Drag Reduction of a Tractor-Trailer Using Planar Boat Tail Plates

advertisement
Drag Reduction of a Tractor-Trailer Using Planar
Boat Tail Plates
J.D. Coon and K.D. Visser
Department of Mechanical and Aeronautical Engineering, P.O. Box 5725, Clarkson
University, Potsdam, NY, USA 13699
Abstract
The use of planar-sided boat tail plates for aft-end drag reduction on a tractor-trailer
was studied numerically, experimentally and on a full scale prototype. Parametric
wind tunnel tests utilized a 1:15 scale Peterbilt 379 tractor and 48 foot (14.6 m)
trailer with cavity plate concepts mounted perpendicular to the trailer base. Yaw
angles up to 9 degrees were examined. Qualitative numerical results confirmed a
pressure increase on the aft face of the trailer. Model drag increments, obtained at
zero yaw and a width-based Reynolds number of 230,000, based on trailer width,
indicated reductions in the drag coefficient, based on frontal area, of up to 0.075 or
about 9% of the baseline model trailer drag. Removal of the top plate degraded the
performance of all devices. Performance also decreased with yaw angle for all
plates mounted perpendicular to the trailer base, contrary to devices with angled
plates. Devices with shorter angled plates indicated better performance with the
top open rather than an open bottom. Drag reduction was more sensitive to plate
inset from the trailer edge than to plate length and a zero inset of the bottom plate
maximized performance. Two full scale prototypes were road tested, the first
utilized rigid composite sides with a flexible top and bottom and the second was an
all rigid-sided aluminum design. The former exhibited cross-country road fuel
savings of about 0.5 miles per gallon (0.2 kilometers/liter), approximately 9%, over
a 10,000 mile (16,093 km) trip, while the latter returned inconclusive results.
Estimated fuel savings for a typical 120,000 miles (193,121 km) per year traveled
were approximately 1500 gallons (5677 liters) per truck.
Introduction
The flow behind bluff bodies, and the desire to reduce the base pressure drag, has
been the subject of numerous investigations (Hoerner 1950; Mair 1965; Kentfield
1984; Rodriguez 1991). The reduction of the drag specifically associated with
tractor-trailers has also been under investigation for many years. In the early
1970’s, NASA, as well as private firms, extensively examined the concept of trailer
drag by modifying the flow over the forward part of the vehicle (Montoya and
250
J.D. Coon and K.D. Visser
Steers 1974) Their results indicated that commercially available devices could
reduce drag by up to 24 percent.
Sovram et al. (1978) reported that Mason and Beebe (1978) examined several
add-on devices to increase the base pressure of a tractor-trailer in 1976. They
included vertical and horizontal splitter plates, guide vanes, and non-ventilated
cavities as illustrated in Figure 1a. The vertical splitter plates had little or no effect
on the tractor-trailer drag, while the vanes exhibited an adverse effect. The only
device that showed positive results was the non-ventilated cavity design, shown in
the bottom right of Figure 1a. Mason and Beebe varied the length of the plates and
found an optimum of 22 inches (0.55 m) for a conventional 48 foot (14.6 m) trailer
which yielded an overall drag reduction of 5 percent.
(a)
(b)
Fig. 1. Drag reduction concepts a) Mason and Beebe in Sovram (1978) b) Bilanin (1985)
Similar research by Hucho (1987) on a small minivan style vehicle in the 1970’s
substantiated the drag reduction behavior of non-ventilated cavities. Cooper (1985)
conducted extensive tests on detailed scaled models of a tractor trailer and a panel
truck using rounded and beveled panels on the rear end of the vehicle. His results
indicated an optimum reduction in the drag of 7-10% of the baseline model drag
and indicated it was not unreasonable to expect this at full scale. A patent was filed
by Bilanin (1985) on a variation of the cavity design concept, illustrated in Figure
1b, that inset the endplates from the trailer perimeter. His claims indicated a drag
reduction on the order of 10%. Recently patents filed by Boivin and Roberge
(2001) describe an attachment of plates to the edge of the trailer and inclination
toward the centerline, but with no bottom plate.
NASA has extensively supported testing of devices aimed to reduce the pressure
drag. A summary of the NASA work from 1973–1982 by Saltzman and Meyer
(1999) indicated that a reduction in base drag, such as by using a truncated boat
tail, will be necessary in light of the increase of base drag associated with forebody
streamlining. In 1988, full-scale tests were performed at NASA’s Ames Research
Center on a full scale, inset, non-ventilated cavity device (Lanser et al. 1991)
similar to Bilanin (1985). Inset, as well as overall plate length, were optimized.
The results showed that the highest drag reduction came with a plate length of
0.36w and an inset on the top and side plates of 0.06w, where w was the width of
Drag Reduction of a Tractor-Trailer Using Planar Boat Tail Plates
251
the trailer. This translates to a plate length of 33 inches (0.84 m) inset at
approximately 7 inches (0.18 m) from each edge for a 48 foot (14.6 m) trailer. The
researchers recorded an overall drag reduction of approximately 10% over a +/- 15°
yaw range. Tests at NASA using innovative testing techniques to validate CFD
methods on bluff body ground vehicle wakes have continued to the present
(McCallen et al. 1999).
Hucho and Sovran (1993) indicated that the major aerodynamic problem of a
bluff type body moving at subsonic speeds through a fluid is at the rear, not the
front. They state that the benefits of drag reduction are threefold: reduced fuel
consumption, increased acceleration, and increased top speed. If the objective is
reduced fuel consumption, the latter two benefits can be used to enhance this
requirement. Re-gearing of the drive train converts increased acceleration to a fuel
savings component, while a reduction in installed engine power accomplishes the
same for the increased top speed capability. It is also worth mentioning, in light of
the recent heightened focus on the environment, that a reduction in emissions will
result from this drag reduction.
The objective of the present study was to investigate and develop a working full
scale prototype of a device to passively reduce the base pressure drag on a tractortrailer. Title 23 of the United States Code of Federal Regulations (Part 658.5)
indicates that devices added to the rear of a semi-trailer, whose function is related
to the safe and efficient operation of the vehicle, are excluded from the length
determination, provided that they are not used for carrying cargo. Revision of this
code is presently underway, however the current regulations stipulate a maximum
length of 5 feet (1.52 m). The US Department of Transportation reported that there
were 1,997,345 combination trucks, including nine-axle tractor trailers, in service
in 1998. Approximately 128.4 billion miles (206.6 billion kilometers) were
traveled and 21.1 billion gallons (79.5 billion liters) of fuel were consumed. A
decrease in fuel consumption of only 5% would incur savings of over 1 billion
gallons (3.79 billion liters) per year and, of course, billions of dollars. Wind tunnel
experiments were utilized in the present study for geometry optimization and
enabled comparison of model scale to full scale results. A key consideration of the
full scale prototype was maintaining the functionality of tractor-trailer.
Numerical Simulation
Although the primary focus of this study was experimental, an effort to gain some
qualitative insight into the effect of endplates, and to provide direction for
experimental tests, was first undertaken using the numerical code FLUENT. The
use of FLUENT is questionable when attempting to estimate drag numbers or flow
structures on such an unsteady, asymmetric, separated bluff body flow, however
the purpose was a quick and “dirty” look to see if there was some time averaged
improvement on the aft trailer face. A simplified tractor-trailer geometry was
modeled three dimensionally in a large, 50ft x 50ft x 150ft, (15.2 m x 15.2 m x 45.7
m) wind tunnel as depicted in Figure 2. A total of 2.64x105 nodes were used at a
Reynolds number of 4.32x106 based on truck width. The grid was not refined in the
present study as the number of nodes were as many as the available computer
hardware could accommodate. The model was run under steady state conditions
only and utilized a k-ε turbulence model. The purpose of the numerical model was
252
J.D. Coon and K.D. Visser
simply to see if there was any pressure increase on the back of the trailer. Cases
were computed with and without plates attached perpendicular to the back of the
trailer. The plates were inset on each side an equal distance, d, from the trailer
edge for a range 0<d<36 inches (0 < d 0.91 m) full-scale in 6 inch (0.15 m)
increments. Plate length, L, was varied from 3 – 6 ft (0.91 – 1.83 m) in 1 ft (0.3 m)
increments.
(a)
(b)
Fig. 2. Simplified numerical tractor trailer representation a) Grid distribution b) Pressure
distribution plane at aft face
An example of the calculated pressure distribution on the baseline (no device)
aft trailer face is illustrated in Figure 3a. The area of lowest pressure occurred near
the bottom of the trailer and extended towards the ground plane. The addition of
the plates was observed to qualitatively increase the pressure on the aft face of the
truck to levels in the flow field around the truck, as illustrated in Figure 3b. Close
inspection of the values in Figure 3b reveals asymmetries, probably due to a lack of
grid resolution in the vicinity of the plates. As noted above, the grid was not
refined due to a lack of available hardware. As the complex nature of this time
dependent, separated flow field limits the accuracy of a code such as FLUENT,
these numerical results were only used to obtain a general feel of what might occur
in the flow with the plates attached and to assist in the wind tunnel experimental
direction. Details can be found in Coon (2001)
Drag Reduction of a Tractor-Trailer Using Planar Boat Tail Plates
(a)
253
(b)
Fig. 3. Numerical pressure distributions on aft face of trailer a) Baseline b) Boat tail plates
added, d = 6 inches (0.15m), L = 48 inches (1.22 m)
Experimental Study
An extensive set of wind tunnel tests were conducted on a scale model of a tractortrailer in the Clarkson University Low Speed Wind Tunnel. The tunnel facility is
an in-draft, open circuit type with a 4.7:1 contraction inlet. The test section is of 48
inch by 36 inch (1.22 m by 0.91 m) cross section with a length of 65 inch (1.65 m).
Honeycomb and fine screens are mounted across the entrance to the contraction
inlet to reduce the turbulence and flow irregularities. The maximum speed
obtainable in the tunnel was approximately 58 miles/hour (26 m/s) which
corresponds to a Re/m of about 1.5e06/m.
Models and Apparatus
A wind tunnel model was constructed at 1:15.25 scale to resemble many of the
common tractor-trailers on the road today. The cab was modeled after a Peterbilt
tractor model 379 and the trailer was modeled after a standard full-scale 48 foot
(14.6 m) trailer. The actual total length of the model was 51.35 inches (1.3 m),
with the trailer extending 37.75 inches (0.96 m) and the cab approximately 11.1
inches (0.025 m) There was a gap between the tractor and trailer of 2.5 inches
(0.064 m) which is approximately 38 inches (0.97 m) full-scale. The cross-section
of the tractor-trailer was approximately 6.30 in. x 7.125 in. (0.16 m x 0.18 m).
Eighteen model wheels were attached, but kept immobile for the experiments.
Figure 4 illustrates the model in the tunnel. The width-based Reynolds number for
the experiment was approximately 2.3x105, significantly below that of full-scale
width-based Reynolds numbers at common highway speeds of approximately
4.5x106.
The model blockage at 0° yaw in the tunnel was 2.6%. A 0.5 inch (1.27 mm)
thick splitter plate was installed to position the model in the center of the test
section, out of the boundary layer, and to model the ground plane. The added
blockage that the modeled ground plane contributed to the overall wind tunnel
blockage was 1.4%. This brought the total test section blockage up to 4%, which is
slightly below the recommended maximum of 5% (Mason et al. 1973).
254
J.D. Coon and K.D. Visser
Fig. 4. Wind tunnel model of tractor trailer
Four different plate cavity models were tested. The first, denoted Equal Inset
(EI), is illustrated in Figure 5a and was similar to the numerical geometries. The
second, Equal Inset, Zero Bottom Plate Inset (EI-0B), was identical to EI, but with
the bottom plate at zero inset, as suggested by previous literature (Bilanin 1985;
Lanser et al. 1991). The last two model types tested were identical to the previous
two except that the top plate was removed, and were referred to as EI-NT and EI0B-NT, respectively. Figure 5b illustrates an EI-0B-NT geometry. Full-scale
issues fueled the interest in top plate removal. If viable from a drag perspective, it
would be more practical as a design as it incorporates less material and would be
far easier to operate than a four-plate design. The geometries test matrix is listed in
Table 1.
(b)
(a)
Fig. 5. Wind tunnel plate cavity geometries a) Equal Inset (EI) b) Equal Inset, Zero Bottom No
Top (EI-0B-NT)
Table 1. Model test geometries: W=EI; X=EI-NT; Y= EI-0B, Z= EI-0B-NT
Plate Inset (Full
Scale)
d, inches (m)
1.91 (0.048)
3.81 (0.096)
5.72 (0.15)
7.63 (0.19)
9.53 (0.24)
11.44 (0.29)
13.34 (0.34)
15.25 (0.39)
19.06 (0.48)
L =36 inches (0.91
m)
L=48 inches (1.22
m)
L=60 inches(1.56 m)
W,X,Y,Z
W,X,Y,Z
W,X,Y,Z
W,X,Y,Z
W,X
W,X,Y,Z
W,X,Y,Z
W,X,Y,Z
W,X,Y,Z
W,X,Y,Z
W,X,Y,Z
W,X
W,X
W,X,Y,Z
W,X,Y,Z
W,X,Y,Z
W,X,Y,Z
W,X,Y,Z
W,X,Y,Z
W,X
W,X
W,X
Drag Reduction of a Tractor-Trailer Using Planar Boat Tail Plates
255
A model of the Boivin and Roberge (2001) design was also constructed,
containing three plates (top and sides) inclined at an angle towards the centerline
with no inset from the perimeter of the trailer. An estimated plate length of 18
inches (0.46 m), full-scale, and angle of 15º , based on photographs, was modeled.
A bottom plate was also added to the original design, perpendicular to the trailer
face, as an additional configuration.
A biaxial, drag and side force, balance was designed and built to take
measurements on the tractor-trailer model. It uses two IKO International linear
translators with crossed roller bearings (CRWU 80-125) mounted perpendicular to
each other. They enabled motion in the direction of flow and perpendicular to the
flow. Two Precision Transducers model PT1000 load cells, were used to take
measurements for drag and side force on the truck model. Both have an instrument
error of 0.03% of the applied load. Voltage output data from the load cells was
signal conditioned with a National Instruments (NI) Strain Gauge Board Model
SG-2043 and acquired with a NI data acquisition (DAQ) card model PCI-6024E.
Zero Yaw Test Results
The differences in drag coefficient, ∆CD, and the percentage of drag reduction,
%∆CD, based on maximum frontal area for the truck model were used to compare
the baseline, no device, geometry to each different device where:
and
CD =
∆C D = C Dbase − C Ddevice
∆CD
%∆CD =
*100
CDbase
Drag
Dynamic pressure * Frontal area
(1)
(2)
(3)
The dynamic pressure is one half of the tunnel air density multiplied by the tunnel
velocity squared.
The experimental results for 0° yawed geometries will be presented first
followed by the yawed conditions. The method of Kline (1985) was used to
estimate the uncertainty in CD , ∆CD and %∆CD. Although the typical uncertainty
error in CD was ± 0.015 or approximately 2% of the measured value, the percentage
of error associated with ∆CD and %∆CD can be significantly higher depending on
the actual drag savings of the device as the uncertainty in ∆CD can represent a
significant portion of the drag savings increment. Consider the uncertainty in ∆CD,
U∆C D , which is determined as:
U∆C D
2
2

∂∆CD  
∂∆CD 
 + UC DDevice

= UC DBase
∂CDBase  
∂CDDevice 

(4)
256
J.D. Coon and K.D. Visser
Since
UC DBase is equal to UC DDevice this becomes:
U∆C D = 2 UC DBase
(5)
Here the uncertainty in ∆CD is roughly 40% greater than that of CD. The
percentage of error, however, is much higher, especially as the difference in drag,
∆CD, decreases.
% error in ∆C
D
= 100
U∆C D
∆CD
(6)
The same is true for %∆CD. As a result, the % error in ∆CD, typically on the
order of ± 0.02, becomes about 28% for an incremental drag savings of 10% over a
baseline CD of, say, 0.8. A drag savings of only 5% would incur % error in ∆CD of
over 55%. Percent error magnitude in %∆CD is similar.
It is important that this error associated with these data be kept in mind when
observing the trends in the following plots. Although many repeat cases were
taken and data was acquired over significant time periods to ensure time averaged
results, consideration of the error bars obscures many of the smaller trends
observed between individual geometric details. Thus, despite the fact that the
overall drag savings of these devices, relative to the baseline geometry, were
statistically significant, the specifics of the data presented below associated with
individual geometric differences should be taken only as indicative of potential
trends and not definitive conclusions. Note that the error bars have been left off the
figures for clarity.
Equal Inset
The behavior of the EI case is plotted in Figure 6a. The data indicated an optimum
geometric range of plate length, L, of 45in. to 55in. (1.14m to 1.40 m) and inset, d,
of 3in. to 7in. (0.076m to 0.18m) full-scale. For comparison, the values of %∆CD
are presented in Figure 6b to illustrate the danger of only presenting percentages of
drag reduction instead of ∆CD. Since the values of %∆CD are scaled by the
baseline, an unrealistic value of the magnitude of drag of the baseline geometry can
lead to erroneous conclusions. This is especially critical at model scales and
Reynolds numbers. As an example, note the difference in perceived %∆CD
performance of the 3 foot (0.91 m) case at an inset of 4 inches (0.10 m) compared
to the values of ∆CD. Also, the incremental values of ∆CD can more readily be
applied to a full scale geometry, as percentages require the baseline model CD. The
maximum %∆CD was achieved with a geometric configuration of L=48in. (1.22 m)
and d=5.72in. (0.14 m), yielding an 8.84% reduction in drag, however in terms of ∆
CD, the L = 36 inch (0.91 m), d = 3.8 inch (0.097 m) geometry performed as well as
the 48 inch (1.22 m) case.
Drag Reduction of a Tractor-Trailer Using Planar Boat Tail Plates
257
10
0.08
9
0.07
8
0.06
∆ CD
% ∆CD
7
0.05
Full Scale Plate Length
0.04
6
5
Full Scale Plate Length
5 feet
4
4 feet
3 feet
0.03
5 feet
4 feet
3 feet
3
0.02
2
0
2
4
6
8
10
12
14
16
18
20
0
2
4
6
Full Scale Inset (inches)
8
10
12
14
16
18
20
Full Scale Inset (inches)
(a)
(b)
Fig. 6. Effect of plate length and inset on drag reduction, equal inset geometry a) ∆CD b)
%∆CD
Effect of Zero Bottom Plate Inset
The EI-0B configuration performed best in similar geometric ranges to that of the
EI cases, however, as seen in Figure 7, a local maximum in length is not as
obvious. For optimum performance, the plate length should be L>45in. (1.14 m),
but no strong upper limit for L is indicated. The optimum range for plate inset was
within 3in.<d<8in. (0.076m<d<0.20m), with a %∆CD range from 8.5-9.0%. The
maximum ∆CD for the EI-0B case was achieved with L=48in. (1.22 m) and
d=5.72in. (0.14 m), yielding a ∆CD = 0.0754. Similar results were achieved for the
case with L=60in. (1.52 m) and d=5.72in. (0.14 m) where ∆CD = 0.0747.
The EI-0B configuration forces a streamline to extend from the bottom-most
part of the trailer, unlike the plate inset which allows separation at the trailing edges
of the trailer and vortex formation in the step region. This vortex strategy for drag
reduction appears to work best for the top and sides of the trailer, but the bottom
plate seems most effective without this formation.
0.08
10
0.07
9
8
0.05
%∆CD
∆ CD
0.06
Full Scale Plate Length
5 feet 0bottom
0.04
4 feet 0bottom
7
6
Full Scale Plate Length
5 feet 0bottom
5
4 feet 0bottom
3 feet 0bottom
0.03
3 feet 0bottom
4
0.02
3
0
2
4
6
8
10
Full Scale Inset (inches)
(a)
12
14
0
2
4
6
8
10
12
14
Full Scale Inset (inches)
(b)
Fig. 7. Effect of plate length and inset on drag reduction, zero bottom geometry a) ∆CD b)
%∆CD
258
J.D. Coon and K.D. Visser
Removal of Top Plate
The effect of removing the top plate decreased the performance for both the EI-NT
and EI-0B-NT four-plate configurations. For the EI-NT case, no indication of an
optimum length was observed between the configurations. Best results were
obtained with the lowest plate inset, d. Although it was not tested, the optimum
configuration may even be a zero plate inset (d=0). The maximum ∆CD of the EINT configurations tested occurred at L=48in. (1.22 m) and d=1.91in. (0.049 m),
yielding a drag reduction of ∆CD = 0.0517 or about 6.2%.
In contrast, the model type EI-0B-NT operated most effectively when plate inset
was in the range 2in.<d<5in, (0.051m<d<0.13m) yielding a maximum ∆CD =
0.0532 at L=60in. (1.52 m) and d=3.81in. (0.097 m) or a %∆CD=6.3%. This was
closely followed by the configuration having L=48in. (1.22 m) and d=3.81in.
(0.097 m) for a ∆CD = 0.0521 or %∆CD=6.3%.
The impact of no top plate is summarized in Figure 8 for the 48 inch (1.22 m)
length plate geometry. The EI-NT geometry performed at 57% below that of the
EI, while the EI-0B-NT dropped to 59% of the corresponding EI-0B maximum
value. Both configurations also exhibited a shift in the optimum inset value. The
EI results indicated an optimum range for plate inset of 3in.<d<7in
(0.076m<d<0.18m) while the optimum range for the EI-NT case occurs at a lower
range of 0in.<d<4in. (0.0m<d<0.10m) Similarly, the EI-0B case had a local
maximum in the range 3in.<d<8in. (0.076m<d<0.20m), but the EI-0B-NT case
indicated maximum performance within the range 2in.<d<5in. (0.051m<d<0.13m).
0.08
0.07
0.06
∆ CD
0.05
0.04
0.03
Four Foot Plate Cases
0.02
EI
EI-0B
EI-NT
EI-0B-NT
0.01
0
0
2
4
6
8
10
12
14
16
18
Full Scale Inset (inches)
Fig. 8. Effect of no top plate, plate length and inset on drag reduction of 4 foot (1.22 m) plate
length
Yaw Test Results
As Cooper (1982) has indicated, yawed conditions can play a major role on the
drag reduction, however yaw angles greater than 9 degrees are relatively
insignificant for trucks at 56 mi/hr (90 km/hr) based on statistical wind estimates.
Since these yaw angles are exceeded < 10% of the time, the present study restricted
testing to < 10 degrees. Limited yaw measurements were performed on select
Drag Reduction of a Tractor-Trailer Using Planar Boat Tail Plates
259
model configurations from the zero degree tests from –3° to +9°, with an increment
of 3°.
For all perpendicular plate configurations tested, and at low angles, the behavior
of the drag with yaw angle was concave down. In other words, the ∆CD
improvement decreased with increasing yaw angle. The device with maximum
∆CD from the zero degree tests, the EI-0B, L=4ft. (1.22 m) , d=5.72in. (0.15 m) is
illustrated in Figure 9. For all configurations tested with 4 foot (1.22 m) plate
length, drag benefits decreased with yaw angle. Erratic behavior was observed at
higher yaw angles for the model with L=3 foot (0.91 m). This may be attributed to
the effect of less area exposed to the flow with a shorter plate length L. Removal
of the top plate had a consistent, negative offset on drag reduction over the yaw
angle range, similar to that of the 0° yaw data of Figure 8.
Yaw tests were also conducted on variations of the Boivin and Roberge (2001)
design. As previously mentioned, this design does not include a plate inset for any
of the three plates which were placed at a 15° angle with respect to the trailer top
and sides. Contrary to the perpendicular plate behavior, at low yaw angles the
trend is somewhat concave up. That is, ∆CD increases with increasing yaw angle
over the range ±3°. The behavior is similar for the inverted case which even
indicated better performance. The addition of an orthogonal bottom plate to the
design improved the drag yet further, but the performance remained below the 4
foot (1.22 m) optimum at 0° yaw.
It should be noted that the curves are not ideally symmetric about a yaw angle of
zero degrees. Two possible explanations can be given for this behavior. First,
although the centerline of the model was aligned with the tunnel free-stream, the
addition of the aft devices could be cause for a non-symmetric geometry. Each aft
device was carefully hand-built and attached with magnetized strips to the trailer,
however the impact of small scale construction and alignment errors at this scale
could be more serious than it was thought to be, certainly more so with a tunnel
model than the full scale device. Secondly, the asymmetries present at +/- 3
degrees can be seen to fall with in the estimated error bands of a ∆CD of about +/0.02, indicating that there may be no asymmetry present at all.
0.1
0.08
0.06
∆ CD
0.04
0.02
0
4 foot, zero bottom, 6 inch inset
-0.02
Makka original
Makka, inverted
-0.04
Makka fourth plate added
-0.06
-3
0
3
Yaw Angle (Degrees)
6
9
Fig. 9. Effect of yaw angle on drag of selected designs
At higher yaw degrees, the results became more erratic and somewhat
unexplainable, possibly due to measurement uncertainty, as seen in Figure 9. At
+6°, the inverted case continues its convex behavior (as experienced over ±3°).
260
J.D. Coon and K.D. Visser
This differs from both the original and 4-plate cases. The common element in the
original and 4-plate cases is the presence of a top plate. It is possible that a
negative effect from the top plate is occurring at +6°. This is inexplicable at the
present, however, since at +9° the trend is reversed. That is, at +9°, the 4-plate
angled outperforms both the original and the inverted Boivin and Roberge (2001)
design. It is suspected that the low Reynolds number effects are contributing to
this behavior in addition to measurement uncertainty. The general conclusion in
the yaw tests is that further data is necessary at more yaw angles and higher
Reynolds numbers to substantiate any claims made for yaw trends of the angled
plate design, or any variations.
Literature Comparison
The optimum geometrical results from the present experimental data were
compared to that of Bilanin (1985) and Lanser, et al. (1991) using the definitions of
Figure 10. Plate insets, G and D, and the plate length, L, were nondimensionalized by the trailer width, W.
Fig. 10. Geometry definitions for literature comparison
Table 2 presents results for the optimum geometries. In terms of plate
placement, both the Bilanin (1985) and Lanser, et al. (1991) published geometries
compare most closely to the present experimental case with zero bottom plate inset,
since they all have the same bottom plate position. The G/W and D/W inset ratios
reported by Bilanin were twice as great as the current experimental inset ratios,
indicating higher optimum plate insets for the former design. Bilanin also specifies
a range for optimum inset ratio of 0.1<G/W=D/W<0.2, but the current
experimental values still lie outside this range at G/W=D/W=0.06. The plate length
ratio reported by Bilanin (1985) of L/W=0.44 is less than that of the present
optimum at L/W=0.50, however, he also specified that the full-scale range for
optimum plate length should be in between 40in.<L< 56in. (1.02m <L< 1.42m)
within which the present study falls.
Drag Reduction of a Tractor-Trailer Using Planar Boat Tail Plates
261
Table 2. Optimum geometry for current and published results
Case
Equal inset
Bottom plate inset = 0
Bilanin (1985)
Lanser et al. (1991)
Length, L/W
0.50
0.50
0.44
0.36
Inset, G/W = D/W
0.06
0.06
0.13
0.06
The results of Lanser et al. (1991) indicate similar geometries to the present
study with identical insets of G/W=D/W=0.06. Plate length ratio differs, from
L/W=0.50 in the present study to L/W=0.36 for Lanser, et al, but recall, as
discussed earlier, that plate inset had a much stronger influence on drag reduction
than does the plate length.
Although it is tempting to speculate that an increase in the drag savings would
occur as Reynolds numbers increased into the full scale regime, other
investigations have indicated this is not always the case (Watkins et al. 1987;
Cooper 1985) provided the model scale Re is great enough A full-scale tractortrailer operates at a Re of about 4x106 based on width of the trailer. Full-scale wind
tunnel tests have shown that the drag coefficient for tractor-trailers lies in the range
0.6 to 0.9 at typical road speeds (McCallen et al. 1999). It has been indicated (SAE
J1252) that although the Re based on width need not match full scale for acceptable
results, it should be a minimum of 7x105. Others have indicated a minimum of
1x106 (Watkins et al. 1993) or 2x106 based on the square root of frontal area
(Cooper 1985).
Full Scale Study
A prototype of the most promising configuration was constructed for the road
testing to compare with tunnel results and to provide an opportunity for the
students involved in the project to get “real world” data. As pointed out by
Saunders et al. (1985), wind tunnel results tend to be quite optimistic, not only
because of some unknown Re scale effects, but by other factors such as turbulence
levels. The primary objectives of the prototype were to maintain full functionality
of tractor-trailer unit with regards to loading and unloading of cargo and to find a
balance between manufacturing requirements and aerodynamic ideals.
The first full scale prototype was engineered to be collapsible such that it could
be folded to within three inches of the doors. The collapsing sequence is illustrated
in Figure 11. The nature of the design development timescale was that the
prototype was fabricated as the tunnel tests continued. Hence, the length of 32
inches (0.81 m) was chosen for the prototype, based on the work of Lanser et al.
(1991), before the tunnel tests indicated 48 inches (1.22 m) may be better.
The main frame of the device was constructed using 2 inch by 2 inch aluminum
(0.05m x 0.05 m) box beams TIG welded to 2 inch (0.05 m) by 1 inch (0.025 m)
horizontal box beams on the top and bottom. Both the top and bottom panels were
constructed of a flexible canvas to allow the unit to fold in. The side plates were
constructed out of fiberglass panels on separate aluminum frames. To facilitate
access to the trailer, the device was only hinged to the right door. Weatherstripping was placed on the frame where each beam met the doors to improve
sealing. To alleviate concerns about the added weight of the main frame and flaps,
262
J.D. Coon and K.D. Visser
the standard extruded aluminum hinges were replaced with steel hinges.
Deployment required manual actuation of the two side flaps, both of which are
within reach of the truck operator from the ground
(a)
(b)
(c)
Fig. 11. Full scale prototype storage sequence a) open b) collapsing c) closed
A tensioning system was devised so that the cloth stayed taut and the side flaps
remained opened during operation. High-strength climbing rope was attached to
the flaps via eyelets on the top and bottom of each flap. A ratcheting tie-down
tensioner (a “come-along”) pulled the flaps into the open position. In order to close
the device the tensioners were relaxed and the rope unhooked. A locking
mechanism latched the device to the driver side door when the doors were closed
and the trailer in motion. The flaps were secured in place when non-operational by
two keyed hasp locks.
Issues associated with fitting the device to the truck doors were alleviated by
purchasing doors from the local trucking company, LaValle Trucking Inc. (LTI),
who were going to road test the device, and installing the device on the doors
before the doors were put on the trailer. Originally, the prototype was to be
centered on the back face of the trailer, inset from all sides by about 7 inches (0.18
m). Due to complexities within the trailer doors, such as hinges and latches, the
prototype placement had to be changed. The bottom of the prototype framing was
inset about 20 inches (0.51 m), because of interference with the latches used to get
in and out of the trailer, and the sides were inset 10 inches (0.25 m). Figure 12
illustrates the installed prototype.
(a)
(b)
Fig. 12. Full scale prototype configuration on test vehicle a) closed b) open
The prototype was tested on a cross-country trip that covered approximately
10,000 miles (16,093 km). Several factors were compared before, during, and after
the trip took place, including fuel mileage, fuel consumed, payload, idle times, and
Drag Reduction of a Tractor-Trailer Using Planar Boat Tail Plates
263
average speeds. The truck was fitted with an onboard fuel consumption and
condition monitoring computer. It should be noted that as this was a one-truck test
and not a standard SAE Type II test (SAE J1321 1981) the data collected could
have an appreciable bias due to the particular ambient atmospheric and road
conditions experienced during the trip.
The results from the test indicate that the addition of the device increased the
fuel mileage from 5.8 to 6.3 miles per gallon (mpg), or 2.5 to 2.7 km/liter, ± 0.1%,
for a savings of approximately 0.5 mpg (0.2 km/liter) This corresponds to an 8.3%
average increase in fuel mileage. Based on these values, an estimated savings for a
typical 120,000 miles (193,121 km) per year use would be on the order of 1500
gallons (5677 liters) per truck. Single trip and single truck bias without a control
issues aside, the preliminary data did indicate encouraging results in the right
direction. Observation of the device before the cross-country road test did indicate
an increase in pressure in the aft region of the trailer. The upper and lower canvas
surfaces bulged outwards from the stationary position, indicating a pressure
difference across the material.
A second generation device was constructed based on experience gained with
the first test. It utilized all aluminum rigid sides with a split plate design as
illustrated in Figure 13a. Unfortunately, uncertainty in results and a short 5500 mile
(8851 km) trip, combined with questionable data reporting led to inconclusive
results. In addition, geometry and fatigue issues, as illustrated in Figure 13b,
caused complications. Excessive vibration, according to the driver, was the cause
of the cracking along the hinges, breakage of a pin joint in the top flap, and
resulting geometry mismatch that led to increased cavity ventilation.
(a)
(b)
Fig. 13. Full scale aluminum prototype a) installed on trailer b) hinge fatigue
Conclusions
The wind tunnel and full scale data presented in this study verify results from the
literature that significant drag reduction can occur through the use of a cavity
device mounted on the aft face of a tractor-trailer rig. The best results were
obtained with a four foot (1.22 m) plate width yielding a drag savings of about
264
J.D. Coon and K.D. Visser
CD=0.075 or about 9% of the baseline drag. It should be noted that because the
incremental drag is a difference of two substantially larger values, the error in ∆CD
was appreciable.
All the wind tunnel results indicated a higher sensitivity for plate inset on drag
reduction than for plate length. Setting the bottom plate inset close to the bottom of
the trailer base appears to be essential to maximizing drag reduction. The optimum
experimental plate inset, d, was found to agree very closely to that of Lanser, et al.
(1991). Removal of the top plate decreased performance by about 30% in all cases,
but there still was an observable drag reduction and this may be a key factor in
realizing the practical role of such a device on a full-scale level.
Device performance decreased with yaw angle for plates mounted perpendicular
to the trailer base. Each of the angled plate design variations, however, displayed a
concave up behavior at low yaw angles, contrary to the concave down behavior of
the perpendicular plate-cavity devices. At small yaw angles, the four-plate angled
plate device achieved the highest drag reduction, indicating the importance of the
presence of the bottom plate to maximize performance.
A full-scale, fully operational prototype was road tested and indicated that the
addition of such a cavity device increased the fuel mileage by 0.5 mpg (0.2
km/liter) or about 8%. Estimated savings for a typical 120,000 miles (193,121 km)
per year use would be on the order of 1500 gallons (5677 liters) per truck
The positive results obtained from the scale and full size tests have prompted a
continued study of the potential use of a cavity design for drag reduction, especially
the possibility of a design employing only three sides. A new series of models is
being constructed to optimize length, inset, and boat tail angle. Yaw behavior,
optimization of three plate geometries, and downstream effects obtained with seven
hole probe flow field surveys will be conducted. In addition, two more full scale
prototypes are being designed with the aim of improving functionality in addition
to obtaining full scale data.
Acknowledgements
The authors would like to thank the National Collegiate Inventors and Innovators
Alliance (NCIIA) for their support of this project through an E-Team grant and the
New York State Energy Research & Development Authority (NYSERDA) Grant
Agreement #6436 for their support. The contributions by the local trucking firm,
LaValle Trucking Inc of Potsdam, NY, of time and labor on the road are also much
appreciated. Finally, a big thank you to the enthusiastic undergraduate Clarkson
students who participated in this project: Anna Sawabini, Matthew Pausley,
Matthew Menotti, Jamison Coon, and Matthew Allen.
References
Bilanin AJ (1985) Vehicle Drag Reducer. US Patent 4682808
Boivin and Roberge (2001) Vehicle Drag Reducer. US Patent Number 6257654
Bureau of Transportation (2000) National Transportation Statistics
Coon JD (2002) The Effects of Non-Ventilated Plate-Cavity Devices on Drag Reduction of
Tractor Trailers. Masters Thesis, Clarkson University, Potsdam NY.
Drag Reduction of a Tractor-Trailer Using Planar Boat Tail Plates
265
Cooper K (1982) The Wind Tunnel Testing of Heavy Trucks to Reduce Fuel Consumption.
SAE 821285
Cooper K (1985) The Effect of Front Edge Rounding and Rear-Edge Shaping on the
Aerodynamic Drag of Bluff Vehicles in Ground Proximity, SAE 850288
Cooper KR, Mason WT, Bettes WH (1982) Correlation Experience with the SAE Wind
Tunnel Test Procedures for Trucks and Buses. SAE 820375
Davidson L, Krajnovic S (2001) Large Eddy Simulation of Flow around a Ground Vehicle
Body AIAA 2001-01-0702
Hoerner SF (1950) Base Drag and Thick Trailing Edges. J Aeronautical Sciences:622-628
Hucho W (1987) Aerodynamics of Road Vehicles
Hucho W, Sovran G (1993) Aerodynamics of Road Vehicles. Annual Review of Fluid
Mechanics
Kentfield, JAC (1984) Short, Multi-Step, Afterbody Fairings. J Aircraft: Vol.21 No. 5
pp.351-2
Kline SJ (1985) The Purposes of Uncertainty Analysis. J. of Fluids Eng: 153 - 160
Lanser WR, Ross JC, Kaufman AE (1991) Aerodynamic Performance of a Drag Reduction
Device on a Full-Scale Tractor/Trailer. SAE 912195.
Mair WA (1965) The Effect of a Rear-Mounted Disc on the Drag of a Blunt-Based Body of
Revolution. The Aeronautical Quarterly Royal Aeronautical Society:350-360
Mason Jr WT, Beebe PS, Schenkel FK (1973) An Aerodynamic Test Facility for ScaleModel Automobiles International Automotive Engineering Congress
Mason Jr WT, Beebe PS (1978) The Drag Related Flow Field Characteristics of Trucks and
Buses, Aerodynamic Drag Mechanisms of Bluff Bodies and Road Vehicles, General
Motors Research Laboratories, pp. 45-93
McCallen R, Couch R, Hsu J, Browand F, Hammache M, Leonard A, Brady M, Salari K,
Rutledge W, Ross J, Storms B, Heineck JT, Driver D, Bell J, Zilliac G (1999) Progress
in Reducing Aerodynamic Drag for a Higher Efficiency of Heavy Duty Trucks (Class
7-8). SAE 1999-01-2238.
Montoya LC, Steers LL (1974) Aerodynamic drag reduction tests on a full-scale tractor
trailer combination with several add-on devices. NASA TM X-56028
Rae WH, Pope A (1984) Low Speed Wind Tunnel Testing
Rodriguez O. (1991) Base Drag Reduction by Control of the Three-dimensional Unsteady
Vortical Structures;. Experiments in Fluids Vol. 11 p218-226
SAE J1252 (1981) SAE Wind Tunnel Test Procedures for Trucks and Buses
SAE J1321 (1981) Joint ATA/SAE Fuel Consumption Test Procedure, Type-II
Saltzman EJ, Meyer RR (1999) A Reassessment of Heavy Duty Truck Aerodynamic Design
Features and Priorities. NASA/TP-1999-206574
Saunders JW, Watkins S, Hoffmann PH, Buckley FT (1985) Comparison of On-Road and
Wind Tunnel Tests for Tractor Trailer Aerodynamic Devices, and Fuel Saving
Predictions. SAE 850286
Sovran G, Morel T, Mason W (1978) Aerodynamic Drag Mechanisms of Bluff Bodies and
Road Vehicles, General Motors Research Laboratories
Watkins S, Saunders JW, Hoffmann PH (1993) Comparison of Road and Wind Tunnel Drag
Reductions for Commercial Vehicles. J. of Wind and Industrial Aerodynamics: 411420
Watkins S, Saunders, JW, Hoffmann PH (1987) Wind Tunnel Modeling of Commercial
Vehicle Drag Reducing Devices: Three Case Studies. SAE 870717
Download