Discharge Coefficients of Ports With Stepped Inlets
Discharge Coefficients of Ports With Stepped Inlets
Discharge Coefficients of Ports With Stepped Inlets
Article
Discharge Coefficients of Ports with Stepped Inlets
Adrian Spencer
Department Aeronautical and Automotive Engineering, Loughborough University,
Loughborough LE11 3TU, UK; [email protected]; Tel.: +44-1509-227-254
Received: 13 August 2018; Accepted: 16 September 2018; Published: 19 September 2018
Abstract: Components of aeronautical gas turbines are increasingly being constructed from two
layers, including a pressure containing skin, which is then protected by a thermal tile. Between
them, pedestals and/or other heat transfer enhancing features are often employed. This results in
air admission ports through the dual skin having a step feature at the inlet. Experimental data have
been captured for stepped ports with a cross flow approach, which show a marked increase of 20% to
25% in discharge coefficient due to inlet step sizes typical of combustion chamber configurations.
In this respect, the step behaves in a fashion comparable to ports with inlet chamfering or radiusing;
the discharge coefficient is increased as a result of a reduction in the size of the vena contracta brought
about by changes to the flow at inlet to the port. Radiused and chamfered ports have been the subject
of previous studies, and empirical correlations exist to predict their discharge coefficient as used in
many one-dimensional flow network tools. A method to predict the discharge coefficient change due
to a step is suggested: converting the effect of the step into an equivalent radius to diameter ratio
available in existing correlation approaches. An additional factor of eccentricity between the hole
in the two skins is also considered. Eccentricity is shown to reduce discharge coefficient by up to
10% for some configurations, which is more pronounced at higher port mass flow ingestion fraction.
Keywords: discharge coefficient; air admission port; orifice; tolerance; eccentricity; step inlet;
counterbore; crossflow
1. Introduction
Ports for metering gas flow are used extensively in gas turbines, for example, in the internal
air system, for blade cooling, and for combustor walls (and of course, there are many none gas
turbine related examples). Despite an increased use of CFD in the design stages of combustors, good
quality experimental data on discharge coefficients is still essential for validating and determining
flow splits between the numerous ports. This is true in the initial design stage where one-dimensional
flow network calculations are most appropriate, because these methods rely heavily on empirical
correlations, and also in complex networks where hundreds of flow paths are possible. Rubini [1]
describes such a network methodology for combustor preliminary design and these methods continue
to be used. With more complex cooling strategies in cooling tiles or turbine blades, similar approaches
are used in cooling circuits, such as within turbine blades as in Kukutla and Prasad [2], for example,
where several hundred discharge coefficients are required to determine the network flow solution.
Even in the detailed design phases, the CFD model requires at least some validation because air
admission ports present their own difficulties to performing accurate predictions with a wide range of
flow scales that require capturing. This is particularly true when sharp edges are present that result
in significant separation and vena contracta formation. Moreover, the ports are often at a small scale
compared with typical CFD grid cell sizes when full systems are modelled, and thus require significant
grid distortion or local refinement to accurately define them. This is even worse if port inlet features
such as radiusing, chamfering, or—as considered here—sharp steps are present, which are usually
present, which are usually another order of magnitude smaller in scale than the port itself. The
another order of magnitude smaller
predicted discharge coefficient, C in scale than the port itself. The predicted discharge coefficient, Cd ,
d, is also found to be sensitive to approach flow conditions that may
is also found to be sensitive to approach flow conditions that may not be well predicted or may be only
not be well predicted or may be only crudely defined as a boundary condition to the calculation, as
crudely defined as a boundary condition to the calculation, as discussed by McGuirk and Spencer [3].
discussed by McGuirk and Spencer [3].
Improving engine efficiency requires reducing parasitic losses. Cooling tiles have thus become
Improving engine efficiency requires reducing parasitic losses. Cooling tiles have thus become
desirable in modern
desirable in modern engine combustors
engine combustors because they require
because less air for
they require wall
less cooling
air thancooling
for wall conventional
than
cooling methods such as z-rings, as outlined in the literature [4]. This has led to a two-skin combustor
conventional cooling methods such as z‐rings, as outlined in the literature [4]. This has led to a two‐
wall construction shown schematically in Figure 1, see the work of [5] for an example of a patent
skin combustor wall construction shown schematically in Figure 1, see the work of [5] for an example
application of this nature. The outer ‘cold’ skin provides the structural integrity of the combustor.
of a patent application of this nature. The outer ‘cold’ skin provides the structural integrity of the
The inner ‘hot’
combustor. The skin, or ‘cooling
inner tile’ provides
‘hot’ skin, the tile’
or ‘cooling thermal barrierthe
provides andthermal
wall cooling mechanisms
barrier and wall between
cooling
the flame and the cold skin. A primary air admission port thus must penetrate
mechanisms between the flame and the cold skin. A primary air admission port thus must penetrate two skins. The outer
skin hole is made larger than the inner skin hole to account for manufacturing and assembly tolerances,
two skins. The outer skin hole is made larger than the inner skin hole to account for manufacturing
as well as differential
and assembly thermal
tolerances, as expansion. Without this
well as differential provision,
thermal an inadvertent
expansion. reduction
Without in open area
this provision, an
could result as the layers become misaligned—leading to a reduction in required
inadvertent reduction in open area could result as the layers become misaligned—leading flow rate. Of more
to a
concern could be the regions of hot gas recirculation behind the resulting ‘fence’ that could occur,
reduction in required flow rate. Of more concern could be the regions of hot gas recirculation behind
leading to premature failure of the cooling tile.
the resulting ‘fence’ that could occur, leading to premature failure of the cooling tile.
Aerospace 2018, 5, 97 3 of 16
fairly insensitive to chamfer depth [6] and chamfer tooling mis-alignment [10], making them attractive
because of the lower manufacturing tolerances required to produce them.
Other parameters, such as port length to diameter ratio, Reynolds number, and proximity to other
ports, also influence the discharge coefficient. These effects have been the subject of previous research
and several empirical correlations exist, but are not investigated in detail here. Interdependence
between the port inlet shape and these other parameters is found to be small, provided that the inlet
feature is small compared with the port diameter and length. The first of these criteria is easily met
within the context of this work, with step sizes typically being an order of magnitude smaller than that
of the port diameter. Port length could be important in terms of the reattachment of flow within the
port. Separation of the flow from a sharp-edged port takes around half a port diameter to reattach.
Once this occurs, pressure recovery within the port leads to an increase in the discharge coefficient.
With inlet shaping, the resulting increase in discharge coefficient implies a reduction in the size of flow
separation at inlet, thus reattachment will occur much sooner. For the stepped ports described without
chutes, it is probable that the flow is marginally reattached and further comment will be given on this.
All these geometrical and crossflow effects on discharge coefficient discussed above have been
summarized by various empirical correlations. Notable ones and those relevant here are McGreehan
and Schotsch [7], Chin et al. [11], Adkins and Gueroui [12], and more recently Fesker et al. [9],
which allow prediction (within typical experimental uncertainty) of a port discharge coefficient to be
determined. Each uses a different methodology and base flow parameters and each has their own
limits of applicability. The effect of stepped inlets on discharge coefficients is currently not reported
in open literature and not included in any available correlation methodology. It is thus difficult to
predict combustor pressure loss without an indication of their effect. Indeed, a recent full-scale test of
a combustor with stepped ports performed very differently than expected because sharp-edged port
discharge coefficients had been assumed. Without hindsight, this is perhaps a reasonable assumption.
The combustor total pressure loss was close to 20% lower than its design value and had weaker jet
penetration than required.
The objective of this work is thus to test stepped holes with crossflow feed for port ingestion
fractions of 30% to 100% (i.e., annulus bleed ratios, α, 0–70%). This will include investigating the effect
of changing the port diameter to feed annulus height and step angle, for example. Benchmarking
the stepped port against plain port data is useful to determine the influence of the step and provide
validation of the experimental techniques. An initial indication of how to include the effect of the inlet
step on the Cd can then be made.
These tests do not consider the effects of compressibility or explore full ranges of geometrical
parameters that are already well captured by existing Cd correlations (such as length to diameter ratio
(L/D), for example). Instead, only two or three values of other main parameters are studied to ensure
stepped port behavior is consistent with these existing correlations and exhibits no significant coupling
between the step feature geometry and other parameters.
Aerospace 2018, 5, x FOR PEER REVIEW 4 of 16
Aerospace 2018, 5, x FOR PEER REVIEW 4 of 16
2.1. Water Flow Facility
2.1. Water Flow Facility
2.1. Water Flow Facility
The test rig is an isothermal, vertically flowing, constant head, water flow rig. Figure 2 shows a
The test rig is an isothermal, vertically flowing, constant head, water flow rig. Figure 2 shows a
schematic of the rig, and it is described in detail in McGuirk and Spencer [15]. The test section has two
The test rig is an isothermal, vertically flowing, constant head, water flow rig. Figure 2 shows a
schematic of the rig, and it is described in detail in McGuirk and Spencer [15]. The test section has
schematic of the rig, and it is described in detail in McGuirk and Spencer [15]. The test section has
inlet flows and two outlet flows that can be controlled by two return flow valves and one inlet valve
two inlet flows and two outlet flows that can be controlled by two return flow valves and one inlet
to two inlet flows and two outlet flows that can be controlled by two return flow valves and one inlet
provide a range of jet Reynolds numbers, jet to downstream cross flow ratios, and annulus bypass
valve to provide a range of jet Reynolds numbers, jet to downstream cross flow ratios, and annulus
flowvalve to provide a range of jet Reynolds numbers, jet to downstream cross flow ratios, and annulus
ratios independently. Setting these three parameters effectively defines the mass flows through
bypass flow ratios independently. Setting these three parameters effectively defines the mass flows
eachbypass flow ratios independently. Setting these three parameters effectively defines the mass flows
path of the test section.
through each path of the test section.
through each path of the test section.
Figure 2. Water flow rig: constant head closed circuit system.
Figure 2. Water flow rig: constant head closed circuit system.
Figure 2. Water flow rig: constant head closed circuit system.
The
The working
working section
section shown
shown inin Figure
Figure 33 isis comprised
comprised ofof two
two circular 600-mm
circular 600‐mm long
long pipes, the
pipes, inner
the
one The
being working
held section
concentric shown
with the in Figure
outer at 3
eachis comprised
end via a of
set two
of circular
six NACA 600‐mm
0015
inner one being held concentric with the outer at each end via a set of six NACA 0015 struts. The long
struts.pipes,
The the
inner
inner one being held concentric with the outer at each end via a set of six NACA 0015 struts. The
pipe contains the ports approximately half way along its length. Wall static pressure
inner pipe contains the ports approximately half way along its length. Wall static pressure tappings tappings were
inner pipe contains the ports approximately half way along its length. Wall static pressure tappings
positioned 70 mm upstream of the ports on the inner and outer surface of the liner to monitor the wall
were positioned 70 mm upstream of the ports on the inner and outer surface of the liner to monitor
were positioned 70 mm upstream of the ports on the inner and outer surface of the liner to monitor
static pressure drop.
the wall static pressure drop.
the wall static pressure drop.
5 mm thick
a) 5 mm thick b)
a) Centrelines of n equi‐spaced ports
b)
20 mm
Centrelines of n equi‐spaced ports
20 mm
100 mm dia.
100 mm dia.
Rig centreline
Rig centreline
Static pressure tappings
Static pressure tappings
Figure 3. Waterflow test section. (a) schematic and (b) photo of inner liner with stepped ports.
Figure 3. Waterflow test section. (a) schematic and (b) photo of inner liner with stepped ports.
Figure 3. Waterflow test section. (a) schematic and (b) photo of inner liner with stepped ports.
Aerospace 2018, 5, 97 5 of 16
Aerospace 2018, 5, x FOR PEER REVIEW 5 of 16
The liner pressure tappings are seen in Figure 3 embedded in the Perspex wall of the inner
The liner pressure tappings are seen in Figure 3 embedded in the Perspex wall of the inner liner,
liner, which has been removed from the test rig in the right-hand photograph. The tapping location
which has been removed from the test rig in the right‐hand photograph. The tapping location was
was positioned upstream to avoid regions of recirculation within the core pipe induced when low
positioned upstream to avoid regions of recirculation within the core pipe induced when low internal
internal crossflow
crossflow to jet velocities
to jet velocities are used.
are used. It has
It has previously
previously been
been investigatedif ifthe
investigated the jet
jet to
to downstream
downstream
crossflow velocity ratio has any effect on discharge coefficients and findings have shown
crossflow velocity ratio has any effect on discharge coefficients and findings have shown the effect to the effect to
be less than 1%, provided the measurement of liner pressure drop is not influenced by the tapping
be less than 1%, provided the measurement of liner pressure drop is not influenced by the tapping
location. The inner liner can be rotated, and this had no effect on the pressure measurements, giving
location. The inner liner can be rotated, and this had no effect on the pressure measurements, giving
confidence in the axisymmetric nature of the approach flows. More details on these aspects can be
confidence in the axisymmetric nature of the approach flows. More details on these aspects can be
found in Spencer [16]. Using water as an essentially incompressible media implies the findings can be
found in Spencer [16]. Using water as an essentially incompressible media implies the findings can
equated to port
be equated to behavior in air in
port behavior at low pressure
air at ratios. The
low pressure behavior
ratios. of compressibility
The behavior is well captured
of compressibility is well
by existing by
captured discharge
existing coefficient
discharge correlations
coefficient through compressibility
correlations corrections, for example,
through compressibility in the
corrections, for
works of [7,9,11], and is thus not considered further.
example, in the works of [7,9,11], and is thus not considered further.
Exhaust
Fan
2.3. Instrumentation
2.3. Instrumentation
It was important to know actual port mass flow and port pressure drops to determine the port
It was important to know actual port mass flow and port pressure drops to determine the port
discharge coefficient. Temperature is recorded to determine viscosity for Reynolds number and density
discharge coefficient. Temperature is recorded to determine viscosity for Reynolds number and
of water is taken as a 1000 kg/m3 .
density of water is taken as a 1000 kg/m3.
Three types of instrumentation were used for this study: orifice plates were used to measure the
Three types of instrumentation were used for this study: orifice plates were used to measure the
flow rates exiting the test section (core and annulus), a 1D laser Doppler anemometer was used to
flow rates exiting the test section (core and annulus), a 1D laser Doppler anemometer was used to
measure the velocity profiles entering the test section (again, core, and annulus), and finally a pressure
measure the velocity profiles entering the test section (again, core, and annulus), and finally a
transducer was used to measure the liner pressure drop. With these measurements, the discharge
pressure transducer was used to measure the liner pressure drop. With these measurements, the
coefficient could be calculated, as will be described in the next section. These devices are described in
discharge coefficient could be calculated, as will be described in the next section. These devices are
a little detail below and summarized in Table 1, but for a fuller description, refer to Spencer [16].
described in a little detail below and summarized in Table 1, but for a fuller description, refer to
Orifice plates were used to measure the return flow from the annulus and core passages of the test
Spencer [16].
section. Designed and made to BS 1042, the uncertainty in the orifice plate’s Cd obtained from the Stolz
Orifice plates were used to measure the return flow from the annulus and core passages of the
equation was always below 2.0% (refer to the standard for further details). The pressure drop across
test section. Designed and made to BS 1042, the uncertainty in the orifice plate’s Cd obtained from the
Stolz equation was always below 2.0% (refer to the standard for further details). The pressure drop
Aerospace 2018, 5, 97 6 of 16
each plate was monitored using an inverted water manometer connected to D and D/2 static pressure
tappings. Typically, each return mass flow rate could be measured to better than ±2.5% over the flow
conditions used for these tests. This has been verified by checking the measured outlet mass flows
against the mass flows obtained by integrating the Laser Doppler Anemometer (LDA) determined
diametral velocity profiles through the annulus and core of a blank test section (i.e., with no ports).
A 1D LDA system was used to measure the axial velocity profiles 150 mm upstream of the ports
in both the annulus and core inflow streams. By assuming the flow at inlet is axis-symmetric, the mass
flow rate into the core and annulus may be obtained by integration of the measured axial velocity
profiles. Within the core, the assumption of axis-symmetry has been shown to be good. However, in the
annulus, a circumferential variation is found because of the six struts used to hold the liner concentric
to the outer casing. It has been established that the effect of the wakes is that the actual mass flow rate
is about 2% lower than if axis-symmetry is assumed. Only three mass flow rates are required to specify
the flow split through the rig, as if the two exit and one of the inlet mass flow rates are known, then by
mass continuity, the unknown inlet mass flow may be calculated. Because the annulus inlet mass flow
rate was determined to have the highest error associated with its measurement, it was decided to use
this purely as a check on the mass flow balance through the test section. It was considered acceptable if
the inferred and measured annulus inlet mass flows agreed to within 3%. Considering the other three
mass flows were measured to within 2.5%, this could be considered a stringent condition, however,
all measurements reported were within this limit. Subsequent calculations requiring knowledge of the
annulus mass flow rate used the value inferred from the other three measured values for consistency.
.
q
m p,ideal = A p 2ρ(( p a − pc ) + q a )
. .
Cd = m p /m p,ideal
and the port to annulus inlet mass flow ratio, α, calculated via the following:
. . .
mp mo − mc
α= . = . . .
ma mo − mc + m a
noting that the annulus inlet mass flow is not used in any calculation to keep the uncertainty associated
with Cd and α minimised for the current experimental setup. Inputs to this calculation were the
inlet axial velocity profiles, the exit orifice plate pressure drops, and the pressure transducer voltage.
Calibration coefficients for each of these then allows the respective inlet, exit, and ideal mass flow rates
to be calculated in kg/s.
Direct comparison was made between the inferred and measured value of annulus inlet mass
flow rate of around 1% to 2% was found, which was typical of most tests, and for all measurements,
Aerospace 2018, 5, 97 7 of 16
the accepted maximum difference between measured and calculated annulus inlet mass flow rate was
3%, as previously stated.
Uncertainties were determined from the contributions from the individual measured value
uncertainties for all measured conditions. In combining the sources of uncertainty, the overall absolute
uncertainty in the discharge coefficient of 4% is calculated with 95% confidence for conditions with
α > 0.5. This is typical of most measurements though, with the highest annulus bleed ratios (i.e., low α,
of less than 0.4), the uncertainty rises to around 6% because of compounding the effect of port mass
flow being derived from the decreasing difference of two mass flows. Taking into account only the
systematic errors indicates that the repeatability of a typical test should be within 2.5%. The level of
scatter within data supports these findings.
The port pressure ratio of the airflow tests was typically around 1.02 to obtain sufficiently high
Reynolds numbers, but stay comfortably within an incompressible regime that has been assumed
within the analysis. Peak through port velocities of the order 50 m/s were obtained. Again, Cd is
defined as the actual to ideal mass flow ratio and assuming constant density locally and using the
calibrated venturi meter (Cd ,v = 0.994) as the reference (actual mass flow), the port Cd was calculated
using the following:
∆pv 1/2
ρv Av
Cd = Cd,v · · ·
ρp Ap ∆p p
Airflow rig results are also at sufficiently high Reynolds numbers (the Cd is quoted as an average
of the tests over the range 6 × 104 and 3 × 105 ) and uncertainty analysis shows the Cd is reported at
better than 3% with 95% certainty.
Figure 5. Parametric stepped port definition.
Figure 5. Parametric stepped port definition.
Table 2. Plain port test configurations, waterflow rig with crossflow.
Table 2. Plain port test configurations, waterflow rig with crossflow.
Dimension Symbol Plain Plain Plain Plain Plain
PortDimension
Diameter (mm) Symbol
D Plain
20 Plain
20 Plain
20 Plain
20 Plain
20
Port Diameter (mm)
Number of Holes nD 620 320 620 320 220
Pitch to Diameter
Number of Holes Z/Dn 2.626 5.243 2.626 5.243 7.852
Annulus Height/D
Pitch to Diameter H/D
Z/D 1.00
2.62 1.00
5.24 0.25
2.62 0.25
5.24 0.257.85
Port Length/D L/D 0.25 0.25 0.25 0.25 0.25
Annulus Height/D
Port/Annulus Area H/D
β 1.00
0.250 1.00 1.143
0.125 0.25 0.571
0.25 0.381
0.25
Port Length/D L/D 0.25 0.25 0.25 0.25 0.25
Port/Annulus Area β 0.250 0.125 1.143 0.571 0.381
Table 3. Stepped port test configurations, waterflow rig with crossflow.
Aerospace 2018, 5, 97
Aerospace 2018, 5, x FOR PEER REVIEW 99 of 16
of 16
3. Results
3. Results
3.1. Plain Port
3.1. Plain Port
A collection of
A collection discharge coefficient
of discharge data measured
coefficient data measured for plain ports
for plain ports (including
(including repeat
repeat tests)
tests) is
is
shown in Figure 6. Two configurations (with the lowest H/D values) show variability of C
shown in Figure 6. Two configurations (with the lowest H/D values) show variability of Cd at high α, d at high α,
which is thought to be because of the formation of an unsteady vortex over a variable number of the
which is thought to be because of the formation of an unsteady vortex over a variable number of the
ports at this high ingestion fraction with low port to annulus area ratio. Otherwise, the scatter of the
ports at this high ingestion fraction with low port to annulus area ratio. Otherwise, the scatter of the
measured C
measured Cdd is generally reasonable and within ±3% as expected from the uncertainty analysis.
is generally reasonable and within ±3% as expected from the uncertainty analysis.
Values
Values ofof
CdCpredicted
d predicted by the correlation of Chin et al. [11] for the plain circular port
by the correlation of Chin et al. [11] for the plain circular port configurations
configurations are
are also shown in Figurealso 6shown in Figure
by the solid 6 by
lines (the the solid lines
configuration they(the
haveconfiguration
been calculatedthey
for ishave been
indicated
calculated for is indicated by unfilled versions of the respective symbol for the experimental data).
by unfilled versions of the respective symbol for the experimental data). The correlation has been
The correlation
further modifiedhas been
to also further the
consider modified
L/D ofto thealso consider
ports using thethe method
L/D of the ports by
outlined using the method
McGreehan and
outlined by McGreehan and Schotsch [8].
Schotsch [8].
Figure 6. Plain port C Variation with alpha, experimental data and empirical correlation.
Figure 6. Plain port Cdd Variation with alpha, experimental data and empirical correlation.
Reasonable but not perfect agreement between the correlations and the measurements can be seen.
Reasonable but not perfect agreement between the correlations and the measurements can be
Two main factors are suspected to account for the discrepancy. Firstly, in the range 0 < L/D < 1, the Cd
seen. Two main factors are suspected to account for the discrepancy. Firstly, in the range 0 < L/D < 1,
of a port is very sensitive to the port’s length. Significant scatter can be seen in the measured Cd ’s of
the Cd of a port is very sensitive to the port’s length. Significant scatter can be seen in the measured
different workers in this range because of the L/D sensitivity (Lichtarowicz [13]). This effect could be
Cd’s of different workers in this range because of the L/D sensitivity (Lichtarowicz [13]). This effect
further compounded by the crossflow inducing non-axisymmetric flow. Secondly, and more important
could be further compounded by the crossflow inducing non‐axisymmetric flow. Secondly, and more
to consider here, is that from experience, some doubt may be cast on Cd correlations as the annulus
important to consider here, is that from experience, some doubt may be cast on Cd correlations as the
height is reduced in comparison to the port diameter. For the plain port geometry, this ratio, H/D,
annulus height is reduced in comparison to the port diameter. For the plain port geometry, this ratio,
was 1.0, and it has been seen for this geometry that unsteady through-port vortices can be formed at
H/D, was 1.0, and it has been seen for this geometry that unsteady through‐port vortices can be
high α (Spencer [16]). The higher levels of scatter in the measurements for α > 0.9 seen in Figure 7
formed at high α (Spencer [16]). The higher levels of scatter in the measurements for α > 0.9 seen in
are thought to be caused by these vortices. To avoid this problem in the subsequent stepped port
Figure 7 are thought to be caused by these vortices. To avoid this problem in the subsequent stepped
measurements, it was decided to scale the ports so that H/D (and all other geometrical ratios) were
port measurements, it was decided to scale the ports so that H/D (and all other geometrical ratios)
higher and more representative of the engine configuration. This was indeed seen to reduce scatter in
were higher and more representative of the engine configuration. This was indeed seen to reduce
subsequent data.
scatter in subsequent data.
Aerospace 2018, 5, 97 10 of 16
Aerospace 2018, 5, x FOR PEER REVIEW 10 of 16
Figure 7. Stepped port Cdd for cases A to E. Chin et al. [11] correlation best fit by adjusting r/D.
Figure 7. Stepped port C for cases A to E. Chin et al. [11] correlation best fit by adjusting r/D.
Aerospace 2018, 5, x FOR PEER REVIEW 11 of 16
Aerospace 2018, 5, 97
Aerospace 2018, 5, x FOR PEER REVIEW 1111 of 16
of 16
Figure 8. Effective r/D value to produce best fit correlation with varying step angle.
8. Effective r/D value to produce best fit correlation with varying step angle.
Figure 8. Effective r/D value to produce best fit correlation with varying step angle.
Figure
3.3. Stepped Ports, Eccentricity
3.3. Stepped Ports, Eccentricity
3.3. Stepped Ports, Eccentricity
It was seen in the previous section that with inlet step features, the variation in Cd with alpha
was seen in the previous section that with inlet step features, the variation in Cd with
ItIt was seen in the previous section that with inlet step features, the variation in C alpha was
d with alpha
was not well captured by empirical correlations at low α, suggesting a different flow direction
not well captured by empirical correlations at low α, suggesting a different flow direction sensitivity
was not well captured by empirical correlations at low α, suggesting a different flow direction
sensitivity of stepped ports compared with plain or radiused ports. As a result, it was considered
of stepped ports compared with plain or radiused ports. As a result, it was considered sensible to
sensitivity of stepped ports compared with plain or radiused ports. As a result, it was considered
sensible to investigate if eccentricity of the holes in the hot and cold skins had an effect. A single
investigate if eccentricity
sensible to investigate if of the holes in
eccentricity of the
the hot andin the
holes cold skins hadcold
hot and an effect. A single
skins had magnitude
an effect. of
A single
magnitude of eccentricity was considered of ε/D = 0.083, but in three directions, as indicated in the
eccentricity was considered of ε/D = 0.083, but in three directions, as indicated in the bottom right of
magnitude of eccentricity was considered of ε/D = 0.083, but in three directions, as indicated in the
bottom right of Figure 9. The correlation of the work of [11] did not fit this data very well in variation
Figure 9. The correlation of the work of [11] did not fit this data very well in variation with α, therefore,
bottom right of Figure 9. The correlation of the work of [11] did not fit this data very well in variation
with α, therefore, to compare the change in C d between the configurations, a best fit quadratic was
to compare the change in Cd between the configurations,
with α, therefore, to compare the change in C a best fit quadratic was fitted to the measured
d between the configurations, a best fit quadratic was
fitted to the measured data. The percentage difference between these smoothed curves and the datum
data. The percentage difference between these smoothed curves and the datum (concentric) case is
fitted to the measured data. The percentage difference between these smoothed curves and the datum
(concentric) case is then shown in Figure 10.
then shown in Figure 10.
(concentric) case is then shown in Figure 10.
Figure 9. Affect of step concentricity with main port on discharge coefficient.
Figure 9. Affect of step concentricity with main port on discharge coefficient.
Figure 9. Affect of step concentricity with main port on discharge coefficient.
Aerospace 2018, 5, x FOR PEER REVIEW 12 of 16
Aerospace 2018, 5, 97 12 of 16
Aerospace 2018, 5, x FOR PEER REVIEW 12 of 16
Figure 10. Sensitivity of change in Cd with alpha for eccentric cases ε1, ε2, and ε3.
Figure 10. Sensitivity of change in Cdd with alpha for eccentric cases ε1, ε2, and ε3.
Figure 10. Sensitivity of change in C with alpha for eccentric cases ε1, ε2, and ε3.
Moving the cold skin upstream has the largest effect on Cd, reducing by 7% to 10% across the
Moving
range the cold skinAt
of α considered. upstream hasmass
high port the largest effect(α),
ingestion Cd , reducing
on moving by 7%skin
the cold to 10% across the range
downstream has a
Moving the cold skin upstream has the largest effect on C d, reducing by 7% to 10% across the
of α considered. At high port mass ingestion (α), moving the cold skin downstream has
reduced effect, from −7% (ecc.1) down to −2% (ecc.2) at α = 1.0. At lower α, both eccentricity directions a reduced
range of α considered. At high port mass ingestion (α), moving the cold skin downstream has a
effect, from −7% (ecc.1) down to −2% (ecc.2) at α = 1.0. At lower α, both eccentricity directions have a
have a similar effect of a 10% reduction. Sideways eccentricity does not appear to be too detrimental
reduced effect, from −7% (ecc.1) down to −2% (ecc.2) at α = 1.0. At lower α, both eccentricity directions
similar effect of a 10% dreduction.
to Cd at high α, but C Sideways eccentricity does not appear to be too detrimental to Cd at
is reduced by 5% at α = 0.4. These figures should be seen in the context of a
have a similar effect of a 10% reduction. Sideways eccentricity does not appear to be too detrimental
high α, but C is reduced
systematic uncertainty in C
d by 5% at α = 0.4. These figures should be seen in the context of a systematic
d of around 2.5%, relevant when considering ratios to the concentric datum
to Cd at high α, but Cd is reduced by 5% at α = 0.4. These figures should be seen in the context of a
uncertainty
case Cd0. in C d of around 2.5%, relevant when considering ratios to the concentric datum case Cd0 .
systematic uncertainty in C d of around 2.5%, relevant when considering ratios to the concentric datum
Cooling tiles are often constructed using additive manufacturing techniques to allow complex
Cooling tiles are often constructed using additive manufacturing techniques to allow complex
case C d0.
features
features to
to be introduced.
be One approach
introduced. One approach to
to avoid
avoid eccentricity effects outlined
eccentricity effects outlined above
above would
would be
be to
to
Cooling tiles are often constructed using additive manufacturing techniques to allow complex
introduce a raised feature to eliminate the effects of the step. This is sketched in Figure 11 with the
introduce a raised feature to eliminate the effects of the step. This is sketched in Figure 11 with the
features to be introduced. One approach to avoid eccentricity effects outlined above would be to
inclusion of an inlet chamfer to help increase the discharge coefficient and help jet directionality.
inclusion of an inlet chamfer to help increase the discharge coefficient and help jet directionality.
introduce a raised feature to eliminate the effects of the step. This is sketched in Figure 11 with the
inclusion of an inlet chamfer to help increase the discharge coefficient and help jet directionality.
Figure 11. Method to avoid inlet step effects on discharge coefficient.
Figure 11. Method to avoid inlet step effects on discharge coefficient.
3.4. Axisymmetric Stepped Ports, Low Step Angles
Figure 11. Method to avoid inlet step effects on discharge coefficient.
3.4. Axisymmetric Stepped Ports, Low Step Angles
To compliment the work described so far, a study was undertaken to compare the effect of a
3.4. Axisymmetric Stepped Ports, Low Step Angles
To compliment the work described so far, a study was undertaken to compare the effect of a step
step inlet feature to a chamfer of similar size. The step angle range, shown in Table 5, was chosen to
inlet feature to a chamfer of similar size. The step angle range, shown in Table 5, was chosen to align
alignTo compliment the work described so far, a study was undertaken to compare the effect of a step
with values typically used when chamfering ports where previous data exists, e.g., [6]. Common
with values angles
typically
areused when
45◦ due chamfering
to ease ports
of tooling, or where
30◦ as itprevious
is known data
thatexists, e.g., [6].
inlet feature to a chamfer of similar size. The step angle range, shown in Table 5, was chosen to align
chamfering this value Common
produces the
chamfering angles are 45° due to ease of tooling, or 30° as it is known that this value produces the
with
highestvalues typically
Cd value, whichused
thewhen
resultschamfering
here have ports where The
confirmed. previous data exists, ports
Cd of chamfered e.g., [6]. Common
is maximised
highest C
theird value, which the results here have confirmed. The C
depth exceeds ~0.08D, beyond this value, increasingd of chamfered ports is maximised once
chamfering angles are 45° due to ease of tooling, or 30° as it is known that this value produces the
once the size of the chamfer has little effect.
their depth exceeds ~0.08D, beyond this value, increasing the size of the chamfer has little effect. Thus,
Thus, in Figure 12, the high Cd values of just above 0.95 at a chamfer angle of 30◦ are constant with t/D.
highest C d value, which the results here have confirmed. The C d of chamfered ports is maximised once
in Figure 12, the high C d values of just above 0.95 at a chamfer angle of 30° are constant with t/D.
from a sharp edge (0◦ ) to 45◦ , a steady increase in Cd is seen for L/D
their depth exceeds ~0.08D, beyond this value, increasing the size of the chamfer has little effect. Thus,
As the step angle increases
As the step angle increases from a sharp edge (0°) to 45°, a steady increase in C
in Figure 12, the high C
of both 0.12 and 0.2. The d is seen for L/D
d values of just above 0.95 at a chamfer angle of 30° are constant with t/D.
shorter port sees a larger percentage increase than the longer port (~25% cf.
of both 0.12 and 0.2. The shorter port sees a larger percentage increase than the longer port (~25% cf.
~10%) As the step angle increases from a sharp edge (0°) to 45°, a steady increase in C
in discharge coefficient compared with the sharp-edged ports (Cd0 ), and the maximum d is seen for L/D
value in
~10%) in discharge coefficient compared with the sharp‐edged ports (Cd0), and the maximum value
of both 0.12 and 0.2. The shorter port sees a larger percentage increase than the longer port (~25% cf.
~10%) in discharge coefficient compared with the sharp‐edged ports (C
d0), and the maximum value
Aerospace 2018, 5, x FOR PEER REVIEW 13 of 16
Aerospace 2018, 5, 97 13 of 16
in this range occurs at 45°. Insensitivity to increased step height of t/D = 0.16 for an L/D of 2.0 is also
seen for two points that map closely to the t/D = 0.08 curve. It would appear the Cd is increased using
this range occurs at 45◦ . Insensitivity to increased step height of t/D = 0.16 for an L/D of 2.0 is also
stepped inlets, with the peak value (at θ = 45°) being about half of the increase obtained by chamfering
seen for two points that map closely to the t/D = 0.08 curve. It would appear the Cd is increased using
at θ = 30°. However, it does appear from this and the previous sections that there is limited sensitivity
stepped inlets, with the peak value (at θ = 45◦ ) being about half of the increase obtained by chamfering
of Cd to step height, as is the case for chamfer depth.
at θ = 30◦ . However, it does appear from this and the previous sections that there is limited sensitivity
of Cd to step height, as is the case for chamfer depth.
Table 5. Axis‐symmetric chamfered and stepped port flow testing configurations, airflow rig.
Dimension
Table Symbol and stepped
5. Axis-symmetric chamfered Chamfer Step configurations,
port flow testing Step airflow Step
rig.
Port Diameter (mm) D 25 25 25 25
Dimension Symbol Chamfer Step Step Step
Port Length/D L/D 2.00 2.00 0.24 2.08
Port Diameter (mm)
Step Height/D t/D D 25
0.08 250.08 25
0.08 25
0.16
Port Length/D L/D 2.00 2.00 0.24 2.08
Step Length/D s/D 0, 0.029, 0.037, 0.046, 0.056, 0.067, 0.080 0.046, 0.080
Step Height/D t/D 0.08 0.08 0.08 0.16
Step Angle θ = tan
Step Length/D(s/t)
−1 θs/D 0°, 20°, 25°, 30°, 35°,40°, 45°
0, 0.029, 0.037, 0.046, 0.056, 0.067, 0.080 30°, 45°
0.046, 0.080
Step Angle θ = tan−1 (s/t) θ 0◦ , 20◦ , 25◦ , 30◦ , 35◦ ,40◦ , 45◦ 30◦ , 45◦
Chamfer
Steps
(a) Absolute values (b) Fractional change compared to sharp edged
Figure 12. Stepped inlet and chamfered inlet comparison for a range of step/chamfer angles.
Figure 12. Stepped inlet and chamfered inlet comparison for a range of step/chamfer angles.
Aerospace 2018, 5, 97 14 of 16
Aerospace 2018, 5, x FOR PEER REVIEW 14 of 16
Figure 13. Effective r/D for all stepped inlet cases and chamfered inlet comparison.
Figure 13. Effective r/D for all stepped inlet cases and chamfered inlet comparison.
4. Conclusions
4. Conclusions
A range of port configurations with stepped inlets have been tested with and without approach
crossA range of port configurations with stepped inlets have been tested with and without approach
flow. Discharge coefficients have been measured and compared to plain port benchmark tests for
cross flow. Discharge coefficients have been measured and compared to plain port benchmark tests
validation purposes.
for validation purposes.
Stepped inlets at entrance to ports produce a significant increase of the port discharge coefficient.
Stepped inlets at entrance to ports produce a significant increase of the port discharge coefficient.
For a stepped port geometry similar to a typical engine combustor primary port, an increase in Cd
For a stepped port geometry similar to a typical engine combustor primary port, an increase in
of 25% over the equivalent sharp-edged port was found (L/D = 0.25, t/D = 0.08).
Cd of 25% over the equivalent sharp‐edged port was found (L/D = 0.25, t/D = 0.08).
To include the effect of the step in empirical correlations, an equivalent r/d factor may be used.
To include the effect of the step in empirical correlations, an equivalent r/d factor may be used.
The equivalent r/d depends primarily on the port step angle (arctan of ratio of step length to height).
The equivalent r/d depends primarily on the port step angle (arctan of ratio of step length to height).
Step height to diameter ratios in the range 0.06 to 0.2 had an insignificant effect on discharge coefficient
Step height to compared
in comparison diameter with
ratios step
in the range
angle, and 0.06 to 0.2 had for
is compensated an in
insignificant effect on discharge
existing correlations.
coefficient in comparison compared with step angle, and is compensated for in existing correlations.
Eccentricity of the two diameters that define the stepped port is a significant factor influencing the
Eccentricity of the two diameters that define the stepped port is a significant factor influencing
overall discharge coefficient. Cd was reduced by up to 10% for the orientations of eccentricity tested
the
andoverall
approachdischarge coefficient. behavior
flow directionality Cd was reduced by up compared
was changed to 10% for the concentric
with orientations of features.
inlet eccentricity
tested and approach flow directionality
◦ ◦ behavior was changed compared with concentric
Step angles of around 45 to 50 appear to produce the highest Cd , though further work is required inlet
features.
to verify this because this value fell between the two testing configurations.
Step angles of around 45° to 50° appear to produce the highest Cd, though further work is
Funding: This research was funded in part by Rolls Royce plc.
required to verify this because this value fell between the two testing configurations.
Conflicts of Interest: The author declares no conflict of interest.
Funding: This research was funded in part by Rolls Royce plc.
Nomenclature
Conflicts of Interest: The author declares no conflict of interest.
Symbol
Nomenclature
A Area
Cd Discharge Coefficient
.Symbol
m Mass flow rate
A Area
pCd Static pressure
Discharge Coefficient
q Dynamic pressure
Mass flow rate
Vp Velocity
Static pressure
αq Port ingestion fraction
Dynamic pressure
V Velocity
Aerospace 2018, 5, 97 15 of 16
Appendix A
The correlations of Chin et al. [11] have been fitted to the experimental data presented in this
paper in most figures associated with the crossflow approach cases. This correlation accounts for
crossflow effects, compressibility, and annulus to port area ratios, among other parameters. Interested
readers may refer to this paper for further details of the full correlation. To compensate for length
to diameter ratio (L/D) and effective radius to diameter ratio (r/D), the method of McGreehan and
Schotsch [8] has been used. Here, the baseline (sharp edged, thin port) discharge coefficient (Cd0 ) is
corrected for inlet radiusing via the following:
Cd:r = 1 − f (1 − Cd0 )
where
r r 2
f = 0.008 + 0.992e−5.5( D )−3.5( D )
where
L 2 L
g = [1 + 1.3e−1.606( D ) ](0.435 + 0.021 )
D
When both L/D and r/D effects are taken into account, compensation is made by reducing the
effective L/D by the r/D value used, and making a cross compensation for Cd:r , not explained fully
here for brevity. These formulae have been used to calculate an effective r/D that best matches the
stepped or chamfered port behavior. ‘r/D’ in the f equation has thus been changed to minimize the
rms error in fitting the correlation of the literature [11] (taken as the baseline ‘Cd0 ’ value) to the data for
cross flow cases. For axis-symmetric flow, Cd0 has been taken to be the sharp-edged value (no step/no
chamfer case) for the same L/D port. ‘r/D’ has then been changed to obtain the best fit to experimental
values of Cd .
References
1. Stuttaford, P.J.; Rubini, P.A. Preliminary Gas Turbine Design Using a Network Approach. ASME J. Eng. Gas
Turbines Power 1997, 119, 546–552. [CrossRef]
2. Kukutla, P.R.K.; Prasad, B.V.S.S.S. Network analysis of a coolant flow performance for the combined
impingement and film cooled first-stage of high pressure gas turbine nozzle guide vane. Proc. Inst. Mech.
Eng. Part G 2018. [CrossRef]
3. McGuirk, J.J.; Spencer, A. Coupled and Uncoupled CFD Prediction of the Characteristics of Jets from
Combustor Air Admission Ports. ASME J. Eng. Gas Turbines Power 2001, 123, 327–332. [CrossRef]
Aerospace 2018, 5, 97 16 of 16
4. Lefebvre, A.H. Gas Turbine Combustion, 3rd ed.; CRC Press: Boca Raton, FL, USA, 2008; ISBN 9781420086058.
5. Pidcock, A.; Close, D. Double Wall Combustor Tile Arrangement. U.S. Patent 2003/0145604 A1, 7 August 2003.
6. Hay, N.; Spencer, A. Discharge Coefficients of Cooling Holes with Radiused and Chamfered Inlets.
J. Turbomach. 1992, 114, 701–706. [CrossRef]
7. Dittmann, M.; Dullenkopf, K.; Wittig, S. Discharge Coefficient of Rotating Short Orifices with Radiused and
Chamfered Inlets; ASME Paper No. GT2003-38314; ASME: New York, NY, USA, 2003.
8. McGreehan, W.F.; Schotsch, M.J. Flow Characteristics of Long Orifices with Rotation and Corner Radiusing.
ASME J. Turbomach. 1988, 110, 213–217. [CrossRef]
9. Feseker, D.; Kinell, M.; Neef, M. Experimental Study on Pressure Losses in Circular Orifices with Inlet Cross
Flow. ASME J. Turbomach. 2018, 140. [CrossRef]
10. Chernukha, P.; Spencer, A.; Colwill, J.A. The Effect of Manufacturing Tolerances on the Performance of Gas
Turbine Air System Metering Holes with Chamfered Inlets. In Proceedings of the ASME Turbo Expo 2018:
Turbomachinery Technical Conference and Exposition, Oslo, Norway, 11–15 June 2018.
11. Chin, J.; Shi, X.; Zhou, K. Analysis and Experimental Study on the Discharge Coefficient of Liner Holes. Int. J.
Turbo Jet Engines 1993, 10, 97–106. [CrossRef]
12. Adkins, R.C.; Gueroui, D.D. An Improved Method for Accurate Prediction of Mass Flows through Combustor
Liner Holes. ASME J. Eng. Gas Turbines Power 1986, 108, 491–497. [CrossRef]
13. Lichtarowicz, A.; Duggins, R.K.; Markland, E. Discharge Coefficients for Incompressible Non-Cavitating
Flow through Long Orifices. J. Mech. Eng. Sci. 1965, 7, 210–219. [CrossRef]
14. Deckker, B.E.L.; Chang, Y.F. An Investigation of Steady Compressible Flow through Thick Orifices. Proc. Inst.
Mech. Eng. 1965, 180, 312–323. [CrossRef]
15. Spencer, A.; McGuirk, J.J. LDA Measurements of Feed Annulus Effects on Combustor Liner Port Flows.
ASME J. Fluids Eng. 2001, 123, 219–227. [CrossRef]
16. Spencer, A. Gas Turbine Combustor Port Flows. Ph.D. Thesis, Loughborough University, Loughborough,
UK, 1998.
© 2018 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).