dynamic stall
Intl. mechanical engineering congress & expo., Novem-
around vertical axis wind turbine blades. In: 46th AIAA
ber, pp. 5–11 (2000)
Aerospaces Sciences Meeting and Exhibit, Reno, Nevada
70. Laursen, J., Enevoldsen, P., Hjort, S.: 3d cfd quantifica-
(2008)
tion of the performance of a multi-megawatt wind tur-
54. He, L., Chen, T., Wells, R., Li, Y., Ning, W.: Analysis of
bine. In: Journal of Physics: Conference Series, vol. 75,
rotor-rotor and stator-stator interferences in multi-stage
p. 012007. IOP Publishing (2007)
turbomachines. Journal of Turbomachinery 124(4),
564–571 (2002) 71. Lin, C.l., Tawhai, M.H., McLennan, G., Hoffman, E.A.:
Computational fluid dynamics. Engineering in Medicine
55. He, L., Sato, K.: Numerical solution of incompressible
and Biology Magazine, IEEE 28(3), 25–33 (2009)
unsteady flows in turbomachinery. Journal of fluids en-
gineering 123(3), 680–685 (2001) 72. Liu, Y., Claus, R.W., Litt, J.S., Guo, T.H.: Simulating
56. Hellström, J., Marjavaara, B., Lundström, T.: Parallel effects of high angle of attack on turbofan engine per-
cfd simulations of an original and redesigned hydraulic formance. In: 51st AIAA Aerospace Sciences Meeting
turbine draft tube. Advances in Engineering Software (2013)
38(5), 338–344 (2007) 73. Lucius, A., Brenner, G.: Unsteady cfd simulations of a
57. Hills, N.J., Chew, J.W., Turner, A.B.: Computational pump in part load conditions using scale-adaptive sim-
and mathematical modelling of turbine rim seal inges- ulation. International Journal of Heat and Fluid Flow
tion. In: ASME Turbo Expo 2001: Power for Land, 31(6), 1113–1118 (2010)
Sea, and Air, pp. V003T01A078–V003T01A078. Amer- 74. Luksch, P.: Parallel and distributed implementation of
ican Society of Mechanical Engineers (2001) large industrial applications. Future Generation Com-
58. Horlock, J., Denton, J.: A review of some early design puter Systems 16(6), 649–663 (2000)
practice using computational fluid dynamics and a cur- 75. Mai-Duy, N., Tran-Cong, T.: Numerical solution of
rent perspective. Journal of Turbomachinery 127(1), navier–stokes equations using multiquadric radial basis
5–13 (2005) function networks. International Journal for Numerical
59. Jafarzadeh, B., Hajari, A., Alishahi, M., Akbari, M.: Methods in Fluids 37(1), 65–86 (2001)
The flow simulation of a low-specific-speed high-speed 76. Medvitz, R.B., Kunz, R.F., Boger, D.A., Lindau, J.W.,
centrifugal pump. Applied Mathematical Modelling Yocum, A.M., Pauley, L.L.: Performance analysis of cav-
35(1), 242–249 (2011) itating flow in centrifugal pumps using multiphase cfd.
60. Jasak, H., Beaudoin, M.: Openfoam turbo tools: From Journal of Fluids Engineering 124(2), 377–383 (2002)
general purpose cfd to turbomachinery simulations. 77. Medvitz, R.B., Kunz, R.F., Boger, D.A., Lindau, J.W.,
In: ASME-JSME-KSME 2011 Joint Fluids Engineering Yocum, A.M., Pauley, L.L.: Performance analysis of cav-
Conference, pp. 1801–1812. American Society of Me- itating flow in centrifugal pumps using multiphase cfd.
chanical Engineers (2011) Journal of Fluids Engineering 124(2), 377–383 (2002)
61. Jaworski, Z., Dyster, K., Nienow, A.: The effect of size, 78. Meng, H., Wang, C.Y.: Large-scale simulation of poly-
location and pumping direction of pitched blade turbine mer electrolyte fuel cells by parallel computing. Chem-
impellers on flow patterns: Lda measurements and cfd ical Engineering Science 59(16), 3331–3343 (2004)
Computational Fluid Dynamics in Turbomachinery: A Review 13

79. Muggli, F.A., Holbein, P., Dupont, P.: Cfd calculation of 95. Stewart, S.F., Paterson, E.G., Burgreen, G.W., Hariha-
a mixed flow pump characteristic from shutoff to maxi- ran, P., Giarra, M., Reddy, V., Day, S.W., Manning,
mum flow. Journal of fluids engineering 124(3), 798–802 K.B., Deutsch, S., Berman, M.R., et al.: Assessment of
(2002) cfd performance in simulations of an idealized medical
80. Navarro, M.A., Santos, A.A.: Evaluation of a numeric device: results of fdas first computational interlabora-
procedure for flow simulation of a 5× 5 pwr rod bundle tory study. Cardiovascular Engineering and Technology
with a mixing vane spacer. Progress in Nuclear Energy 3(2), 139–160 (2012)
53(8), 1190–1196 (2011) 96. Um, S., Wang, C.Y., Chen, K.: Computational fluid
81. Nennemann, B., Vu, T., Farhat, M.: Cfd prediction of dynamics modeling of proton exchange membrane fuel
unsteady wicket gate-runner interaction in francis tur- cells. Journal of the Electrochemical society 147(12),
bines: A new standard hydraulic design procedure. In: 4485–4493 (2000)
HYDRO 2005, LMH-CONF-2006-006 (2005) 97. Wang, J.f., Piechna, J., Mueller, N.: A novel design of
82. Oyama, A., Liou, M.S., Obayashi, S.: Transonic axial- composite water turbine using cfd. Journal of Hydrody-
flow blade optimization: Evolutionary algorithms/three- namics, Ser. B 24(1), 11–16 (2012)
dimensional navier-stokes solver. Journal of Propulsion 98. Wang, X., Hirsch, C., Kang, S., Lacor, C.: Multi-
and Power 20(4), 612–619 (2004) objective optimization of turbomachinery using im-
83. Palacios, F., Colonno, M.R., Aranake, A.C., Campos, proved nsga-ii and approximation model. Computer
A., Copeland, S.R., Economon, T.D., Lonkar, A.K., Methods in Applied Mechanics and Engineering 200(9),
Lukaczyk, T.W., Taylor, T.W., Alonso, J.J.: Stan- 883–895 (2011)
ford university unstructured (su2): An open-source in- 99. Wang, Y.X., Zhang, L.L., Liu, W., Che, Y.G., Xu, C.F.,
tegrated computational environment for multi-physics Wang, Z.H., Zhuang, Y.: Efficient parallel implementa-
simulation and design. AIAA Paper 287, 2013 (2013) tion of large scale 3d structured grid cfd applications on
84. Pasaogullari, U., Wang, C.Y.: Computational fluid dy- the tianhe-1a supercomputer. Computers & Fluids 80,
namics modeling of proton exchange membrane fuel cells 244–250 (2013)
using fluent. In: 2002 Fluent Users Group Meeting, 100. Westra, R., Broersma, L., Van Andel, K., Kruyt, N.:
Manchester (2002) Piv measurements and cfd computations of secondary
85. Peigin, S., Epstein, B., Rubin, T., Seror, S.: Parallel flow in a centrifugal pump impeller. Journal of Fluids
large scale high accuracy navier-stokes computations on Engineering 132(6), 061,104 (2010)
distributed memory clusters. The Journal of Supercom- 101. Wu, J., Shimmei, K., Tani, K., Niikura, K., Sato, J.:
puting 27(1), 49–68 (2004) Cfd-based design optimization for hydro turbines. Jour-
86. Pierret, S.: Multi-objective and multi-disciplinary opti- nal of Fluids Engineering 129(2), 159–168 (2007)
mization of three-dimensional turbomachinery blades. 102. Xin, L.Y.Y.Y.: Blade stacking optimization based on ga
Proceedings of the 6th world congresses of structural and sqp methods [j]. Journal of Engineering Thermo-
and multidisciplinary optimization (2005) physics 1, 015 (2003)
87. Puigt, G., Gazaix, M., Montagnac, M., Le Pape, M.C., 103. Xudong, C.H.H., Song, F.: Cfd investigation on stall
de la Llave Plata, M., Marmignon, C., Boussuge, J.F., mechanisms and casing treatment of a transonic com-
Couaillier, V.: Development of a new hybrid compress- pressor (2006)
ible solver inside the cfd elsa software. AIAA Paper 104. Yang, S., Kong, F., Chen, B.: Research on pump vo-
3379, 2011 (2011) lute design method using cfd. International Journal of
88. Reboul, G., Polacsek, C.: Towards numerical simulation Rotating Machinery 2011 (2011)
of fan broadband noise aft radiation from aeroengines. 105. ZHANG, D.s., SHI, W.d., Bin, C., GUAN, X.f.: Un-
AIAA journal 48(9), 2038–2048 (2010) steady flow analysis and experimental investigation of
89. Ron-Ho, N.: A multiple-grid scheme for solving the euler axial-flow pump. Journal of Hydrodynamics, Ser. B
equations. AIAA journal (2012) 22(1), 35–43 (2010)
90. Sezer-Uzol, N., Long, L.N.: 3-d time-accurate cfd sim-
ulations of wind turbine rotor flow fields. AIAA paper
394, 2006 (2006)
91. Shahpar, S., Lapworth, L.: Padram: Parametric design
and rapid meshing system for turbomachinery optimi-
sation. In: ASME Turbo Expo 2003, collocated with the
2003 International Joint Power Generation Conference,
pp. 579–590. American Society of Mechanical Engineers
(2003)
92. Simmendinger, C., Kügeler, E.: Hybrid parallelization of
a turbomachinery cfd code: performance enhancements
on multicore architectures. In: Proceedings of the V
European conference on computational fluid dynamics
ECCOMAS CFD (2010)
93. Spence, R., Amaral-Teixeira, J.: A cfd parametric study
of geometrical variations on the pressure pulsations and
performance characteristics of a centrifugal pump. Com-
puters & Fluids 38(6), 1243–1257 (2009)
94. Stewart, M.E.: Towards a tool for rigorous, automated
code comprehension using symbolic execution and se-
mantic analysis. In: Software Engineering Workshop,
2005. 29th Annual IEEE/NASA, pp. 89–96. IEEE
(2005)

View publication stats