Development of a CFD simulation methodology for adjusting the axial turbine in a turbocharger

115 Pages·2014·9.66 MB·English

by  Thibault De Jaeger

CONFIDENTIAL UP TO AND INCLUDING 06/30/2014 - DO NOT COPY, DISTRIBUTE OR MAKE PUBLIC IN ANY WAY
Development of a CFD simulation methodology for
adjusting the axial turbine in a turbocharger
Thibault De Jaeger
Supervisors: Prof. ir. Erik Dick, Prof. dr. ir. Joris Degroote
Counsellor: Ir. Dieter Fauconnier
Master's dissertation submitted in order to obtain the academic degree of
Master of Science in Electromechanical Engineering
Department of Flow, Heat and Combustion Mechanics
Chairman: Prof. dr. ir. Jan Vierendeels
Faculty of Engineering and Architecture
Academic year 2013-2014

CONFIDENTIAL UP TO AND INCLUDING 06/30/2014 - DO NOT COPY, DISTRIBUTE OR MAKE PUBLIC IN ANY WAY
Development of a CFD simulation methodology for
adjusting the axial turbine in a turbocharger
Thibault De Jaeger
Supervisors: Prof. ir. Erik Dick, Prof. dr. ir. Joris Degroote
Counsellor: Ir. Dieter Fauconnier
Master's dissertation submitted in order to obtain the academic degree of
Master of Science in Electromechanical Engineering
Department of Flow, Heat and Combustion Mechanics
Chairman: Prof. dr. ir. Jan Vierendeels
Faculty of Engineering and Architecture
Academic year 2013-2014
Deauteurenpromotorgevendetoelatingdezescriptievoorconsultatiebeschikbaartestellen
endelenervantekopi¨erenvoorpersoonlijkgebruik. Elkandergebruikvaltonderdebeperkin-
gen van het auteursrecht, in het bijzonder met betrekking tot de verplichting uitdrukkelijk
de bron te vermelden bij het aanhalen van resultaten uit deze scriptie.
The author and promoter give the permission to use this thesis for consultation and to copy
parts of it for personal use. Every other use is subject to the copyright laws, more specifically
the source must be extensively specified when using from this thesis.
Gent, Juni 2014
De promotor De begeleider De auteur
The promotor The supervisor The author
Prof. dr. ir. E. Dick Prof. dr. ir. J. Degroote Thibault De Jaeger
iii
Development of a CFD simulation methodology for
adjusting the axial turbine in a turbocharger
Thibault De Jaeger
Supervisors: Prof. dr. ir. Erik Dick, Prof. dr. ir. Joris Degroote
Master’s dissertation submitted in order to obtain the academic degree of
Master of Science in Electromechanical Engineering
Deparment of Flow, Heat and Combustion Mechanics
Chairman: Prof. dr. ir. Jan Vierendeels
Faculty of Engineering and Architecture
Academic year 2013-2014
Summary
The performance of a turbocharger exhaust gas turbine on a medium speed diesel engine is
studied. Duetocontinuousredevelopmentofbothengineandturbochargerformorestringent
emission legislations, expensive engine and turbocharger test are necessary to achieve a good
matching.
Based on the theory of Computational Fluid Dynamics (CFD), a numerical model of the
turbine is developed in order to analyse the performance of the M40 T266 turbine from
KompressorenBauBannewitz(KBB)GmbHonthe16VDZCengine,manufacturedbyAnglo
Belgian Corporation (ABC) NV. A possible solution for improving the engine performance
was found to be an increase of the stator flow area, enlarging the choking mass flow rat.
A geometrical extrapolation of the turbine blade is performed to analyse the effect of an
increased stator flow area, and a data map for this adjusted turbine is constructed for use in
1D simulation software.
Keywords
CFD, turbocharger, axial turbine, mass flow rate choking
Development of a CFD simulation methodology for adjusting the axial
turbine in a turbocharger

Thibault De Jaeger
Supervisors: Erik Dick and Joris Degroote

Department of Flow, Heat and Combustion Mechanics, Ghent University,
Sint-Pietersnieuwstraat 41, B-9000 Ghent, Belgium

Abstract International Maritime Organization (IMO). Their most stringent
new emission legislation is the IMO Tier III legislation, which will
come into effect in 2016.
The performance of a turbocharger exhaust gas turbine on

a medium speed diesel engine is studied. Due to
continuous redevelopment of both engine and
turbocharger for more stringent emission legislations,
expensive engine and turbocharger test are necessary to
achieve a good matching. Based on the theory of
Computational Fluid Dynamics (CFD), a numerical model of
the turbine is developed in order to analyse the
performance of the M40 T266 turbine from Kompressoren
Bau Bannewitz (KBB) GmbH on the 16VDZC engine,
manufactured by Anglo Belgian Corporation (ABC) NV. A
possible solution for improving the engine performance
was found to be an increase of the stator flow area,
enlarging the choking mass flow rat. A geometrical
extrapolation of the turbine blade is performed to analyse
the effect of an increased stator flow area, and a data map
for this adjusted turbine is constructed for use in 1D
simulation software.
Introduction
For over a century, the medium speed diesel engine has found
a widespread use in maritime, locomotive, traction and power Figure 1. IMO NOx emission legislations [1]
generation applications. Its reliability, efficiency and robustness
Engine manufacturers like Anglo Belgian Corporation (ABC) NV
have made the medium speed diesel engine the backbone of
have to respond to these emission legislations with new
the transport industry today. One of the major features in
concepts such as Exhaust Gas Recirculation (EGR) and Miller
achieving this success was the use of turbochargers. A modern
timing. Miller timing lowers the combustion temperature by
turbocharged engine uses the exhaust gas to spin a turbine,
changing the intake valve close (IVC) time. The lowered
which drives a centrifugal compressor to compress the ambient
combustion temperature is beneficial for NOx emissions, but
air and deliver it to the engine cylinders. Because a turbocharger
implies a shortened compression stroke in the cylinder. This loss
is intrinsically a form of energy recuperation, turbochargers play
in compression has to be compensated by a higher boost
a big role in the race towards more efficient internal combustion
pressure delivered by a turbocharger. This poses high demands
engines. And so, in recent years, special attention has been
on the turbocharger design, and requires re-development of
given to the development and improvement of turbocharger
both turbocharger and engine.
systems.
A constraint posed on the re-development of engines
Diesel engines have great advantages as energy
for more stringent emission legislations is the difficult matching
providers in many applications. However, the diesel engine has
between the turbocharger and the engine. The reciprocating
high nitrogen oxides (NOx) and Particulate Matter (PM) motion of the cylinders results in pulsating flow in intake and
emissions due to its lean operation and the short available
exhaust systems, which is detrimental to the turbocharger
mixing time in the combustion chamber. With the increasing
performance. Simple algorithms, like the algorithm that KBB
environmental awareness of the last decades, several
uses for the matching of turbochargers, can give an idea
organisations have imposed emissions legislations for medium
towards the selection of an appropriate turbocharger, but in
speed diesel engines. One of those organisations is the
order to achieve a good match, multiple expensive engine tests
Page 1 of 6
and turbocharger tests have to be performed. With the Pre-processing
constantly changing emission legislations, this method seems
inefficient time- and money-wise. In order to reduce costs of this The steps performed and the software used are presented in
re-matching, many manufacturers have been leaning towards Figure 2.
specialist software. One dimensional engine simulation software The geometry is imported in ANSYS TurboGrid as
like GT-Power, commercially available at Gamma Technologies, blade profiles at different span locations. The turbine consists of
allow for analysis of a wide range of parameters and phenomena 20 stator blades and 45 rotor blades, with a blade height 53mm
related to engine performance, while three dimensional and outside diameter 266mm. With TurboGrid, the grid is
Computational Fluid Dynamics (CFD) software like Fluent and constructed separately for stator and rotor. The pre-processing
CFX, both commercially available at ANSYS Inc., allow for is then performed in ANSYS CFX-pre
analysis of the fluid flow in the engine components.
The object of this thesis is the T266 axial exhaust gas Grid Generation
turbine of the M40 turbocharger from KompressorenBau
Bannewitz (KBB) GmbH, of which two are placed on the
In order to construct a grid for numerical simulations of the flow
16VDZC engine from ABC. During an engine test performed in
in turbo machinery, a correct topology around the blades has to
May 2013, it was found that the performance of the engine was
be selected. The topology acts as a framework for the grid
not sufficient. Some important parameters from this engine test
around the blade.
are given in the table below.
Traditional grid topologies consist of combinations of

H,J,G and L grids. Preferably an O-grid is included. This O-grid
Table 1. Experimental test data for engine operating point 100%
adds a ring around the blade with a very fine mesh, for accurate
Engine Load 100% boundary layer results on the blade surface. These conventional
mesh topologies typically require a substantial amount of user
Engine Speed 1000 rpm
manipulation to construct a grid of acceptable quality. With
Mechanical Power 3400 kW complex blade geometries like torsioned rotor blades, this
Specific Fuel Consumption 214.6 g/kWh method of meshing is very inefficient. Furthermore, these
traditional topologies often result in an excessive mesh
Specific Air Consumption 9.03 kg/kWh
resolution within the blade passage when a sufficient boundary
Static Compressor Pressure Ratio 3.838
layer resolution is required [2]. ANSYS TurboGrid provides an
Total-to-Static Turbine Pressure Ratio 3.31 alternative. The Automatic Topology and Meshing (ATM)
optimized topology method is preferred to generate high-quality,
structured grids without the constraints of the traditional
The goal of this dissertation is to provide ABC and KBB with a
topologies. With the ATM method a structured, hexahedral
roadmap to improving the turbine design and its match with the
mesh is created separately for the stator and the rotor.
16VDZC engine short term. With the use of CFD software (CFX)
For the generation of turbine characteristics, a
the turbine characteristics are analysed and a proper possible
relatively coarse mesh for a single blade passage is
solution is proposed. A second, more long-term envisioned, goal
constructed, in order to create a data map of a turbine in a
is to develop a simulation methodology for determining the
reasonable timespan. This mesh has a number of cells of
turbine characteristics, and construct a data map of the axial
1.3E+06.
turbine for the use in 1D-engine simulation software like GT-
Power, and thus providing an alternative to expensive engine
and turbocharger tests.

Figure 3. Stator and Rotor Grid at 50% span
Solver Settings
The Shear Stress Transport (SST) turbulence model was
applied to model the flow in the blade passage. This model
combines the advantages of the k-ε model away from the walls
and the k-ω model near the walls.
Flow problems that involve moving parts (as in
turbomachinery) cannot be modelled with one reference frame.
Figure 2. The ANSYS – CFD environment
The geometry consists of two fluid zones, stator and rotor, with
an interface boundary separating the zones. The stationary
zone can be solved with the stationary frame equations,
whereas the rotational section of the geometry can be solved
using moving reference frame equations. The mixing plane
Page 2 of 6
approach was used to treat these equations at the interface. In
the mixing plane model each fluid zone is treated as a steady-
state problem. Flow-field data from the zones are passed as
boundary conditions that are spatially averaged at the mixing
plane interface. This mixing removes any unsteadiness due to
circumferential variations in the passage-to-passage flow field (
wakes, shock waves, separated flow), therefore yielding a
steady-state result. Despite the simplifications inherent in the
mixing plane model, the results can provide reasonable
approximations of the time-averaged flow field. Another option
is the Sliding Mesh model (SM). The Sliding Mesh method
models the relative motion of the two zones, where the rotor
position is adjusted with every time step. This method is more

accurate than a MP model but it is more computationally
intensive. Figure 4. Grid independence test
As boundary conditions, the rotor speed, total inlet
Validation with Experimental Data
pressure, static outlet pressure and total inlet temperature were
specified. The flow direction at the inlet is normal to the
boundary, and the transported turbulence quantities at the inlet In order to validate the CFD calculations, experimental values
is defined by a turbulence intensity of 5% and a turbulence are provided by KBB of the turbine. These were provided as a
length scale of 0.038∗ 𝑑 , where 𝑑 is the hydraulic diameter at data map. The data is experimentally determined on a
ℎ ℎ
the inlet (𝑑 −𝑑). turbocharger test bed, and contains four performance
𝑜 𝑖
The fluid flowing through the turbine is the cylinder parameters of the turbine: reduced speed, reduced mass flow
exhaust gas. As an approximation, the fluid data from air was rate, total-to-static pressure ratio, turbine effective efficiency and
used. This is a good approximation because of the high air to total inlet temperature. The reduced speed and reduced mass
fuel ratio in a diesel engine. However, due to the high exhaust flow rates are calculated by:
temperature, the specific heat capacity Cp of the gas is adjusted
to 1120 J/kgK. This is a value commonly used for exhaust 𝑛 =𝑛 ⁄√𝑇 [𝑟𝑝𝑚/√𝐾]
gasses and was recommended by KBB. 𝑟𝑒𝑑𝑢𝑐𝑒𝑑 𝑎𝑐𝑡𝑢𝑎𝑙 00

Validation √𝑇00 𝑘𝑔 √𝐾
𝑚̇ =𝑚̇ [ ∗ ]
𝑟𝑒𝑑𝑢𝑐𝑒𝑑 𝑎𝑐𝑡𝑢𝑎𝑙 𝑝 𝑠 𝑘𝑃𝑎
00
Grid Independence
The effective turbine efficiency is calculated as:
In order to analyse the influence of mesh size on the results, a
grid independence study is executed with 4 different grids. Each 𝜂𝑒𝑇 = 𝜂0,𝑖𝑠𝑇∗𝜂𝑚 =𝜂0,𝑖𝑠𝑇∗0.97
grid is constructed with the same topologies for both the stator
and the rotor section, but with coarser parameters. For each grid Where 𝜂 is the total-to-static isentropic turbine efficiency.
0,𝑖𝑠𝑇
the calculation is executed with inlet conditions corresponding The mechanical efficiency 𝜂 consists mainly of bearing and
to the measured data from the ABC engine test at 100% engine 𝑚
friction losses (all mechanical losses in a turbocharger are
load. The influence of the number of nodes on the mass flow
attributed to the turbine side).
rate and the effective turbine efficiency is shown in Figure 4. This
Figure 5 and Figure 6 show the comparison between
graph shows that if the amount of cells is higher than 1.3 million,
the experimental results and the numerical results obtained from
the difference in the properties displayed are negligible. It is then
the CFD computations. It is clear from these figures that the CFD
safe to assume that 1.3 million is the correct choice. For a higher
calculated results are close to the experimental values. The
amount of cells the increase in calculation time outweighs the
maximum difference is 3.1% for the efficiency. CFD calculated
small increase in accuracy.
mass flow rates are higher than the experimental values and the
Table 2. Grid numbers and some mesh parameters average difference between experimental and numerical values
is 7.8%. An exact cause for this difference in mass flow rate is
Nr of cells grid 1 grid 2 grid 3 grid 4 unknown, but it is probably due to slightly different properties of
Stator 391680 538110 716000 1153200 the fluid. A higher mass flow rate compared to experimental data
Rotor 370168 414494 580404 955264 is not uncommon for steady flow simulations for exhaust gas
turbines [3].
total 761848 952604 1296404 2108464

Span wise 80 90 100 120
Shroud tip 10 12 14 18
Size factor 1 1.1 1.2 1.4
Min. y+ 2.75E-01 2.72E-01 2.69E-01 1.42E-01
Max. y+ 3.89E+01 2.93E+01 2.58E+01 7.24E+00
Page 3 of 6
Figure 5. Comparison of CFX calculated efficiencies to experimental
values
Figure 7. Turbine efficiencies for engine operating points plotted on the
efficiency curve
The turbine operating point for 100% engine load is situated at
a high pressure ratio for the turbine. A comparison with the
operating point of the turbine on the ABC 8DZC engine learns
that the pressure ratio for the 16VDZC engine at 100% load is
3.31, and for the 8DZC engine this is 2.86, thus the 16VDZC
engine has a higher inlet pressure for the turbine. Small
differences between these engines might be expected due to
the different construction of exhaust manifolds, but
nevertheless, this indicates that the stator flow area (100cm²) for
the turbine is too small. As a result from the high pressure ratios,

the operating point of the turbine is close to choking conditions,
Figure 6. Comparison of CFX calculated mass flow rates to experimental which results in a lowered efficiency.
values
Choking means that the flow reaches sonic state in a
blade passage and may occur at high pressure ratios in either
Numerical Simulations with Engine Test the stator or the rotor. For a turbine with low degree of reaction,
Data choking occurs primarily in the stator. The choking mass flow
rate is independent of rotational speed. The corresponding
stator pressure ratio and mass flow rate are [4]:
Texhpee irnimpuetn dtaalt ae nfogri neen tgeinste a onpde araretin sgh opwoinn tisn wtaabsle d 3e.r ived from an 𝑝1 ℎ1 𝑛⁄𝑛−1 2 𝑛⁄𝑛−1
=[ ] =[ ]
𝑝 ℎ 𝛾+1
Table 4. CFX input data derived from experimental engine test 01 01
1 1
Load % 75 90 100 100+ 𝑚̇ =𝐴 𝜌 𝑐 [ℎ1]𝑛−1+2
𝑐 1 01 01 ℎ
Engine Speed rpm 1000 1000 1000 1000 01
Power kW 2550 3060 3400 3621 Where n=2.32 is the polytropic exponent and 𝛾=1.4, the ratio of
specific heat capacities. With the numerical results it is
𝑚̇ kg/s 3.87 4.27 4.47 4.56
𝑒𝑥ℎ
calculated that the mass flow rate for choking in the stator is
Turbo Speed rpm 31261 33520 35419 36520
4.606 kg/s.
p00 kPa 284 316 343 358 Similarly for the rotor [4]:
p2 kPa 103 103 104 104 𝑝1 ℎ2 𝑛⁄𝑛−1
=[ ]
T00 K 726 766 806 846 𝑝 ℎ
01 01
ℎ 𝑛−11+12
For every engine load point, a CFX numerical simulation was 𝑚̇ =𝐴 𝜌 𝑐 [ 2]
executed and the resulting turbine efficiencies are plotted on the 𝑐 2 01 01 ℎ01
acquired efficiency characteristic in Figure 7. The efficiency is
The mass flow rate for choking in the rotor is 6.209kg/s. It is clear
decreasing for increased engine load, while the pressure ratio is
that due to the turbine’s relative low degree of reaction, the
increasing. The turbine is then not working at optimal efficiency
stator is more prone to choking. The typical curve for mass flow
for the engine maximum load. For marine diesel engines, this is
rate as a function of total pressure ratio is provided in Figure 8
not an uncommon observation, since the turbocharger needs
[4]. At 100% engine load, the average mass flow rate through
high efficiency at lower loads as well. The higher turbocharger
the exhaust gas turbine is 4.47kg/s, which is very close to the
efficiencies at lower loads then result in a slightly deteriorated
“maximum” mass flow rate (choking mass flow rate) of the
efficiency at maximum load. However, for an engine used for
turbine.
power generation applications like the 16VDZC, a high efficiency
is needed at maximum load, and the efficiency at lower loads is
of less importance.
Page 4 of 6
the trailing edge of the neighbouring blade. After the normal
shock, the boundary layer becomes wider. The normal shock at
the end of the supersonic zone at the suction side, results in
shock losses and due to small radial differences in the shock
intensities, strongly rotational flow. This all leads to a decrease
of the turbine efficiency.
At a total-to-static pressure ratio of 3.8, the flow
reaches sonic speed at the throat section between the suction
side of the blade and the trailing edge of the neighbouring blade.
At this point, the outlet velocity of the stator is supersonic. In the
flow field, the large supersonic zone is visible, with a strong,

normal shock at the end. The wake flow of the neighbouring
Figure 8. Mass flow rate in function of pressure ratio for choking in the blade is again very influential on the shape of this supersonic
stator [4] zone.
Mass Flow Choking in the Stator
In order to understand why the efficiency is decreasing for
pressure ratios above 2.7 (see figure 7), a flow analysis of the
stator is executed for different pressure ratios. For this CFD
analysis, the flow fields in the stator are numerically calculated
with a different mesh then before. The grid independence test
indicated that the mesh was accurate for calculating the
properties in the previous section, but in order to accurately
describe the flow near boundary layers, lower y+ values are
required. This was not achieved in the previous numerical
simulations, in order to keep calculation time reasonable. With a
similar methodology, but a fine mesh (min. y+ 8.388E-02, max.
y+ 1.102E00) the flow fields in the stator are analysed for
different pressure ratios. The most important figures of this
analysis are presented here.

Figure 10. Mach number distribution in the stator for a total-to-static
pressure ratio of 3.8.

Figure 9. Mach number distribution in the stator for a total-to-static
pressure ratio of 2.824. The minimum Mach number in this figure is set
to 0.6, to be able to distinguish the supersonic zone better
A pressure ratio of 2.824 is just greater than the pressure ratio
at which maximum in the best fitted efficiency curve is expected.

Figure 9 shows the Mach number and pressure plots of the flow
field. Around this pressure ratio, the flow field alters from being Figure 11. Mach number distribution in the stator for a total-to-static
entirely subsonic, and at the suction side of the vane, sonic pressure ratio of 5.
speed is reached in one point. With increasing pressure ratios,
this point grows into a supersonic zone, ending in a normal
shock. The supersonic zone is influenced by the interaction with
Page 5 of 6