Gas Turbine (GT) design has always become competitive for low fuel consumption,
meets the environmental codes, while keeping the GT light in weight and compact
in size. This delivers great advantage in aero engines, which lowers the overall
weight and structure integrity of the aircraft, as well as keeping the space
efficient for related industry. Also, GT engines are widely used in oil and
gas industry, to generate electricity power as well as operation of rotational
machines for oil gas lift. The bulky size of gas turbine-compressor/generator
set has always become a great factor and limit in designing oil and gas offshore
platform. It also limits the flexibility of improving the existing oil and gas
platform by exchanging different model of the GTs according to needs.
The best design idea to favor the function range of gas turbine is to reduce
the size of gas turbine. A diffuser is located in between the High Pressure
Stages (HP) and Low Pressure Stage (LP) turbine, Fig. 1, namely
intermediate turbine diffuser with the function of pressure recovery.
||Schematic representation of the double shaft gas turbine
This diffusers size increases with higher power output, due to higher
combustor exit flow.
The diffuser function is to increase the flows
static pressure. However, by shortening this diffuser while retaining its inlet
and outlet size (increasing the cone angle), flow separation will occur, creating
a boundary layer that will significantly reduce the diffusers
efficiency, thus the gas turbine engines
Numerous researches are carried out to study the diffusers behavior to
improve the GT performance. Mehta and Bradshaw (1979)
used an aggressive diffuser as part of a wind tunnel design.
|| Flow control mechanism (a-b) Active and (c) Passive measures
for re-energization of boundary layer (Lord et al.,
They mentioned one of the problems that they have faced while using this type
of diffuser is flow separation due to the cone angle of the diffuser is higher
than 5° and he managed to solve it by introducing screens made up by woven
wire gauze in the diffuser.
Lord et al. (2000) investigated on active or
passive flow controls to design more aggressive transition duct geometries with
larger radial offsets. The first type can either be energization of boundary
layer by injecting high energy fluid or removal of low energy fluid from critical
wall region, as in Fig. 2.
Lin (2002) performed a thorough review on low-profile
vortex generator and their ability to prevent flow separation. The working principle
is to transport high momentum fluid from the core flow into the boundary layer
by means of stream wise vortices.
Fukudome et al. (2005) did an analysis on separation
control of high angle of attack air foil for vertical axis wind turbines. One
of the methods they used is by using turbulence promoters. They had concluded
that the present of turbulence promoter is useful to modify the aerodynamic
performance of the vertical axis wind turbine by altering the separation.
Sieker and Seume (2007) discussed that the power and
efficiency of turbines are strongly depend on the performance of the exhaust
turbine diffuser. They did an experimental analysis to relate the influence
of rotating wakes on separation in the turbine exhaust diffusers.
Zulkefli and Ahmad (2010) had reported numerical simulation
results of the effect of streamwise vortices on turbulent flow structure. Their
objective in their simulation is to obtain the optimum parameter of sub-boundary
layer vortex generator. They had used commercial code fluent 6.3TM
to simulate their model. They have stated two different types of flow control
devices, which are passive and active control device.
Intermediate passageway and energy promoters geometry: Reducing the
size of gas turbine can have many advantages. This includes increasing the stability
of the shaft, where shorter shaft has better stability and balance and reducing
its total weight. Other than that, this can contribute to more flexible position
allocation for GT at offshore platform structure, as the space available is
Two patents held by general electric, namely Graziosi and
Kirtley (2006) and Widenhoefer et al. (2009)
introduces inter turbine diffusers with different type of method in eliminating
flow separation. In both inventions, secondary air is injected to energize the
boundary layer to prevent the separation. The air will be taken from the compressor
section of the gas turbine, due to suitable static pressure ratio between suction
port and the injection slot. Santer et al. (2010)
explained that passive flow control is less complex than active flow control,
because there is no need for handling additional fluid streams at unsteady flow
rates. Merely the installation of fixed components at the right position would
be very beneficial to re-energize boundary layer. He also installed a work package,
namely EU project AIDA to evaluate the application of passive flow control devices
in both compressor and turbine transition ducts. Low vortex generators have
been designated for one of their super aggressive intermediate turbine diffuser
setup, AIDA and TTTF at Graz University of Technology. These ducts shows fully
separated flow on casing wall and therefore suitable for the study of passive
flow control devices in order to show improvements after installation.
Therefore, the objective of the present work is to investigate the possibility
of shortening the intermediate diffuser in a GT by CFD simulation. Passive and
aggressive diffusers were modelled and simulate assuming 2-D incompressible
gas flow using FLUENT commercial software. The installation of energy promoters
with five different heights (0.5, 1.0, 1.5, 2.0 and 2.5 mm), at five different
promoters location in the upstream part of the diffuser were investigated.
MATERIALS AND METHODS
For this study, CFD simulation was carried out using both GAMBIT version 2.2.30
for the diffuser design and FLUENT version 6.3.26 for simulation. The simulations
is to prove the existence of flow separation as the diffusers
cone angle increases and reduce the diverse effect of the separation by introducing
energy promoters. Accordingly, two main configuration were to be considered
in the simulation, namely, the diffuser and secondly, the promoter.
|| The benchmark S-shape diffuser
|| The aggressive diffuser
The diffusers: Firstly, S-shape diffuser with normal cone angle was
designed to create a benchmark for comparison, as shown in Fig.
3. The total length of the diffuser is 482 mm with inlet height of 50 mm
and exit height of 80 mm. The radii of curvature of the bends are 191 mm. The
total length of the stream wise from inlet to outlet is 600 mm. Then, the diffuser
was modified by shortening its length to create another model of an aggressive
diffuser, as shown in Fig. 4.
A very aggressive and separating duct design with an area ratio AR of 1.62
and L/hin of 2.56 was chosen based on the recommendations of Gopaliya
et al. (2011). The length was reduced from 482-300 mm, while the
inlet and outlet were maintained with 50 and 80 mm, respectively.
The aggressive diffuser has 100 mm radii of curvature and the total stream
wise total length of 314 mm.
The promoters: The energy promoter were introduced and tested at different
position and different energy promoter height.
||Geometries of the energy promoters, H: Various heights of
0.5, 1.0, 1.5, 2.0 and 2.5
A vortex generator model has been adopted for the investigation of various
configurations within a design of experiments for a flow controlled intermediate
turbine diffuser by Wallin and Eriksson (2006, 2008).
Four parameters with influence on the vortex generator performance were allowed
to vary within the bounds:
||A non-dimensional location of the trailing edges relative
to the baseline separation line ΔSVG/hVG = 0-11
||The minimum height corresponds with the boundary layer thickness height
||hVG = 1.3-2.9 mm
||LVG/hVG = 25-5.5xhVG
||Angle of attack (10-26°)
According to that, the final selected shape of the promoters in the present
investigation is shown in Fig. 5. The promoters were selected
with triangular shape of 25° frontal edge. Five different heights of 0.5,
1.0, 1.5, 2.0 and 2.5 mm were adopted in this investigated. And, five different
promoters location in the upstream part of the diffuser were simulated.
SIMULATION AND EVALUATION PROCEDURE
Numerical implementation: The modeling of the diffusers and promoters
geometries with various was carried out by GAMBIT version 2.2.30. The CFD simulation
was carried out by FLUENT version 6.3.26 software. The field was meshed using
the volume meshing tool. Only one element type was successfully applied that
is the triangular quadrilateral type. The mesh was unstructured, as shown in
Fig. 6. Errors were received when attempting to use any other
element type. The mesh was refined gradually to prove the grid independency.
The boundary conditions applied for the simulation purpose are 200 m sec-1
inlet gas velocity, with 10% turbulence intensity.
At the outlet, the pressure was set at 0.0 Pa gauge scale. The solid boundaries
have roughness of 0.01 and no slip shear conditions. When the governing equations
of mass, momentum and energy conservation solved, the turbulence model selected
is the k-ε model.
The performance evaluation: The exit static pressure is then recorded
to identify the efficiency of the diffuser, which would be the ultimate goal
for comparison, in determining the diffuser performance at various configurations
and promoters geometries.
The pressure rise coefficient given by:
as the dynamic head at inlet.
||The unstructured meshing with tetrahedral element type
Since T = 200°C, then:
the diffusers efficiency, is:
where, Cpi is the ideal pressure raise coefficient, as:
RESULTS AND DISCUSSION
Results of different configurations are plotted accordingly. The results are
presented in form of velocity vectors, pressure contours and static pressure
profiles at inlet and outlet, at different promoters
positions and sizes.
Flow analysis: The velocity vector plots are used to indicate the flow
behaviour. The flows in normal S-diffuser and aggressive diffuser are shown
in Fig. 7 and 8, respectively.
Qualitative analysis of the flow could be dropped from this simulation results.
The comparison will focus on the critical portion of the diffuser at the bend
of the lower surface, where the flow decelerates dramatically due to the adverse
|| Velocity vectors by magnitude of the flow in normal diffuser,
Vin = 200 m sec-1
|| Velocity vectors by magnitude in aggressive diffuser without
promoters, Vin = 200 m sec-1
|| Velocity vectors by magnitude of the flow in aggressive diffuser
with promoters, Vin = 200 m sec-1
The same flow passes through the normal diffuser smoothly, as in can be noticed
in Fig. 5. While, by reducing the diffuser length from 482
mm to 300, a reversal flow and separation are taking place in the case of the
aggressive diffuser, by passing the same gas flow. The resulted circulation
bubble in the second case will effects the performance and reduces the pressure
recovery at the exit of the diffuser. To maintain the advantage of reducing
the length and increasing the pressure losses due to separation, an energy promoter
(H = 1.5 mm) is introduced in the flow of the aggressive diffuser just upstream
of the allocated separation starting point. The resulted simulation of the flow
structure is shown in Fig. 9. It is quite clear that the promoter
added momentum to the flow particles in the high bend portion and successfully
eliminated the occurrence of the reversal flow.
Analysis of promoters position:
Five different locations of the promoters in the upstream of the bend have been
simulated with promoter of 1.5 mm height. Those positions are at 5, 10, 15,
20 and 25° counterclockwise of the horizontal of the central point of the
bend, as below:
The five positions have been simulated in the aggressive diffuser. The results
of the static pressure are shown in Fig. 10. The predicted
values of the pressure are in the centerline of the duct. The inlet static pressure
is the same in all promoter position cases. At outlet, best performance to reduce
the back pressure is gained at promoter position at 10° upstream the second
bend, as shown in Fig. 11.
The exit static pressure is highest value at position of energy promoter at
10°. Positioning the promoter at angles of 25, 20 and 15, the exit static
pressure shows slightly lower pressure recovery. However, in all cases of position
angles, the promoters increase the momentum of the fluid to overcome the adverse
pressure and result in smaller separation region as compared with cases of with
no promoter installation. On the other hand, the exit static pressure bounces
back when the position of the energy promoter exceeds 10 to 5°. The simulation
analysis shows that the best contribution of the energy promoter is at 10°
|| The promoter positioning angle
Analysis of the promoters size: Having the best position of the
promoters found, which 10° is, the height effect of the energy promoter
is tested at that angle. For all cases, the promoter frontal angle of attack
is fixed at 25°, as recommended by Wallin and Erriksson
(2006, 2008), while the height, Hpr is
varied. and the length of the promoter is changing, accordingly.
Looking at Fig. 12, the energy promoter at height of 2.0
mm delivers the highest static pressure recovery. The other heights, 0.5, 1.0,
1.5 and 2.5 mm are resulted in lowers static pressure recovery, compared to
the case of 2.0 mm height.
Efficiency analysis: Comparison of static pressure recovery between
the normal S-shaped diffuser, aggressive diffuser and aggressive diffuser with
energy promoters shows that energy promoters are able to increase the aggressive
diffusers efficiency. Simulation
results summarized in Table 1 show a comparison of the predicted
static pressure at the three cases of diffusers. The normal S-shaped diffuser
is simulated, showing good diffusers
efficiency, at 71.2 %. The S-shaped diffusers
length is reduced by 38.4% and flow separation occurs, with efficiency at 16.4
% only. After the energy promoter is introduced, efficiency of the aggressive
diffuser is increased to 38%.
But, it is nowhere near a benchmark normal diffusers static pressure
recovery. Some of the reason contributes to this pattern, may be due to over-aggressive
diffuser design (very short centre-line length), which creates a large separated
region that is not fully eliminate-able of separation.
||Predicted static pressure results in the aggressive diffuser
at various promoters positions, Hpr = 1.5 mm, Vin
= 200 m sec-1
||Predicted static pressure results in the aggressive diffuser
at various promoters heights, position angle = 10° and Vin
= 200 m sec-1
||Efficiency comparison between different cases of promoters,
all cases with Cpi = 0.61
Other than that, due to insufficient resources on current existing aggressive
diffuser design geometry, a typical S-shaped diffuser was chosen instead. This
raises the problem that, perhaps this diffuser design, was not in the effective
range of energy promoters utilization.
The gas flow through normal S-diffuser, aggressive S-diffuser with and without
energy promoters have been simulated by FLUENT commercial software to investigate
turbine size reduction by shortening the intermediate turbine diffusers downstream
of HP turbine stages. The reduction of the normal S-diffuser length from 482
mm to be aggressive with 300 mm, have reduced the diffusion efficiency from
71 to 16%. Insertion of energy promoters has enhanced the aggressive diffuser
efficiency to be 38%. This simulation findings are demonstrated the possibility
of reducing the S-diffuser length but energy promoters are required to be inserted
upstream of the separation bubble to reduce the reversal flow and enhance the
static pressure recovery. The optimum promoters (with 25° frontal angle)
height is 2.0 mm and location at upstream the second bend at 10° position
Since the investigation is performed in 2-D, a similar simulation is recommended
to be conducted in 3-D CFD to further investigate the flow behavior and its
effect and relate them to this 2-D simulation.
The authors acknowledge Universiti Teknologi PETRONAS for the support to conduct
the work and the presentation of the results in the present study.