Indian Journal of Engineering & Materials Sciences

Vol. 20, April 2013, pp. 125-131

Flow characteristics of a symmetric Y-shaped diffusing duct with zero yaw angle
Netrapal Singh, Abdur Rahim* & Mohd Islam
Department of Mechanical Engineering, Jamia Millia Islamia, New Delhi 110 025, India
Received 15 June 2011; accepted 20 March 2013

An experimental investigation of a Y-shaped diffusing duct with an aspect ratio and area ratio of 2 and turning angle of
22.5°/22.5° is reported for flow and performance characteristics. A calibrated five-hole pressure probe in null mode is used
to measure the flow parameters. The performance parameters are evaluated based on mass-averaged quantities. It is
observed that the static pressure recovery coefficient and the total pressure loss coefficient increase continuously up to the
exit of the diffusing duct at a constant rate. It is also observed that the flow is divided into two flow regimes whereas at the
merger, the two streams mixed with each other resulting a centrally core flow with almost uniform intensity and a pair of
vortices. The longitudinal velocity distribution at the exit is seen to be nearly uniform.

Keywords: Y-shaped diffusing duct, Five-hole probe, Secondary flows

The conversion of kinetic energy into pressure energy velocity ratio improves the pressure recovery. The
within a fluid-mechanical system finds wide limitations of several types of inlets of jet powered
application in aero-engines. This is necessitated by the engines (ducting system, large external drag,
use of curved diffusers due to space restrictions and insufficiency of the air intake area, etc.) led to the
design capability. The Y-shaped ducts are used as development of submerged inlet (submerged in the
intake ducts in aircrafts as well as in heating ventilating fuselage) and the experimental investigations on these
and air-conditioning (HVAC) ducts. These ducts fulfill inlets have been carried out and presented by Frick et
the requirements of higher flow and space constraints al.3 for one side inlet in a small scale wind tunnel.
due to geometrical variations accompanied by diffusion Simon et al.4 proposed that an external compression
which make the flow characteristics quite complex. bump inlet is superior to the internal compression ramp
The curvature of both limbs of the Y-duct generates inlet for a supersonic fighter aircraft twin intake
centrifugal forces, which get balanced by the pressure induction system. Pai5 has discussed two design
gradient in the plane of the bend. The central part of the aspects–intake-airframe integration and engine intake
fluid is forced outwards to satisfy the continuity compatibility to bring out the complex nature of intake
resulting in generation of secondary flows. The Y- flows in an air intake system of modern combat
shaped duct investigated in the present study is a aircrafts. Sudhakar and Ananthkrishnan6 have
symmetric duct, which was fabricated to feed flow suggested a simple model to establish the transition
from two inlet limbs into a single exit duct of circular from symmetric to asymmetric conditions in the
shape. supersonic flight on air intake system. Jolly et al.7 have
The studies reported on Y-ducts in open literature proposed the methodology of design and development
refer to high-speed aircraft breathing systems and these of bifurcated Y-ducts. Patel et al.8 have carried out
are not very numerous. The first investigation on investigations on Y-shaped diffusing intake ducts to
phenomenon of originating aerodynamic instability or analyze the flow characteristics within the Y-shaped
buzz was published by Ferri and Nucci1 for supersonic rectangular diffusing duct having an angle of turn as
inlets. Martin and Holzhauser2 extended the work of 22.5°/22.5° and an overall area ratio of 2.0 with an
Ferri and Nucci and investigated an NACA submerged circular exit. The inlet of this duct was through two
twin air intake on a full-scale model of a fighter rectangular cross-sections with aspect ratio of 2.0.
aircraft. They have reported that the change in inlet Detailed analysis carried out using CFD code
______________
‘FLUENT’ for a skewed axial velocity in one limb and
*Corresponding author (E-mail: rahim_ark@rediffmail.com, uniform flow in the other one at average inlet velocities
arahim@jmi.ac.in) of 30, 60 and 90 m/s showed that static pressure
126 INDIAN J ENG. MATER. SCI., APRIL 2013

recovery reduces with increase in skewness and strong order to get the static wall pressure profile along the
secondary flows are observed throughout the length of curved arms of the Y-shaped diffusing duct. The
the diffuser with a highly complex flow at the merging external top wall of the curved duct is provided with
plane. No effect of Reynolds number was observed on appropriate number of holes so as to facilitate the
the performance of the Y-duct for uniform flow at the insertion of a five-hole pressure probe as shown in
inlet for both limbs of the Y-duct. Longitudinal Fig. 1a for pressure and velocity measurement. When
velocity contours depicted the deflection of the flow not in use these holes are sealed using moulding clay to
towards the outer wall all along the length of the ensure proper working of the experimental set-up for
diffuser for all inflow conditions. Saha et al.9 have other measurements such as static wall pressure or
examined the influence of the inlet shape on the airflow parameter at some other cross-sections. Total
performance of twin intake duct. They have also number of measurement sections are ‘10’, i.e., ‘4’ in
observed a twin pair of vortices after the merger plane. arms ‘A’ and ‘4’ in arm ‘B’ (same positions as in arm
Each pair of vortices originates from each limb in all ‘A’), and ‘1’ after the merger (plane A) and ‘1’ near the
cases due to change in the centerline curvature. Saha et exit plane (plane B). So in this way, there are total ‘10’
al.10 have investigated that the influence of inlet shape measurement sections in the test duct from inlet to
on the performance and flow characteristics single outlet as shown in Fig. 1a. Each section has different
offset S-Duct having angle of turn of 22.5°/22.5° using number of stations with different number of measuring
CFD code ‘FLUENT’. They investigated S-ducts points to entirely cover the whole curved area. With the
having 300 mm centerline length and circular exit. help of a traversing mechanism three dimensional
They have concluded that elliptical shaped inlet duct movement of the probe allows in the experimental set-
gives the best performance in terms of pressure up to measure the airflow parameters at sufficient
recovery, loss coefficient and flow distortion at the number of vertical and inclined positions in the whole
compressor faces where as square duct produces the duct. A five-hole pressure probe is used in null mode to
worst flow characteristics. Singh et al.11 have analyzed measure the velocity and pressure. The probe
the effect of corner rounding on the performance and orientation in different planes and the details of the
flow characteristics of a double offset Y-shaped aircraft five-hole probe are shown in Figs 1b and 1c.
intake duct using CFD. They observed that the
Y-shaped duct
performance and flow characteristics of the Y-shaped
intake duct improve significantly with rounding of the The Y-shaped duct is made up of epoxy resin.
sharp corners up to a fixed arc radius. In the present Primarily the duct is designed with the help of Gambit
investigation a symmetric Y-shaped diffusing duct has geometry generation module of Fluent (CFD) package,
been used for analyzing the flow behavior at different then the wood pattern created and finally with the help
cross-sections at zero yaw angle. of wood pattern, plaster of paris mould created and
epoxy resin mixed Y-duct fabricated for
Experimental Set-up experimentation. The aspect ratio is kept 2 at both the
The Y-shaped curved and diffusing transition duct inlet limbs while the area ratio of entire duct is 2. The
having rectangular inlets of arms A and B with a centerline length and radius of curvature for both limbs
circular exit is used for capturing experimental data. of this Y-shaped duct is 300 mm and 382 mm
The two arms of the duct having rectangular inlet respectively. The turning angle is 22.5°/22.5°. The free
cross-sections merge with each other to again form a stream velocity as measured by a calibrated Pitot-static
single circular cross-section at the merging as well as at tube in the wind tunnel was found 12.55 m/s, with the
the outlet section. The duct with fore-body (fuselage) is average inlet suction velocity in the range of 15.06 m/s
coupled with an open circuit subsonic wind tunnel and this results in a velocity ratio of 1.2 at Re = 0.685 × 105.
installed at the exit of the settling chamber with the The circular exit section is connected to the pipeline
help of a pipeline network. The flow in the wind tunnel network, which is fixed to a fitted control valve with
is created by suction with the help of control valve and orifice meter and further connected to the inlet of
a centrifugal blower as shown in Fig. 1. Sufficient centrifugal blower. This blower is running at 3000
numbers of pressure taps are provided on external RPM and equipped with a flow control door for fixing
curved side-walls of the duct to measure the static wall the constant velocity ratio.
pressure. The holes drilled for pressure taps are of 1.5 A rectangular inlet section having dimensions of 50
mm diameter each and are placed sufficiently close in mm × 100 mm for one of the limbs of the Y-shaped
 SINGH et al.: Y-SHAPED DIFFUSING DUCT 127

Fig. 1—Schematic layout of experimental set-up (a) diagram of divided intake duct, (b0 orientation of five-hole probe at a point of
measurement and (c) details of the five-hole probe
128 INDIAN J ENG. MATER. SCI., APRIL 2013

duct and a circular outlet having a diameter of 160 velocity at both limbs is 15.06 m/s with Reynolds
mm is selected. During experimentation at the inlet number, Re = 0.685 × 105. Measurements were taken
there are 5 horizontal stations, i.e., 5 mm far from at six different traverse planes along the test duct.
both vertical walls and 19 vertical measurement
points at each horizontal station to capture the whole Longitudinal velocity distribution
inflow to form a grid of experimental data points to Figure 2a shows normalized longitudinal velocity
plot all the parameters at this plane. Similarly the (Ulong/Uave(in)) contour plots at different cross-sections
horizontal stations and vertical points increase from for the intake duct A (Right Limb) and duct B (Left
inlet to outlet, as per cross-section of different Limb). The longitudinal velocity contours at section
measuring curved sections. At the merger and outlet (1) show that the fluid flow at the inlet of Limb A and
there are 8 vertical as well as inclined locations with Limb B duct is nearly uniform except close to the
31 points at each location, i.e., 5 mm far from both wall. The fluid core as it moves into the duct
walls to cover up the entire circular cross-section for undergoes distortion as a result of the curvature and
data capturing. In this way from inlet to outlet, the the bulk flow moves towards the concave wall in the
number of horizontal, vertical and inclined stations first half of the bend [Fig. 2a, section (1&2)]. It
with different cross-sections are increasing and continues to develop along the same wall in the
accordingly the five-hole probe traverses with the second half [Fig. 2a, section (3&4), convex wall in
help of traversing mechanism having three the second half]. A flow separation can be seen at the
dimensional movement to capture all the data points convex wall at the inflexion plane. [Fig. 2a, section
to plot all the contours and graphs. (5&6)] gives the velocity distribution in the form of
iso-velocity contours at the merging and exit plane. It
Uncertainty analysis is seen that the flow is divided into two flow regimes.
The maximum error due to probe blockage area At the merger, two streams are mixed with each other,
takes place while taking the measurements very close a centrally core flow with almost uniform intensity of
to the bottom wall. In the case of five-hole probe pair of vortices found. The flow velocity is higher in
measurements the sensing head of the probe is not in the central region and reasonably uniform and slowly
that plane due to its ‘L’ shape. The length of ‘L’ is 1.5 decreasing to the outer walls. This implies smooth
times of the stem diameter and due to its length, the merger of the flow from the two ducts and the
effect of blockage in the plane of measurement is separation bubble seen in the two limbs has
significantly reduced and hence no correction has disappeared. A secondary flow induced in the cross
been applied for the blockage effect. The positioning planes of curved ducts caused by the imbalance of the
of the sensing head offset due to probe rotation was centrifugal force and radial pressure gradient in the
always brought back to its original position using the plane of curvature. The flow beyond the merger plane
traverse mechanism. The experimental errors in is seen to develop well and is seen to be nearly
different measured quantities are estimated following uniform at the exit with the intensity of secondary
the method of Kline12 and are given below in Table 1. flow also reducing.
Results and Discussion
Experimental investigations were carried out to 3-D longitudinal velocity distribution at inlet
establish the effect of turning angle on overall Figure 2b shows normalized longitudinal 3-D
performance of the Y-shaped diffuser with velocity (Ulong/Uave(in)) contour plots at 0° yaw angle at
rectangular inlet (aspect ratio AS = 2) and circular inlet cross-sections for the intake duct A(Right Limb)
outlet with area ratio (AR) of 2. The average inlet and duct B(Left Limb). The 3-D wire mesh plot
supports velocity distribution at both the inlets and
Table 1—Estimated uncertainties for the measured quantities reveals that the flow induced in the duct by suction.
Quantity Estimated Uncertainty
Velocity less than Velocity above Total pressure variations
5 m/s 6 m/s The total pressure distributions at the merger and
Mean velocity ±5% ±2% outlet sections are presented in Fig. 2c. The total
Pressure (static & total) ±5% ±2% pressure distribution is almost identical to the
Flow angle ± 0.5° ± 0.5° longitudinal velocity distribution.
 SINGH et al.: Y-SHAPED DIFFUSING DUCT 129

Fig. 2a—Longitudinal velocity contour plots of Y duct at 0° yaw of different sections

Secondary flow velocity distribution at equal distances along the circumference. A multi-
The secondary flow velocity vectors of merger and tube manometer having an accuracy of 0.5 mm of
outlet section are presented in fig. 2d. Normalized water column (approx. 5 N/m2) was used to sense the
cross flow velocity (Usec/Uave(in)) distribution at wall static pressure from these tapings. The normalized
different planes for the 22.5°/22.5°, Y-shaped average wall static pressure along the curved wall for
diffusing duct also analyzed and found that there is no different locations is shown in Fig. 3. At the inlet, there
significant variation observed in the cross flow is no significant variation in the static pressure
velocity distribution up-to section (4) for 22.5°/22.5°, coefficient for all the tapings and this supports the
Y-shaped diffusing duct. The cross flow velocity longitudinal velocity distribution at the inlet. Due to
pattern is nearly similar throughout, in both inlet corner effect of curvature some drop observed initially
limbs of Y-duct. At the merger, two streams are which further increases. The wall pressure coefficient
mixed with each other, centrally core flow with increases continuously along the outer wall of the first
almost uniform intensity of pair of vortices found. bend and inner wall of the second bend, whereas on the
The secondary flow pattern at the exit plane shows opposite side it increases up to the outlet section. This
presence of two pair of vortices with central pair shows the movement of the fluid core towards the
having weak intensity. concave surface. As the flow moves downstream, the
Wall pressure distribution static pressure coefficient starts to increase again due to
The wall static pressure at any given axial location accumulation of fluid near the inner surface of the
was measured as an average of pressure tapings placed second bend.
130 INDIAN J ENG. MATER. SCI., APRIL 2013

Fig. 2b—3-D normalized longitudinal velocity wire-mesh plots at 0° yaw of both inlets

Fig. 2c—Total pressure contours at 0° yaw of merger and outlet sections

Fig. 2d—Secondary velocity vector plots at 0° yaw of merger and outlet sections

Performance characteristics
Based on the five-hole probe measurements, the
average static pressure and total pressure were
evaluated to estimate the performance of the diffuser
in terms of static pressure recovery coefficient (CP)
and total pressure loss coefficient (CL). These values
are calculated in non-dimensional form as:
CP = (Increase in static pressure between two
measurement stations) / Inlet dynamic pressure Fig. 3—Wall pressure distribution along the duct length
 SINGH et al.: Y-SHAPED DIFFUSING DUCT 131

(i) The velocity pattern at the exit is uniform for this

Y-shaped diffusing duct having a 22.5°/22.5°
turning angle.
(ii) The flow velocities being higher in the central
region and reasonably uniform with slowly
decreasing to the outer walls which implies
smooth merger of the flow from the two ducts
and the separation bubble seen in the two limbs
has disappeared.
(iii) At the merger, two streams are mixed with each
Fig. 4—Variation of performance parameters along the duct
other, centrally core flow with almost uniform
intensity of pair of vortices is found.
CL = (Loss in total pressure between two
(iv) The pressure recovery coefficient increases
measurement stations) / Inlet dynamic pressure
continuously when fluid approaches from inlet to
Variations of static pressure recovery coefficient and
exit and the loss coefficient also increases but at a
total pressure loss coefficient for both the limbs along
lower rate as compared to recovery coefficient.
the length are shown in Fig. 4. For limb A, the mass-
averaged static pressure recovery coefficient increases
continuously in the first bend and a slight reduction in References
1 Ferri A & Nucci L M, NACA RM-L50K30, Jan 1945.
the rate of recovery coefficient is observed near the 2 Martin N J & Holzhauser C A, NACA RM-A8F21, Oct 1948.
inflexion plane due to a change in the direction of 3 Frick C W, Davis W F, Randall L M & Mossman E A,
curvature, which leads to turbulent mixing. In the NACA ACR-5I20, Washington, Oct 1951.
second bend, the fluid recovers its lost energy and the 4 Simon P C, Brown D W & Huff R G, NACA RM-E56L19,
loss coefficient is nearly constant up to the exit of April 1957.
diffusing duct, except at X/L = 0.6 and X/L = 0.9 where 5 Pai T G, Aeronautical Development Agency, Bangalore,
India, 1994.
some depression occurs. This may be due to the growth 6 Sudhakar K & Ananthkishanan N, J Aircraft, (1995)
of boundary layer and increase in the losses due to 438-439.
rapid mixing. More or less similar trend found in limb 7 Jolly R, Pai T G & Jayasimha P, HAL, Bangalore, India,
B. It is observed that the variation of total pressure loss 1996.
is almost linear along the whole length of the duct. The 8 Patel T, Singh S N & Seshadri V, J Aircraft, (2005) 113-120.
overall static pressure recovery and total pressure loss 9 Saha K, Singh S N & V Seshadri, Proc. 3rd Int. Conf. FMFP,
IIT Bombay, India, 2006.
coefficients are 0.38 and 0.25 up to the merger plane. 10 Saha K, Singh S N, Seshadri V & Mukhopadhyay S, J
Aircraft, (2007) 1230-1245.
Conclusions 11 Singh R K, Singh S N & Seshadri V, IMechE 222 Pt G: J
From this study, the following conclusions can be Aerospace Eng, (2008) 969-983.
drawn: 12 Kline S J, Trans ASME, J Fluids Eng, 107 (1985) 153-160.