Professional Documents
Culture Documents
Summary
his document describes the design, construction and subsequent modification of a vertical wind tunnel located at CSIRO Forestry and Forest Products, Yarralumla, ACT. This facility enabled the first study of the aerodynamic and combustion properties of eucalyptus bark as
firebrands.
The 20 m long wind tunnel had a variable-speed fan, horizontal divergence section, and a
turning section below a contractor and a vertical tapered working section. The working section was 4250 mm long and 770 mm H 770 mm in cross-section at its base, diverging to 1000
mm H 1000 mm at the exit. The air velocities within the working section could vary between
0 and 20 m s-1, and had a variation in the uniform section of the air flow of approximately
1% (P = 0.05) of the mean.
Two features in wind tunnel design facilitated the study of firebrands. The vertical working
section had a divergent taper that allowed a firebrand to find its terminal velocity within a velocity gradient. The second feature was the modification of the air flow in the working section
that reduced impact of non-tethered firebrands with the walls of the working section. A series
of screens at the base of the working section resulted in relatively high air velocity adjacent to
the walls, and relatively low air velocity in the centre zone. These features, and the fan speed
control, allowed observation of burning firebrands throughout their viable lifetime.
The centre zone was square in cross-section and was 300 mm H 300 mm at the base of the
working section. The velocity in this zone could reach up to 10 m s-1 and had a variation of
about 4% (P = 0.05) of the mean air velocity across the zone. The air flow in the centre
zone was calibrated so that the air velocity at any location could be derived as a function of
contractor differential pressure (CDP) and the height of the location above the base of the
working section. It was assumed that when there was no apparent vertical movement of a
bark sample, its terminal velocity equalled the air velocity. At such times, the CDP and
height of the sample were recorded.
iii
iv
Contents
Summary ........................................................................................................................iii
Table of Contents ...........................................................................................................v
List of Figures ................................................................................................................vi
List of Plates .................................................................................................................vii
1 General Design............................................................................................................. 1
1.1 Principles of wind tunnel design......................................................................... 1
1.2 Previous firebrand studies using wind tunnels ................................................... 1
1.3 CSIRO design requirements and constraints......................................................2
2 Design and construction of components......................................................................5
2.1 General ..............................................................................................................5
2.2 The fan..............................................................................................................5
2.3 The diffuser.......................................................................................................6
2.4 The screens .......................................................................................................7
2.5 The turning section ...........................................................................................9
2.6 Straightening section ........................................................................................9
2.7 Contractor section ........................................................................................... 10
2.8 Adaptor ........................................................................................................... 15
2.9 Working section .............................................................................................. 16
2.10 The wind tunnel housing ............................................................................... 19
3 Instrumentation ......................................................................................................... 21
3.1 Air speed control ............................................................................................. 21
3.2 Contractor differential pressure ....................................................................... 21
3.3 Working section pitot tube ..............................................................................22
3.4 Pressure transducers and control board ...........................................................23
4 The unmodified air flow in the working section ........................................................24
4.1 The variation in air flow measured for a central transect ................................24
4.2 Comparing the air flow along a central transect with that of the surrounds ....27
4.3 Conclusion ......................................................................................................29
5 Modification of the air flow .......................................................................................30
6 Sources of variation in the air flow ............................................................................33
6.1 The variation in contractor differential pressure ..............................................33
6.2 The variation in pitot differential pressure of the unmodified air flow ...........33
6.3 The variation in pitot differential pressure (PDP) of the modified air flow .....34
7 Calibration of the modified air flow ...........................................................................35
8 Discussion and conclusions .......................................................................................38
9 Acknowledgements....................................................................................................40
10 References ................................................................................................................ 41
Appendix ...................................................................................................................... 41
v
vi
General Design
ll modern wind tunnels that produce a highly uniform flow of air operate on the same
fundamental design principles. Highly turbulent air from the fan is decelerated in a widening
duct called the diffuser. This slowed air is then passed through screens which moderate the velocity extremes produced by the fan.
The air is then passed through a straightening section which straightens the flow by removing
large diameter eddies. This air, now moving slowly with a degree of uniformity in the widest cross-section of the tunnel, then enters the contractor. The contractor constricts the air
flow, forcing the air to accelerate. Energy is added equally to each parcel of air. Variations
in energy (mean or turbulent) of the in-going air stream remain but are an insignificant fraction
of the accelerated flow at the exit. The degree of curvature of the contractor walls is critical;
the flow must be accelerated smoothly because any separation of the flow from the contractor
walls will produce undesirable eddies in the final flow. The contractor is the final step of air
flow preparation, reducing variation in the flow to values as low as 0.5% of mean air velocity.
The uniform accelerated air flow then enters the working section where the aerodynamic experiments are performed.
Most wind tunnels have the critical components of the tunnel aligned on one axis because
bends introduce turbulence and vorticity. When a bend in the tunnel is needed, the tunnel is
normally designed so that the bend is well upstream of the working section. This allows maximum time for unwanted vorticity to dissipate. To reduce the height of the CSIRO vertical
wind tunnel (and the cost of the building required to house it), a turning section was installed
just upstream of the straightening section.
Figure 1.1.
The construction of the CSIRO vertical wind tunnel
was supported by Mayne Nickless Ltd. The facility
was opened by Dr Malcom McIntosh, Chief
Executive of CSIRO, on January 17 1997.
Housing
Screens
Adaptor Section
Straightening
Section
Garage
Contractor
Diffuser
Fan
Turning
Section
4.8 m
2.65 m
Figure 1.2
The general layout and components of the CSIRO vertical working section wind tunnel. The tunnel
is housed in a tower constructed of clad scaffolding.
General Design
struct. A 1/8 scale prototype was built that incorporated a 90E turn immediately after the diffuser and before the straightening section. This enabled the fan and diffuser to be aligned on
a horizontal axis. The prototype performed well and this suggested that the height of a full
scale tunnel would only need to be in the order of 12 metres.
Because burning firebrand samples would lose volume and mass with time, there was a need
to be able to adjust the speed of the air flow within the working section. This would enable
the air flow velocity to be matched with the changing terminal velocity of the samples as they
burnt. In order to further stabilise the vertical movement of the firebrand, the working section was designed with a divergent taper. This taper produced a gradient of air flow velocity
in which the non-tethered sample could settle at a location where the air velocity equalled its
terminal velocity.
The problem of limiting the horizontal component of movement of non-tethered embers to
minimise or eliminate contact with the walls (Clements 1977) was identified but not solved at
the design stage.
All wind tunnels are built to engineering, spatial, local government, and financial constraints.
As a result, each tunnel is unique, even if the experimental requirements are similar. The upper limit on cost was A$100,000; engineering and performance criteria were constrained by
this budget. A schematic of the final design of the CSIRO vertical wind tunnel is shown in
Figure 1.2. The components and modules of the wind tunnel are described in detail in Section 2.
Figure 1.3
The external view of the completed CSIRO vertical wind tunnel.
2.1 General
he components of the wind tunnel were square in cross-section and fabricated from
wood. They were designed and built in a modular manner in order to facilitate on-site
construction. The diffuser, turning, and working sections were all made from 12 mm form ply
(a wax-treated plywood used to form concrete in the construction industry) and framed or
braced with 90 mm H 35 mm section timber. All modules were fitted with 90 mm H 35 mm
external flanges of timber so they could be bolted together. The contractor was constructed
from laminated 3 mm sheets of high-density particle board. The diffuser screens and turning
vanes were constructed from sheet steel, and the straightening section was built of strips of
hollow-section poly-carbonate roofing material.
The expected pressure losses were calculated for air velocities corresponding to a volume flow
rate of 10 m3 s-1.
It was envisaged that up to 5 screens might be installed in order to obtain an acceptable quality air flow: These would produce a pressure loss of around 200 Pa (see section 2.4). The pressure loss due to the straightening section was estimated to be in the order of 2 Pa (see section
2.6). The exit loss was estimated to be less than 70 Pa (see section 2.9). The total pressure
loss was estimated to be less than 300 Pa.
Figure 2.2
Fan motor, fan and diffuser sections of the vertical wind tunnel. These
sections are housed in a pre-fabricated steel garage.
The speed control panel allowed fine-step adjustment of the fan speed. This device was
mounted adjacent to the working section for ease of reach. Fan vibration was isolated from
the diffuser and the rest of the tunnel by physically separating the fan from the diffuser by
about 100 mm. A rubber connector 1250 mm in diameter and 500 mm long was fabricated to
bridge and seal this gap. The intake of the fan and the diffuser sections are shown in Figures
2.1 and 2.2.
Figure 2.3
Dimensions of diffuser modules and location of screens. The modules are constructed from 12 mm
plywood. The modules are framed and joined using external flanges of 90 mm x 35 mm section pinewood.
1
kv 2
2
(2.2)
where P in Pa is the change in pressure across the screen calculated as upstream static pressure (P U) minus downstream static pressure (P D ), is the density of air (1.27 kg m-3), k is the
Design and Construction
coefficient of resistance for the screen, and v is the velocity of air (m s-1) passing through each
screen.
The pressure losses were calculated assuming that the air flow through each screen was 10 m3
s-1 (see section 2.2). The values for v for screens #1 and #2 were 10/(1.218 H 1.358) m s-1 and
10/(1.980 H 1.980) m s-1 respectively. The pressures drop for screens #1 and #2, calculated
using Equation 2, were 22.95 and 7.79 Pa respectively.
The expected pressure drop in the diffuser is the sum of the Bernoulli pressure rise due to the
deceleration of the divergent flow, calculated using Equation 2.3, and the pressure drop due
to the resistance of the two screens (Equation 2.2).
P = P1 P2 =
1
2
2
v 2 v1
2
(2.3)
The static pressure drop over the length of the diffuser was measured using a differential pressure transducer (see section 3) connected to take-off points mounted in the inside walls. At a
fan speed that delivers 3.5 m3 s-1 (32% of maximum fan speed) the static pressure drop was
4.5 Pa. This was about 75% greater than predicted. This difference was attributed to the
use of screens of sheet metal with punched holes rather than woven wire.
Figure 2.3
Schematic diagram of turning section. This houses steel vanes which turn the air flow through
90 degrees.
8 The CSIRO Vertical Wind Tunnel
Figure 2.5
The top of the straightening section which is housed in a module 600
mm deep. Two pressure take-off points are located in each wall of this
module. Two of these points and the base of the contractor are visible.
The straightening section had an open space ratio of 80% (viz. b = 0.8) and hence the coefficient k (Equation 2.1) was 0.31. The calculated pressure drop across the straightening section (Equation 2.2) was 1.24 Pa (assuming that the resistance of the straightening tubes was
comparable to the resistance of a grid of the same open space ratio). Figure 2.5 shows the
inside of the base of the contractor, featuring the top of the straightening module and two
pressure take-off points.
The straightening section was mounted in a 600 mm deep module located above the turning
section.
known as the >inflection point=, w p, of the contractor profile. Borger gives a series of
tabulations as a function of the required contraction ratio which allow the wind tunnel
designer to determine the precise values for the two sets of parameters (y'0, xw and yw) that give
the optimally short contractor profile.
The design of the CSIRO vertical wind tunnel required a contraction area ratio of 7.11:1.
That is, the cross-section of the contractor reduced from 2000 mm H 2000 mm at the inlet to
750 mm H 750 mm at the outlet to the working section. This ratio is adequate for high
quality air flow (Rae and Pope 1984). Borger gives a series of charts that specify parameters
which describe the contractor proportions as a function of E, the square root of the inverse
of the contractor ratio. In the case of the CSIRO tunnel, E = 0.375.
xw =
y
15
xw - w
7
y0
(2.4)
where xw and yw are parameters given by Borger in charts as a function of E. Parameter y'0 is
the gradient common to the two curves and is given by the function :
y 0 =
ws wh
wl
(2.5)
where w s, w h and w l are the physical parameters describing the inlet characteristics determined
by Borger as a function of E.
The value for x$w is then substituted in Equation 2.6:
y=3
15yw
yw 5
y
- 5 w3 x3 +
5 x
8 xw
4 xw
8 xw
(2.6)
to generate a series of points (x$ , y), where x$ is a variable, 0 = x$ = x$ w. By choosing an appropriate interval value for a range of values of x$, within the range 0 to x$ w , Equation 2.6 results in a series of y values. These are used to determine corresponding x values using Equation 2.7 :
x = x - y(
8 xw
1
)
15 y w y0
(2.7)
The resultant coordinate pairs (x, y) give the curve of the contractor profile.
Design and Construction
11
y
0.6
(X w ,Yw )inlet
Gradient y' 0
0.4
0.2
(X w ,Yw )exit
0
0.2
0.4
0.6
0.8
1.0
1.2
-0.2
-0.4
Figure 2.6
Resultant curves for the CSIRO vertical wind tunnel obtained from equations 4 - 7. Starting coordinates
for the inlet curve, (xw, yw)inlet, and exit curve, (xw, yw)exit, were obtained from Borger=s charts of contractor
parameters as a function of E, the normalised half-width of the contractor exit. The common gradient,
yN0 = 1.348, is also obtained from values Borger gives as a function of E. The curve of (xw, yw)exit was
mirrored around both axes to produce the overall shape of the contractor (-xw, -yw)exit to (xw, yw)inlet.
half-width of the exit, was 375 mm and all values calculated for x and y could also be taken as
in millimetres if multiplied by 1000. That is, for the entrance curve xw = 1080 mm and yw =
552 mm and for the exit curve xw = 490 mm and yw = 73 mm. When these two curves were
spliced, an inflecting curve was produced for one side of the contractor where y contracted
625 mm over a distance (x) of 1570 mm.
If the proportions had not been so serendipitous, then the values calculated by Equations 2.6
and 2.7 would have to be scaled appropriately.
1.2
1
Inlet
0.8
0.6
0.4
width (m)
Exit
0.2
0
0
0.25
0.5
0.75
1.25
-0.2
1.5
1.75
length (m)
-0.4
-0.6
-0.8
-1
-1.2
Figure 2.7
The scaled, stretched (+ 30%) and joined curves that resulted for the profile of the CSIRO
vertical wind tunnel contractor. Two sides of the contractor profile are shown, mirrored
around the centre-axis of the contractor. Note that the half-width of the exit of the con-
13
formulated a slight divergence at the exit of the contractor to compensate for boundary layer
buildup which acts to constrict the outlet. This refinement was not incorporated into the
CSIRO contractor because the adaptor and working sections were designed with divergence.
Figure 2.8
The contractor is located on top of the straightening module. It is constructed from highdensity particle board coated with fibre glass to improve strength and hold the critical shape.
1
( v 21 - v 22 )
2
(2.8)
where v 1 and v 2 are the air velocities at the exit and entrance of the contractor respectively,
calculated for an air flow of 10 m3 s-1. In the CSIRO vertical wind tunnel this gives a pressure
drop of 196.7 Pa.
In theory, the energy of each parcel of air which enters the contractor is increased by exactly
196.7 J m-3, so the final kinetic energy is 196.7 + 4.0 " 1.8 J m-3. The initial variation in energy is conserved, but is now a much smaller fraction of the total kinetic energy. Kinetic energy at the exit of 200.7 " 1.8 J m-3 corresponds to a velocity of 17.78 " 0.08 m s-1 which is a
variation of " 0.45%. Thus, a variation in air velocity of 20% at the entrance of the contractor is reduced to a variation of 0.45% at the exit.
In practice, other factors decrease this theoretical gain. A boundary layer builds up against
the walls of all sections and increases in thickness with distance. Manufacturing tolerances
can also reduce this theoretical gain. Departure from axial symmetry will bias the air stream
and imperfect joins between sections can accelerate boundary layer growth. Some residual
turbulence can remain from the straightener and the parallel flow criteria cannot be exactly
met in practice.
15
Figure 2.9
The adaptor section is located at the top of the contractor and below the working section.
Three pressure take-off points are visible. Aluminium brackets are attached to the
adaptor to relocate the working section following maintenance of the screens.
=
x
16 The CSIRO Vertical Wind Tunnel
0 . 3747
R
0 .2
e
(2.9)
Figure 2.10
Schematic diagram of the working section. This section was constructed in two
modules. Two adjacent walls in each
module contain full-length perspex observation windows. Each module has two
full-length unhinged access doors. The
bottom module also has a small hinged
door through which firebrands are
launched. (All measurements in figure are
in mm)
Exit
where d is the boundary layer thickness, and x is the distance in the direction of flow. The Reynolds Number is given by Re = LV/v where L is
the length of the working section, V
is the velocity in the working section,
and v the kinematic viscosity.
2125
Entrance
17
Figure 2.11
The lower half of the working section. Perspex panels enable observation of firebrands as they burn in the air flow. The access hatch is just visible on the right rear
wall of the working section.
Figure 2.12
Access door to the lower half of the working
section. Firebrands are lit and then placed in
the working section through this door.
Figure 2.13
The view through the protective screen at the bottom of the working section to the top of the adaptor section below.
Figure 2.14
The lower half of the working section, the fan control (1) and control board (2). The readout unit (3) for
contractor and pitot differential pressures is mounted above the access door (4) (seen open here).
Firebrand samples are weighed and ignited at workbenches which are visible behind the working section.
Figure 2.15
The internal scaffolding used to build the tower
that houses the vertical components of the
wind tunnel.
Design and Construction
19
Figure 2.17
The housing for the wind tunnel comprises steel cladding fastened to hardwood battens. These in turn are attached to the
scaffold frame. The cladding is standard Colorbond corrugated sheet steel.
20 The CSIRO Vertical Wind Tunnel
3 Instrumentation
he air velocity in the wind tunnel was adjusted by means of a fan speed control box supplied with the fan. This was located next to the
working section for easy access. The fan speed
was determined by the frequency of the alternating
current supplied by the controller and was based
on the motor=s synchronous speed. The maximum
fan speed was associated with 50 Hz and lesser fan
speeds associated with relatively lower frequencies. However, the fan speed controller only supplied power to the fan at discrete frequencies. The
range of frequencies supplied to power the fan, albeit closely spaced, were not continuous. Hence,
the air velocities produced by the fan were also not
continuous but discrete.
Figure 3.1
The fan control box located next to the lower
half of the working section where firebrands
are launched through the access hatch.
21
Static pressure take-off points in the contractor walls (Figure 3.2) were 20 mm diameter
threaded brass plugs, each with a central 3 mm diameter hole, fitted into the walls and sanded
flush with the inside surface.
Eight such points were installed at the inlet of the contractor (high pressure) and six at the
exit of the contractor (low pressure). Each set of take-off points was joined using 3 mm
diameter flexible tubing. The two flexible tubes (viz. high and low pressure) terminate at a
control board on which the pressure transducers are mounted.
Figure 3.3
The bent shaft pitot tube (1) can be traversed remotely on its support beam (2)
to any required position with an accuracy of " 1 mm using a scale mounted
on the support beam. The low (3) and high (4) pressure tubes that lead to the
control board are identified.
22 The CSIRO Vertical Wind Tunnel
Figure 3.4
The control board supports the stopcocks. These allow the pressure lines from either the pitot
tube or low and high pressure (LP and HP) take-off points from the contractor to be connected to
one of two pressure transducers (a and b). Output from either transducer is displayed on the
read-out unit located in the working section. The pitot tube pressure was only measured during
calibration of the wind tunnel. Contractor Differential Pressure was monitored continuously during
experiments.
Instrumentation
23
he air flow in the working section was measured using the Pitot Differential Pressure
(PDP) along transects perpendicular to the flow. The transects revealed two zones:
the boundary layer (BL) where the air velocity increases from zero near the wall to a
maximum at some distance from the wall;
a central uniform zone (UZ) where the mean velocity is equal to the maximum air
velocity in the boundary layer and where the variation in air velocity should be <
0.5% (see section 2.7).
180
160
Height above base of
working section
140
100 mm
1500 mm
4100 mm
120
100
80
60
40
20
0
-600
-400
-200
0
Distance from centre (mm)
200
400
600
Figure 4.1
Half-transects of PDP (Pa) across the working section for CDP of 180 Pa at heights of 100 mm,
The width of the North and South boundary layers and the mean values and standard deviation of the corrected PDP value of the Uniform Zone are shown in Table 1. The speed of the
air flow across the North and South half-transects was different, particularly in the boundary
layer. The mean PDP values for each half-transect were compared using a standard >t= test.
The means and the probability (0.05) of such a difference occurring are shown in Table 1.
4.1.2.1 The width and shape of the boundary layer
The width of the boundary layer at the top of the working section was predicted to be approximately 83 mm (section 2.9) but was in fact about 330 mm on the North wall (Table 4.1).
This difference is attributed to the divergence of the working section. The boundary layer on
each side had the same width at a height of 100 mm, but differed in width and shape at other
heights (Figure 4.1). This asymmetry was most likely due to the effect of imperfect joins between the contractor, the adaptor, and the working section and its modules. Ideally, the join
should be smooth but minor irregularities could produce asymmetry in the air flow and accelerate the growth of the boundary layer.
25
Table 1. The width of the boundary layers on the north and south walls, the mean PDP of the Uniform Zone (x
$ (UZ)), and the standard deviation of this mean (s(UZ)) from measurements made along
a central N-S transect at three heights above the bottom of the working section. The mean values
for PDP for the North and South halves of the transects (x
$ (UZ-N)) and (x
$ (UZ-S)) and the probability
(P) of this difference are also given.
1500
4100
35
160
330
35
85
200
164.3
117.0
68.4
0.9
0.5
0.9
163.9
116.8
67.4
164.6
117.2
68.9
0.001
0.03
< 0.001
(4.1)
where the values for x$(UZ) and s(UZ) are the mean and standard deviation of the PDP for the
uniform zone (Table 1).
The variation in air velocity along the central transect within the uniform zone at 100, 1500
and 4100 mm above the bottom of the working section was approximately "0.51%, "0.45%
and "1.32% of the mean respectively. It was anticipated that measurements of firebrand terminal velocity would be performed in the lower quarter of the working section where the
variation was less than "0.5%. This error was in the order of the error in air flow predicted
upstream of the contractor (see section 2.7), and was considered acceptable for the purposes of
measuring the terminal velocities of firebrand samples.
26 The CSIRO Vertical Wind Tunnel
| x1 - x2| 100
_
1
x1
(4.2)
where /x$1 and /x$2 are the square root of the mean PDP for the North and South half-transect
respectively. The difference in mean air velocity at heights of 100, 1500 and 4100 mm were
0.25, 0.15 and 1.14% respectively. The difference was possibly the result of non-vertical
alignment of the working section and its modules. The effect of a difference of this magnitude was considered insignificant when measuring terminal velocities of firebrand samples.
4.2 Comparing the air flow along a central transect with that
of the surrounds
4.2.1 Method
Measurements of PDP were made at two heights, 100 and 1500 mm above the bottom of the
working section. The pitot tube was rotated around four selected positions within the uniform zone to measure the variation of air flow around each position as well as to determine if
the carriage of the pitot tube itself interfered with the measurement of the air flow. These positions were 225, 250, 275 and 300 mm from the South wall along the N-S transect. The order in which each position was measured was randomly selected.
At each position the pitot tube was rotated to 12 points. These points, which were also selected in random order, were separated by 30E of arc and included two positions directly under the metal guide of the carriage. Measurements directly under the guide were classed as
centre (C) and those away from the guide surround (Sr). The surround class was further subdivided into surround east (Sr-E) and surround west (Sr-W).
Values for corrected PDP were obtained using the method described in section 4.1.
27
Table 4.2. The difference in PDP between centre of transect and surrounds within the
uniform zone at two different heights.
1500
164.44
117
1.92
0.76
168.49
119.24
2.07
1.22
P1
< 0.001
< 0.001
(Sr-E) (Pa)
169.23
119.18
(Sr-W) (Pa)
167.75
119.18
0.01
NA
x$(C) (Pa)
s(C)
x$(Sr) (Pa)
s(Sr)
P2
Table Notes: P1 describes the probability (0.05) of the difference between the
mean PDP for centre (x
$ (C)) and surrounds (x
$ (Sr)). Similarly P2 describes the
probability of the difference between mean PDP for surround east ((Sr-E)) and
4.2 into Equation 4.1. These values were "1.14% and "0.6% of the mean PDPs at heights
of 100 mm and 1500 mm respectively. The corresponding values obtained in section 5.1 for
heights of 100 mm and 1500 mm were " 0.51% and " 0.45% respectively.
The action of a divergent section is the reverse of a contractor. We would expect any variations in mean wind velocity at 100 mm to be amplified as the section diverges and the velocity is reduced. However, the total absolute error in the kinetic energy in the system is constant. That is, the pressure gradient of the divergent working section subtracts kinetic energy
from all parcels of air equally without altering the absolute variation of kinetic energy of the
parcels. By utilising the constant percentage error in the kinetic energy of the system, the percentage error in the velocity at any height can be determined.
At a height of 1500 mm, the increase in the error of the velocity was in the order of 60%,
bringing the absolute error in the flow to 0.8% of the mean from an original 0.5% at the start
of the working section. The reduction in the air velocity at this height was 18%.
The variation in velocity was greater than that predicted from the tunnel geometry. A proportion of this variation may be due to factors such as instrument drift, drift in the fan speed, and
pressure fluctuations in ambient weather (e.g. due to air temperature, relative humidity, wind
speed (see section 6)).
4.2.2.2 The variation in air flow within the surrounds of the Uniform Zone
The variation in air velocity within the surrounds section of the Uniform Zone at the two
heights was calculated by substituting values for x$(Sr) and s(Sr) from Table 4.2 into Equation
4.1. The resultant values were " 1.20% and " 1.00% of the mean air velocity at 100 mm and
1500 mm height respectively.
28 The CSIRO Vertical Wind Tunnel
4.3 Conclusion
The boundary layer in the working section was asymmetrical along a North-South transect,
and attained a width near the exit approximately three times the width predicted. These aberrations were attributed to non-vertical alignment of the working section and upstream irregularities in the joins between modules of the wind tunnel.
The variation in air velocity at heights of 100 and 1500 mm was also described for central
North-South transects underneath the pitot tube support bar and for the surrounding area
within the uniform zone. The measured variation of the air flow within the uniform zone
along North-South transects and within the surrounding area at the two heights was between
" 0.5% and " 1.3 % of the mean air velocity. This error was considered to be acceptable for
the purpose of measuring terminal velocities of firebrand samples.
In addition, the mean air velocity derived from measurements underneath the pitot support
bar underestimated the mean air velocity of the surrounding area within the uniform zone by
1%. The calibration of the wind tunnel (see section 7) related measurements of the mean
PDP under the pitot support beam to measurements of contractor differential pressure (CDP).
This calibration was necessary because measurements of PDP within the uniform zone could
not be made using a pitot tube while an experiment was in progress. However, the CDP
could be constantly and easily monitored. The effect of the support beam during calibration
meant that air velocities derived using measurements of CDP would underestimate the actual
air velocity in the uniform zone by 1%.
29
efore experiments were conducted, it was necessary to install a primary screen at the
base of the working section directly on top of the adaptor section. The purpose of this screen
was to prevent burning samples falling into the contractor and straightening sections. The
screen material was standard aluminum insect mesh with square apertures of about 1.8 mm H
1.8 mm. It was anticipated that the turbulence introduced by this screen would dissipate
within 200 mm above the screen (Bradshaw and Pankhurst 1962).
The firebrand study required the bark samples to remain untethered and free to move during
combustion at their terminal velocity. However, when the tunnel was initially tested the
combusting samples of bark of Eucalyptus obliqua had frequent collisions with the walls. As anticipated (see sections 1.2 & 1.3), the bark tended to become trapped in the boundary layer and
stick against or slide down the walls.
The cause of the wall collisions appeared to be the horizontal component of velocity which is
often developed by both regular- and irregular-shaped free-falling objects. The air flow in the
working section was modified to minimise the horizontal movement of the non-tethered sample. This modification was accomplished by the addition of square concentric layers of mesh
placed at the base of the working section.
Three additional screens of standard insect mesh with sides of 750, 500 and 300 mm were
fastened to the primary screen at the bottom of the working section (Figure 5.1 (a)).
These additional screens were laid so that the orientation of the mesh wire on successive
screens differed by 45E. This was to minimise the occurrence of irregular areas where successive layers of mesh were >in phase=, which caused large variations in air velocity across the
centre zone.
The modified air flow profile had higher velocities near the walls and a central zone of lower
velocity (Figure 5.1 (b)). This centre zone (Z) in plan view was a 300 mm square at the base
of the working section and decreased in width by approximately 20 mm per metre of height
above the screens.
The modified air flow profile was investigated by measuring the maximum and minimum air
velocities at 10 mm intervals along a transect 1000 mm above the screens at contractor differential pressures of 2 Pa and 4 Pa. The profile of mean air velocities from the wall to the middle of the centre zone within the modified working section is shown in Figure 5.2.
The profile had low velocity air flow in the centre, high velocity near the wall, and, in between, a series of steep velocity gradients associated with the changes from screen to screen.
The high velocity zone acted as a barrier to a sample traversing to the wall so that the fire30 The CSIRO Vertical Wind Tunnel
to pitot tube
(not shown)
(b)
(2) Velocity profile of
modified airflow
Boundary
layer
READOUT
UNIT
PRESSURE
TRANSDUCER
LP
(d)
(4)Combusting
(a)
(1) Additional mesh
157
sample
HP
CONTROL
BOARD
Original mesh
screen
to contractor
Figure 5.1
Schematic diagram of the modified working section and the instrumentation used to determine air velocity in the working section during calibration. The diagram shows the arrangement of the screens used to modify the air flow (a). The velocity profile (b) and the location
of the centre zone (c) of this air flow are also shown. A combusting firebrand sample (d) is
shown within this zone.
brand was pushed back towards the centre (Figure 5.1). The modified air flow did not completely eliminate collisions with the walls of the working section for all samples all the time.
However, the amount of impact which occurred with regular-shaped messmate stringybark
(E. obliqua) samples was very much reduced.
The differences in air velocity between zones decreased as mean air velocity in the working
section decreased. The areas of turbulent air flow were characterised by large fluctuations in
PDP at a given position but the magnitude of these fluctuations were not investigated due to
the time constraints of the study.
Combusting samples typically moved in three dimensions throughout the air flow profile, but
tended to spend at least part of their flight within the centre calibrated zone (Z). When in this
Modified air flow
31
zone, samples tended to adopt a flat spin which could be fast or slow. Some samples showed
both clockwise and anti-clockwise spin during their combusting flight. While samples were
moving in the centre zone there was little variation in vertical movement. Although the
variation of air flow in this centre zone was not measured we considered that if this zone had
a wide range of air velocities the vertical movement of samples would have been erratic.
Therefore, we considered the air flow within this section was sufficiently uniform for these
experiments.
3
CDP 2 Pa
CDP 4 Pa
2.5
1.5
0.5
Modifying screens
Wall
Centre
42
0
40
0
38
0
36
0
34
0
32
0
30
0
28
0
26
0
24
0
22
0
20
0
18
0
16
0
14
0
12
0
10
0
80
60
40
20
he air flow in the working section was described in terms of the pitot differential pressure (PDP) measured at points along transects or within designated areas of the air flow.
During these measurements the contractor differential pressure (CDP) was maintained at a
predetermined value which was monitored and adjusted as required. At a constant fan speed
setting, the measurement of CDP (displayed on the readout device) fluctuated by one or two
units. This fluctuation was attributed to changes in ambient wind velocity outside the structure, which affected the air flow inside the structure. The measurement of CDP appeared to
be very sensitive to small changes in external ambient air movement as the readout device
only displayed whole units on a 0 - 250 unit scale. One unit corresponded to 1 Pa when using
the 0 - 250 Pa Furness pressure transducer, and 0.08 Pa when using the 0-20 Pa pressure
transducer. The problem of ambient air movement is discussed more fully below.
Similarly, the reading of PDP could also fluctuate. The causes of this variation in PDP are
also discussed below.
6.2 The variation in pitot differential pressure of the unmodified air flow
The variation in the PDP of the unmodified air flow (the air flow in the working section without any modifications to improve the air flow for firebrand observation) appeared to result
from a combination of fluctuations in ambient wind, velocity fluctuation due to turbulence in
the boundary layer, and low frequency vibration or resonance of the walls of the diffuser and
turning section.
33
he aim of calibration was to express the PDP in the lower 1250 mm of the centre zone
(Zone Differential Pressure or ZDP) as a function of CDP and height (h) above the screens.
The air velocity at that height could then be calculated from the derived ZDP. The measurements of ZDP for calibration were obtained along a single North-South transect of the centre
zone at different heights and for different CDPs.
It was necessary to test if the measurements along this N-S transect were representative of the
measurements throughout this centre zone. The results of the test were: mean pressure (S-N
centre transect) = 2.6 Pa and mean pressure (combined transects) = 2.44 Pa (P=0.25).
Although this test was conducted under less than ideal conditions because of fluctuating ambient wind speeds, there was no significant difference between the mean differential pressures
of the whole centre zone and the single central transect.
The pressure measurements for calibration (ZDP) were made by taking measurements at eight
locations along the North-South transect through the centre of the zone at five different
heights and for six contractor differential pressures. These locations were approximately 10,
30, 70 and 120 mm from each of the south and north boundaries of the centre zone (Figure
7.1).
35
equivalent to an error in derived air velocity of about " 4% or " 0.2 m s-1 and " 0.04 m s-1 at
air velocities of 5 and at 1 m s-1 respectively.
Constrained linear regression was performed for the 48 measurements at each of the five
heights to derive expressions of the form:
ZDP = k CDP
(7.1)
Figure 7.2 shows the mean value for centre zone differential pressure (ZDP) versus each
value for contractor differential pressure (CDP) for each category of height.
18
16
h = 250 mm
h = 500 mm
h = 750 mm
h = 1000 mm
h = 1250 mm
14
ZDP (Pa)
12
10
8
6
4
2
0
0
10
15
20
25
30
35
CDP (Pa)
Figure 7.2
The mean value for centre zone differential pressure (ZDP) versus contractor differential pressure (CDP) for five heights within the working section.
The values obtained for k using constrained regression are shown in Table 7.1:
Table 7.1. Values of coefficient k and its standard error (s.e.) derived by regression
for Equation 12 for each height category.
Height h (m)
k (mean)
s.e.
r2
0.25
0.5
0.0028
< 0.001
0.99
0.5
0.46
0.0044
< 0.001
0.99
0.75
0.41
0.0027
< 0.001
0.99
0.36
0.0029
< 0.001
0.99
1.25
0.32
0.0027
< 0.001
0.99
Linear regression of the mean values for h and k, tabled above gave:
2
r = 0.99
(se = 0.0037) (se = 0.0031)
k = - 0.18 h + 0.55
(7.2)
The air velocity during an experiment at a given height in the centre zone was derived from
the CDP, which was displayed on the readout device, was determined in the manner:
1) the slope k for the given height was derived using Equation 7.2.
2) the ZDP was derived using Equation 7.1.
3) the air velocity was then derived by substituting this value for ZDP into Equation
7.3, which is specific to the pitot tube:
v = 1 . 28
ZDP
1 . 013
T
1
P
288
d
(7.3)
where v is air velocity (m s-1), ZDP is centre zone differential pressure (Pa), P is station
barometric pressure during experiment (Bar), T is ambient temperature during the experiment
(oK) and d is relative air density (1).
The visual estimate of the height above screens of the firebrand sample (h), was a further
source of error in the estimate of air velocity. Height was estimated to the nearest 100 mm.
The consequence of estimating a height of 0.5 m (500 mm) instead of 0.6 m (600) mm was
an error of 4% of ZDP and hence 2% of air velocity.
37
he CSIRO vertical wind tunnel was constructed economically and within a limited
budget. The diffuser, straightening section, contractor, and tapered working section proved
sound in principle. Uniform air flow was maintained for the full length of the working section
without boundary layer separation. The divergent taper working section, at 3.4E total angle,
was sufficient for practical observation of burning embers.
The tunnel exhibited low frequency vibrations which were attributed to resonance of panels
in the diffuser, the turning section and working section, and the length of the tunnel. All
these panels were constructed from 12 mm thick plywood. Sections with dimensions 1.2 m H
2 m are currently unbraced. External bracing could easily be attached and should remove
most of this vibration.
Figure 8.1
A piece of Eucalyptus obliqua bark burning at its terminal
velocity in the working section of the CSIRO vertical wind
tunnel. Contractor differential pressure can be seen on the
digital display in the background.
Turning sections will inevitably introduce turbulence of the scale of the vane spacing (140
mm). These sections are normally placed well upstream from critical working sections to
allow this turbulence to dissipate. However, the placing of the straightener and contractor a
short distance from the turning section did not have a great adverse affect on the quality of air
flow.
38 The CSIRO Vertical Wind Tunnel
Construction difficulties increase exponentially with size. Calculations following the kinetic
energy increases of individual air parcels showed that a 7.1:1 contractor will reduce a 20%
variation in input flow to 0.5 % at the contractor exit. If the widest cross-section was reduced
from 2 metres to 1.5 metres, a 4:1 contraction could be used. A 4:1 contraction would reduce
a 20% variation to 1.1 % which is most probably adequate for firebrand studies. Screens of
similar porosity to those presently used would be needed in the diffuser sections, but note that
a smaller cross-section tunnel will generate higher back pressures. However, fans of the type
selected have a generous reserve of pressure capability to permit wind tunnel builders to
increase the number of screens to improve the flow quality into the contractor and so
compensate for a lower contraction ratio.
The above suggestion
maintains the working section
with a 750 mm entrance into
the adapter. This is a practical
size.
We would not
recommend making it smaller.
Boundary layers will build up
at the same rate regardless of
tunnel width. The boundary
layer will eventually encroach
on a small tunnel, producing an
unworkably small centre zone a
shorter distance into the
working section.
A pitot tube was mounted so
that it could be traversed
remotely to measure
differential pressure of the air
flow in the working section.
Measurements made directly
under the pitot support beam
u n d e r- estimated the air
velocity in the surrounding area
by 1%. The air velocities in
the uniform zone at the bottom
and top of the working section
varied by between " 0.5% and
" 1.5% (P = 0.05) respectively,
of the mean air velocity. This
air flow was adequate for the
study of firebrands.
Figure 8.2
A flaming firebrand being launched through the access door into
the lower half of the working section of the CSIRO vertical wind
tunnel. The firebrand was lit with a gas flame at the work bench
in the background.
Additional screens were placed at the base of the working section to modify the air flow and
reduce the rate of collision of untethered samples with the walls. The screens resulted in a
central zone of lower air velocity than the surrounding areas and tended to restrict the
horizontal movement of samples, reduce collisions with the walls of the working section so
that most samples spent at least a part of their flight in the calibrated centre zone. This
method was successful for experiments burning samples of the bark of E. obliqua. It is
possible that more sophisticated methods exist to achieve this, but time and budget
constraints prevented further investigation.
Discussion and Conclusions
39
The calibration method allowed the terminal velocity of a firebrand sample within the
modified working section to be determined as a function of contractor differential pressure
(CDP) and the height of the sample above the base of the working section. The CDP was
monitored continuously during experiments using a readout device mounted in the north wall
of the working section. The variation in air velocity throughout the centre zone, at a given
height and for a constant CDP, was approximately " 4% (P = 0.05) of the mean air velocity.
Fluctuation in air velocity in the working section due to ambient air movement was a problem
during calibration. Consequently calibration was only performed on calm days. Large
fluctuations in ambient wind did not affect the accuracy of measurements during experiments
but made it difficult to take measurements and did result in the loss of a few samples from the
working section.
While the tapered working section allowed a burning ember to fluctuate over a significant
range of terminal velocities, the ember had to be kept within a practical viewing >window= of
the working section. This was achieved by manually adjusting the overall air flow velocity by
altering the speed of the fan. While the speed of the fan responded quickly to the speed
controller, human reaction times and hand/eye coordination while observing the ember=s
flight were the limiting factors in the operational use of the wind tunnel. It was found that a
good percentage of the 4.25 metre tapered working section length was needed to steady an
ember and keep it captive.
The wind tunnel was successfully used as part of a doctoral study into the aerodynamics and
burning characteristics of E. obliqua (Ellis 2000). Accurate measurements of the terminal
velocity of burning bark samples enabled a model of maximum spotting distance to be
constructed.
9 Acknowledgements
he construction of this wind tunnel was funded by Mayne Nickless Ltd. The tunnel was
prefabricated in the CSIRO Division of Forestry and Forest Products= carpentry workshop at
Yarralumla, ACT, by Ricky Jordan. Members of the Bushfire Behavior and Management
Team assisted with the construction of the housing for the wind tunnel and with its final
installation and assembly: Phil Cheney, Jim Gould, Peter Hutchings, Sean Cheney and John
Coleman.
41
10 References
Bertin, J.J. 1984. Engineering Fluid Mechanics. Prentic-Hall, Inc., Englewood Cliffs, New
Jersey. 461 pp.
Borger G.G. 1976. The optimisation of wind tunnel contractions for the subsonic range. NASA
Technical Translation: NASA TT F-16899. Translation of "Optimierung von
Windkanalduesen fur den Unterschallbereich, Ruhr-Universitat, Fakultat fur
Maschinebau und Konstructiven Ingenieurbau, Doctoral Dissertation, 1973. 151 pp.
Bradshaw, P. and R.C. Pankhurst. 1962. The design of low speed wind tunnels. National Physical
Laboratory, Teddington, UK. Reprinted from NPL Aero Report 1039. 64 pp.
Clements H.B. 1977. Lift-off of forest firebrands. USDA Forest Service Research Paper SE-159.
11 pp.
Ellis, P.F. 2000. The aerodynamic and combustion characteristics of eucalypt bark--A firebrand study.
PhD Thesis, Australian National University, Canberra. 187 pp.
Mentah R.D. 1994. Aspects of the design and performance of blower tunnel components. Thesis (Ph.
D)--London University. Photocopy. Boston Spa, England: British library.
Muraszew, A., J.B. Fedele and W.C. Kuby. 1975. Firebrand Investigation. Aerospace report
ATR-75(7470)-1. The Aerospace Corp., El Segundo, Calif. 103 pp.
Muraszew, A., J.B. Fedele and W.C. Kuby. 1976. Investigation of fire whirls and firebrands.
Aerospace report ATR-76(7509)-1. The Aerospace Corp., El Segundo, Calif. 155 pp.
Rae, W.H. and Pope, A. 1984. Low-speed wind tunnel testing. Wiley, New York.
Raupach, M.R. 1990 Similarity analysis of the interaction of bushfire plumes with ambient winds.
Mathl. Comput. Modelling, 13(12):113-121 .
Tarifa,C.S., del Notario, P.P., and F.G. Moreno. 1965. On the flight paths and lifetimes of burning
particles of wood. Tenth symposium (International) on Combustion, pp. 1021 - 1037.
Tarifa, C.S., del Notario, P.P., F.G. Moreno, and A.R. Villa. 1967. Transport and combustion of
fire brands. Final report of grants FG-SP-114 and FG- SPEED-146, Aeronautical
Institute of Madrid for USDA. 90 pp.
References
efinition of static, dynamic and total pressure help in the description of wind tunnel
performance. These pressures are not measured in absolute terms but as a differential
pressure between two points.
The air flow in the vertical wind tunnel was initially calibrated using pitot tube anemometers
to take measurements of point velocity throughout the working section. The total air flow of
the working section air flow was determined as a function of Contractor Differential Pressure
(CDP). In subsequent experiments, where an unobstructed working section was required, air
flow was determined with a differential pressure reading between two static pressure taps, one
at the inlet of the contractor and one at the outlet of the contractor. The difference between
these two provided the CDP.
Static pressure was obtained by tapping into the tunnel wall. The inside opening was flush
with the inside wall such that the air flow suffered little disturbance. The pressure reading
was that of the free stream pressure within the tunnel at that location. The air flow within the
contractor was essentially non-turbulent and Bernoulli's principal was valid, i.e. the static
pressure (P) plus the kinetic energy of the air stream was a constant. For the contractor the
pressure and velocity at the inlet (v i) is related to the pressure and velocity at the outlet (v o) by
Bernoulli's equation :
Pi + 2 v i2 = Po + 2 v o2
(A1)
or
Poi = 2 v o2 - 2 v i2
(A2)
where P i and P o are the pressures at the inlet and outlet respectively (Pa), P oi is the
difference between the pressures at the inlet and outlet (Pa), and is the density of the air
(kg/m3).
Knowing that the ratio of exit velocity to the inlet velocity was fixed by the contraction ratio :
vo/vi = C
(A3)
meant that the directly measured differential pressure P oi could be converted to velocity.
43
The other means of determining velocity is with a pitot tube anemometer. This is also a
differential pressure device, one that measures both static and total pressure. The difference
between static and total pressure is dynamic pressure. Dynamic pressure provides a measure of
velocity.
Dynamic pressure is defined as the pressure exerted on a stationary object by the motion of a
fluid. For a fluid to exert dynamic pressure on an object, the fluid must be undergoing
acceleration, e.g. deflection by the object.
In the case of a pitot tube anemometer, the pressure measured at the tip inlet of the tube
(which faces into the fluid stream) is the total pressure of the fluid stream. This total pressure
is a the sum of the static pressure and the work done on the fluid in order to bring it to rest in
the inlet. This work is equal to the fluid=s initial kinetic energy (2v 2). The ports further
down the shaft of the pitot tube parallel to the fluid stream, where the fluid is moving close to
the free stream velocity, reads the static pressure. A differential pressure transducer connected
between the static and total pressure inlet ports gives a measure of the dynamic pressure and
provides a measure of the kinetic energy of the airstream.
The kinetic energy of the air, 2v 2, is often loosely referred to as the dynamic pressure.
Bernoulli=s principle is then that the sum of dynamic plus static pressure (i.e. the total
pressure) is constant. However, this is only true for a frictionless or drag-free system. In
practice, wind tunnels have resistance from screens, honeycombs, walls, etc, which are
described as losses in total pressure.
Appendix 1