passive drag reduction of square back road vehicles
TRANSCRIPT
Passive drag reduction of square back road vehicles
Alamaan Altaf a, Ashraf A. Omar b, Waqar Asrar a,n
a Mechanical Engineering Department, Kulliyyah of Engineering, International Islamic University Malaysia (IIUM), Jalan Gombak, 53100 Kuala Lumpur,Malaysiab Department of Aeronautical Engineering, University of Tripoli, P.O. Box 13154, Tripoli-Libya
a r t i c l e i n f o
Article history:Received 10 April 2014Received in revised form10 August 2014Accepted 11 August 2014
Keywords:Bluff bodyPassive drag reductionRoad vehiclesCFDElliptic flaps
a b s t r a c t
Bluff body vehicles such as trucks and buses do not have a streamlined shapes and hence have highdrag which can be reduced to make great savings in operational cost. While rectangular flaps havebeen widely studied as both passive add-ons and in active drag reducing systems for bluff bodies,changing the basic geometry of the flap has not been explored in literature. In this work, a baselinedrag value is obtained for a simplified MAN TGX series truck in a CFD software, and the dragreduction of a proposed elliptically shaped flap is compared to aerodynamically equivalentrectangular flaps. The optimal mounting angle for both flaps is found to be 501. A parametric studyof changing the ellipse semi-major axis is carried out to find the optimal length for drag reduction. Amaximum drag reduction of 11.1% is achieved using an elliptical flap with 0.12 m semi-major axis;compared to 6.37% for a length equivalent rectangular flap, and 6.84% for a surface area equivalentrectangular flap. Results of the pressure distribution and velocity flow behind the rear of the truckare also given and analyzed.
& 2014 Elsevier Ltd. All rights reserved.
1. Introduction
While the importance of reducing aerodynamic drag of cars iswell known and researched, the case of bluff (and/or square back)road vehicles such as trucks and buses has received less attention.In Malaysia alone, there are more than 20.18 million road vehicles,of which about 7.57% are heavy and light duty vehicles (Ministry ofTransport Malaysia, 2010). These figures are suggestive of theeconomic size of the trucking industry. The commercial vehiclesector unlike the private car industry discourages frequent releasesof new models due to different market dynamics. New modelstake longer to design, produce and release to market, and areexpected to stay in the market much longer than new models ofcars. Also, due to the function they perform, and because of theirrelatively uniform prevalence all over the world, a great deal ofmoney is spent on trucks and buses, especially in developingcountries. Reduction of the aerodynamic drag of these vehicles canresult in great savings by decreasing fuel consumption.
Overcoming aerodynamic drag on long haul journeys is thecause of most of the fuel consumption of trucks and buses. Heavyvehicles, due to their large frontal area and bluff shapes areaerodynamically inefficient and take up to 65% of fuel to overcome
drag. As mentioned by Hsu and Davis (2010), it is estimated thatwith a drag reduction of about 40% for trucks, a saving of $10,000per vehicle can be made every year.
Aerodynamic drag force on a body is caused by the body'sprofile, and its surface area. The corners of square back vehiclescause air flowing past the moving vehicle to separate, leading to amajor pressure drop which induces a large wake behind thevehicle. The separation can be attributed to two factors – first,the inability of flow to move past sharp corners and second, lack ofenergetic flow at this point. For such bluff bodies, the profile drag(wake) contributes to 80–90% of total drag, while the remainder isdue to skin friction drag. Thus, it is imperative that more attentionbe paid to drag reducing add-ons that reduces the rear wake.
Rectangular flaps have been widely studied, both in isolationaffixed on the roof as well as on all four edges of the rear of a truckto mimic boat-tailing. However, no significant research has beenconducted on the effect of changing the basic shape of the flap.This work proposes the use of an elliptically shaped flap for dragreduction. As far as the authors know, elliptically shaped flapshave not been employed for bluff body drag reduction.
Furthermore, most studies are conducted on a generic Ahmedbody (Ahmed et al., 1984) rather than a realistically modeled truck.In this work, Computational Fluid Dynamics (CFD) simulations areconducted on a simplified truck model based on the MAN TGXtruck series to establish a baseline drag value. An ellipticallyshaped flap is then designed and added on to the roof of the
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/jweia
Journal of Wind Engineeringand Industrial Aerodynamics
http://dx.doi.org/10.1016/j.jweia.2014.08.0060167-6105/& 2014 Elsevier Ltd. All rights reserved.
n Corresponding author. Tel.: þ60361964557.E-mail addresses: [email protected] (A. Altaf),
[email protected] (W. Asrar).
J. Wind Eng. Ind. Aerodyn. 134 (2014) 30–43
truck to study its drag reducing potential. A parametric study ofthe flap angle is conducted to determine the best angle formounting the flap. A comparison is made with two similarly sizedrectangular and triangular flaps – one with a laterally equivalentdimension; and a second with an equivalent surface area. Resultsshowing the velocity distribution and pressure gradient behindthe truck are given. The remainder of this paper is organized asfollows: Section 2 reviews literature on studies of flaps, Section 3provides details of the model and simulations, Section 4 presentsresults and their discussion and Section 5 concludes the paper.
2. Related work
Drag reduction techniques are divided into two main cate-gories: active and passive. Active methods involve energy expen-diture and usually a sophisticated control-feedback mechanism,while passive methods require no energy input and are morerobust. Factors to be considered while choosing either of thesetechniques include industry standardization (weight/size), costeffectiveness (fabrication, installation, driver training, modifica-tions-compatibilities), energy/power consumption, maintenanceand vehicle usage.
Mohamed-Kassim and Filippone (2010) conducted a numericalstudy of fuel saving potential of various drag reducing retrofits.They found that vehicle parameters alone do not affect total drag;operational parameters have a large effect as well. It was foundthat using aerodynamics devices is optimal when these vehiclestravel at high speeds on long haul driving cycles, as weight doesnot have any direct effect on drag. Driving through urban areasutilizes most fuel for acceleration and deceleration. These factorsare important because every part of a truck contributes, positivelyor negatively, to the total drag of the truck. While studying the useof back-flaps on a full-sized truck, they found a reduction of 4–5%of total drag. This is one of the few studies conducted on a full-sized truck, however it did not address the shape or design ofthe flap, or approach the topic of optimizing the dimensions formaximum drag reduction.
Lee and Ko (2008) studied the flow field behind perforatedGurney-type flaps and concluded that perforated flaps are betterthan solid ones to reduce drag and wake width and unsteadiness.However the study was done on flaps alone, without any actualthree dimensional (3D) body. Fourrié et al. (2010) in their experi-mental study using deflector on a generic car model found that dragreduction of up to 9% was obtained depending upon the deflectorangle. Beaudoin and Aider (2008) did the experimental study on a3D bluff body called Ahmed body, a commonly used 3D bluff bodyfor benchmarking purposes, using flaps at all the edges on the tworear surfaces and found that the most efficient configuration wasthe two flaps on side edges of rear slant. Depending on variousconfigurations, the drag could be reduced by 25% and lift by 107%.Ha et al. (2011) carried out an experimental and computationalstudy of the drag reducing capability of a rear downward flap on apickup truck. They found that the drag coefficient (Cd) was reducedwith increasing the flap length. They deduced that the flapdisplaced the flow attachment enabling more downwash, hencereducing the reverse flow in the wake. With the increase indownward angle, there was an increase in drag reduction. Theyproposed that this happens because the cabin back pressureincreases with the increase in downward angle, which reducesthe drag coefficient. However the performance does not stayconstant with increasing flap length and downward angle. It wassuggested that the rear downward flap needs to be designed whichwould have an optimum downward angle.
Active control is usually obtained by blowing, suction, syntheticjets, actuated flows, etc. Nayeri et al. (2009) experimentally
studied the effect of active and passive control in combinationon a generic tractor trailer. Solid flaps with constant/oscillatorysuction and blowing were used. Techniques such as flow visualiza-tion, six component balance, Laser Doppler Anemometry (LDA)and Particle Image Velocimetry (PIV) were applied. It was foundthat the smaller flaps with active control are more efficient thanlonger flaps without active control. Smaller flaps with constantblowing gave the best drag reduction of 8.81%. Also it was foundthat the flaps at the sides of the truck did not produce anysignificant drag reduction. This was attributed to the ground effectand ineffective slot length. However, as is pointed out in the paperthe slot length for blowing was not optimized. Also the interactionof the lower part of wake with the ground boundary layer needs tobe studied further.
Although it is seen that flaps have been widely studied inliterature, little work has been done to investigate the change inthe basic shape and dimensions of the flaps. The next sectiondescribes the methodology used to develop the CFD models of thetruck and attached elliptical flaps.
3. Models
In this section, the baseline and modified models are describedalong with the simplifying assumptions employed and the generalnumerical procedure is outlined. The main simplifications madeare:
i. Simplified geometry: since the major part of drag for bluffbodies is contributed by the frontal area the shape of cab,appendages, underbody, etc. are not considered (Leuschen andCooper, 2009).
ii. Effect of cab-tailer gap is not studied in this work, hence thecab and trailer are considered to be one continuous segment.
iii. Simplified cut tires: the circular geometry of tires is simplifiedto rectangular shape near the road. These cut tires are used(a) to reduce computational load, and (b) rolling motion oftruck is not considered in this work.
iv. Speed: as mentioned in Section 2, Mohamed-Kassim andFilippone (2010), showed that on long hauls, the major partof fuel is used to overcome aerodynamic drag. Thus thecommon highway speed limit of 30 m/s (108 km/h) is usedthroughout the paper.
3.1. Baseline model
The model used for this paper shown in Fig. 1 is based on theMAN TGX series – a common long haul class 8 truck often used totransfer heavy freight. All the models in this work adhere to the
Fig. 1. Baseline truck.
A. Altaf et al. / J. Wind Eng. Ind. Aerodyn. 134 (2014) 30–43 31
dimensions detailed in the blueprints of the truck (MAN Truck andBuses UK Ltd, 2011).
3.2. Elliptic flap
A generic elliptic flap is shown in Fig. 2 and has parameters ofsemi-major axis (W) and semi-minor axis (Le) as indicated in Fig. 3.
Whilst W is assumed to be the width of the truck, it is possibleto choose different values of Le. Five different values of Le arechosen (0.16 m, 0.14 m, 0.12 m, 0.08 m, 0.06 m) and simulated inorder to discern which size reduces drag the most. The elliptic flapis attached to the top rear end of the truck at an angle θ. A study ofthe flaps at different angles is also performed.
3.3. Rectangular flap
A generic rectangular flap is shown in Fig. 4 and has parametersof its width (W) and length (Lr) as indicated in Fig. 5.
Similar to the elliptical flap, the width of the rectangular flap isalso assumed to be the width of the truck. However, its length (Lr)needs to be chosen carefully. For comparison of the shape of theflaps, it is necessary to establish equivalence (control variable)between them in either of two ways. One way is to ensureequivalent surface area of the ellipse and rectangle, as drag alsoincreases due to skin friction. This can be done by setting thesurface areas of the ellipse and rectangle equal to each other asshown in Eq. (1).
Ae ¼ 0:5πWLe ¼ Ar ¼WLr ð1Þ
After solving for Lr, Eq. (2) is obtained.
Lr ¼ 0:5πLe ð2Þ
Another way to ensure equivalent length (Lr¼Le) as the amount ofprotrusion of the flap over the rear of the truck influences thewake dynamics.
3.4. Triangular flap
A triangular geometry can be considered a transitional shapebetween a rectangle and ellipse, having characteristics of both.This is a worthwhile reason to consider triangular flaps, as they areeasier to manufacture than elliptic flaps, but have similar geome-try. Following the methods used in Section 3.3, the design of atriangular flap is given in Fig. 6 and 7 below.
The width of the triangle is the width of the roof of the truck,while the length can be chosen to make its surface area equivalentto the elliptic flap as shown in Eq. (3).
Lt ¼ πLe ð3Þ
In order to make the triangular flap length equivalent to theelliptic flap, the length can be chosen to be the same as shownin Eq. (4).
Lt ¼ Le ð4Þ
After selecting the length of the triangular flap, it is attached to therear end of the truck as shown in Fig. 6.
Fig. 2. Truck with elliptic flap at angle θ.
Fig. 3. Design of elliptic flap.
Fig. 4. Truck with rectangular flap.
Fig. 5. Design of rectangular flap.
Fig. 6. Truck with triangular flap.
A. Altaf et al. / J. Wind Eng. Ind. Aerodyn. 134 (2014) 30–4332
3.5. Perforated flaps
Recent research has shown that using a layer of perforations onthe surface of a truck can increase its air-porosity and reduce drag(Bruneau et al., 2012). Whilst this may be difficult to do on a wholetruck, it is relatively easy to accomplish on a flap. Two differentsizes of perforations (diameters 0.02 and 0.01 m), based on theboundary layer thickness, were used on all three types of flaps asshown in Figs. 8–10 below.
4. Simulation setup
The above mentioned models were drawn using DassaultSystèmes SolidWorks 2010 and imported into Star CCMþ forCFD simulations. A bounding box was created around the truckto define the computational domain. A clearance of 10L was usedbetween the inlet and the front face of the truck; a clearance of20L was used between the outlet and the truck; a clearance of 10Lwas used between the sides and the truck on both sides andclearance of 8Lwas used between the top of the box and the truck;where L is the overall length of the truck, as shown in Fig. 11. Thesurface and volume mesh were generated in Star CCMþ itselfusing a polyhedral mesher, prim layer mesher and surfaceremesher. The prism layers were only extruded from the trucksurface to model boundary layer around it, as shown in Fig. 13. Therest of the mesh consists of the polyhedral volume cells adjoiningthe prism layers. Volumetric control boxes with finer meshes wereused at the front face and at the back of the truck as shownin Fig. 12.
4.1. CFD model validation
The Ahmed body (Ahmed et al., 1984) is one of the widely usedvalidation and benchmarking model used for road vehicles. It isalso a common benchmark case for RANS models. It was firstproposed by Ahmed et al. (1984). They did wind tunnel testing tostudy the flow around the body and analyzed the wake structuredeveloped. It was found that 85% of the drag was due to pressure
Fig. 7. Design of triangular flap.
Fig. 8. Perforated elliptic flap.
Fig. 9. Perforated rectangular flap.
Fig. 10. Perforated triangular flap.
Fig. 11. Bounding box around the truck.
Fig. 12. Mesh around the truck.
A. Altaf et al. / J. Wind Eng. Ind. Aerodyn. 134 (2014) 30–43 33
drag. They also found that the drag of the body depends on theangle of the slant at the rear of the body, which is varied between01 and 401. The shape and dimensions of an Ahmed body areshown in Fig. 14. The Ahmed body with a slant angle of 251 is used.
Due to the geometric symmetry of the body, only half themodel was simulated at velocity of 40 m/s (Re¼2.77eþ6 andboundary layer thickness¼0.0204 m). The total number of cellsgenerated were 525,817 with the minimum cell size of 0.002 m.The mesh around the Ahmed body is shown in Fig. 15. The meshand physics parameters used in the Star CCMþ simulation aresummarized in Table 1.
The simulation was run for 1000 convergent iterations. Theresultant flow around the Ahmed body is shown in Fig. 16. Thedrag coefficient value was obtained and compared to experimentaldata of Ahmed et al. (1984) as shown in Table 2.
The results obtained from the simulation compare well to theresults given in experimental studies found in literature. Thisvalidates the meshing and physics models being used for thesimulation of bluff bodies in this work, and provides strongevidence for the validity and accuracy of the results obtained fromthe simulations.
4.2. Grid independence
Simulations with varying mesh coarseness were performed onthe baseline truck to test grid independency with respect to thecalculated drag coefficient of the truck. Results of the gridindependence test are shown in Table 3. There is a maximumvariation of only 13% between the coarsest and finest meshingused, and as such, the results are assumed to be largely gridindependent. The medium mesh was used thought out this studybecause it was both computationally efficient and making it finerhas little effect on precision of CD value.
5. Results and discussion
This section gives the results of all the parametric studiesconducted using the baseline model and elliptic, rectangular andtriangular add-ons.
5.1. Flap angle
In order to ascertain the best mounting angle for the flaps, theflap angle for all three flap designs was varied, and the dragcoefficient was calculated in each case. The results are given inFigs. 17–19, together with the values for the baseline dragcoefficient.
It can be seen from the graphs that in all cases, minimum dragcoefficient is achieved when the flap is mounted at angle of 501. Itis interesting to note that the CD value for triangular flap dipssharper than the rectangular flap at 50, but not quite as sharply asthe elliptic flap. This once again supports the notion that atriangular flap can individually be considered as “in-between”rectangular and elliptic shaped flaps.
Fig. 13. Mesh near the truck.
Fig. 14. Dimensions of Ahmed body (Hinterberger et al., 2004).
Fig. 15. Mesh around Ahmed Body.
Table 1Mesh and physics parameters.
Mesh property Parameter
Surface mesh Surface remesherVolume mesh PolyhedralBoundary layers PrismsNumber of prism layers 10Prism layers stretching 1.2%Prism layers thickness 0.0204 mSurface growth rate 1.5%Minimum cell size 0.002 mTotal no. of cells 4525,817
Physics properties ParametersFlow characteristics Steady, stationary, incompressible,
turbulent, segregated flow with energy solver, RANSTurbulence model K-OmegaFlow velocity 40 m/s
A. Altaf et al. / J. Wind Eng. Ind. Aerodyn. 134 (2014) 30–4334
5.2. Elliptic flap length
As mentioned in Section 3.2, the length of the elliptic flap is animportant parameter that can be changed. A study of variouslengths (Le) of the elliptical flap was carried out at the optimal 501angle. The results of this are shown in Table 4.
It can be seen that the maximum drag reduction is given by anelliptical flap of Le¼0.12 m.
5.3. Flap shape comparison
Rectangular and triangular flaps of equivalent length andequivalent surface area are also simulated at the same angle forbenchmarking purposes. It is instructive to take note of the patternof air flow past the rear of the truck. This is shown for the baselinemodel of the truck, the elliptic flap, the length equivalent rectan-gular flap, the surface-area equivalent rectangular flap, the lengthequivalent triangular flap, and the surface-area equivalent trian-gular flap in Figs. 20–25(a and b), respectively.
It can be seen in the case of the baseline model that there is alarge, broad wake at the rear of the truck, extending an appreciabledistance behind it. This consists of two counter-rotating vortices, aclose one near the bottom of the truck, and a further one towardsthe top. While all the flaps seem to taper the wake, they do so indifferent ways. In the case of the rectangular flaps, there is mildstreamlining of the flow past the top of the truck. The elliptic flapstreamlines the flow further, and almost causes a slow tapering off
in the wake at a short distance behind the truck. It also seems todistribute the flow in the wake more symmetrically (notice parti-cularly the positioning of the vortices). In the case of triangularflaps, it is seen from the figures that the length equivalent flapaffects the flow behind the rear of the truck less than the surfaceequivalent flap, as it is smaller in length. However, a remarkablechange is seen in the position of the top vortex. While in all othercases, the top vortex is behind the lower vortex, in the case of thesurface area equivalent triangular flap, the top vortex is now in factin front of the lower vortex and quite close to the rear of the truck.The surface equivalent flap affects the wake more, especially bycausing the flow vectors near the middle of the back of the truck tobecome perpendicular to the surface of the truck.
The drag coefficient and correspondingly, drag reduction withrespect to the baseline drag were also calculated. The results forthe different flap configurations at the optimum angle of 501 aregiven in Table 5.
5.4. Flap shape and wake reduction
In order to understand how flap shape affects the wake, it isimportant to be able to quantitatively represent and assess thechanges in the wake. To help in doing this the Star-CCMþ datastructure of a presentation grid is used. Presentation grids are a2-dimensional regularized set of data points on a plane whoseresolution can be specified. In this case, the presentation gridsare primarily used to represent pressure and velocity valuesbehind the truck. Fig. 26 shows two presentation gridsbehind the truck parallel to the yz-plane, at two differentx-coordinates.
To quantify the wake, 44 presentation grids are used, each witha resolution of 100�100 points with a gap of 0.025 m betweenthem in order to section the wake into thin slices.
The pressure and flow behind the truck is visualized at x¼1.4 m(approximately the location of widest wake for baseline model).Fig. 27 shows the pressure distribution on the plane, and thevelocity flow vectors seeded on the plane, for all the cases atoptimum angle.
From the baseline pressure distribution graph, it can be seenthat there are two concentrated regions of very low pressure eachreaching nearly �115 Pa. This is supported by the velocity flowvectors which exhibit two counter-rotating vortical regions. Thisflow is actually towards the direction of the back of the truck asshown in the figures displaying the pattern of airflow past the rearof the truck. In all other pressure distribution graphs, it can be
Fig. 16. Flow around Ahmed body (shown in mid plane).
Table 2Drag coefficients for Ahmed body.
Model Drag coefficient, CD
Ahmed 1984 0.280STAR CCMþ 0.274
Table 3Grid independent test.
Mesh No. of cells CD
Course 298857 0.73Medium 331901 0.64Fine 1 471249 0.63Fine 2 583726 0.66
A. Altaf et al. / J. Wind Eng. Ind. Aerodyn. 134 (2014) 30–43 35
seen that the concentrated low-pressure regions have beenreduced considerably. In particular, no other case reaches thelowest value of �115 Pa. In the case of the rectangular flap withequivalent length to the elliptical flap, the two regions seem tohave merged and a new low concentration region is seen near thecenter – although it is still considerably smaller compared to thebaseline model. In the case of the rectangular flap with theequivalent surface area, no major concentrated region of lowpressure is really discernable. Some patchy areas reach about
Fig. 17. CD for elliptic flap.
Fig. 18. CD for rectangular flap.
Fig. 19. CD for triangular flap.
Table 4CD for varied Le.
Le (m) Drag coefficient Drag reduction (%)
0.16 0.596 7.650.14 0.590 8.630.12 0.574 11.110.08 0.601 6.900.06 0.601 6.92
A. Altaf et al. / J. Wind Eng. Ind. Aerodyn. 134 (2014) 30–4336
�98 Pa, and the velocity flow vectors seem to be moving down-wards as compared to the baseline model, thereby causing slighttapering of the overall wake. The pressure distribution in the caseof the elliptical flap is the most evenly spread out, and showsalmost no regions of significantly low pressure (blue regions). Thelowest pressure is around �90 Pa, with most of the wake nowgreen in color indicating pressures between �60 Pa and �80 Pa.From the velocity flow vectors, it can be clearly seen that the wakeis “closing in”, especially at the top near the roof, where thevelocity vector heads can be seen directing downwards. Thisexplains the wake tapering seen in the Fig. 21. For triangular flaps,although the wake seems to be much smaller in the surfaceequivalent triangular flap in Fig. 25, there is significantly morecross-flow visible in Fig. 27(f), and hence the surface equivalenttriangular flap would be less effective. The length equivalenttriangular flap achieves a drag reduction of 6.43%, while thesurface area equivalent triangular flap manages to achieve a dragreduction of only 4.73%. It can be concluded that the cross-flowand skin friction effects are more significant in the surface areaequivalent flap, likely due to it being longer.
It is possible to sum the pressure values of the individual datapoints on the presentation grid, and obtain a value for the pressuredistribution on the presentation grid. This is done for each of the44 presentation grids, for each case and given in Fig. 28. From thegraph it can be clearly seen that overall the pressure value forthe elliptic flap remains the highest (least drag) and the baselinethe lowest (most drag). Furthermore, it can be seen that thepressure values for the triangular flaps largely remain between therectangular and elliptic flap, again establishing that it is an “inbetween” shape.
If the integral of the graph for each case in Fig. 28 is taken, itwill represent the volume integral of the pressure distributionover the whole region of the wake, which represents the energy ofthe wake. This value is given in Table 6 for each case.
It can be seen from Table 6 that the wake for the truck modelwith an elliptic flap has considerably higher energy and hencelower drag.
It is also possible to export the presentation grids intoMATLAB and subtract one presentation grid from another toobtain the difference in wake between them. Thus, if the pre-ssure presentation grid for the elliptic flap at a particularx-coordinate is subtracted from the pressure presentation gridfor the baseline model at the same x-coordinate, the result willbe the change in baseline wake caused by the elliptic flap.Fig. 29 (a, b andc) shows the presentation grids for the baseline,elliptic, and the subtraction between them, respectively (nega-tive values show increase in pressure, i.e. less drag). This grid istaken at the x-coordinate where the total changed pressurevalue is the highest.
It can be seen from Fig. 29(c) that the two blue circular regionsin the center show the increase in pressure (hence decrease indrag). It is particularly important to notice that once again, theelliptic flap leads to pronounced symmetric pressure increase(reduction in wake size), hence its significant drag reduction. Allthese results strongly indicate that it is the inherent shape of theelliptic flap which helps in symmetric pressure distribution inthe wake.
Fig. 20. Airflow (a) velocity magnitude and (b)vectors behind the baseline truck-without any flaps.
A. Altaf et al. / J. Wind Eng. Ind. Aerodyn. 134 (2014) 30–43 37
Fig. 21. Airflow (a) velocity magnitude and (b)vectors behind truck with elliptic flap at 501. (For interpretation of the references to color in this figure, the reader is referredto the web version of this article.)
Fig. 22. Airflow (a) velocity magnitude and (b)vectors behind truck with length equivalent rectangular flap at 501.
A. Altaf et al. / J. Wind Eng. Ind. Aerodyn. 134 (2014) 30–4338
Fig. 23. Airflow (a) velocity magnitude and (b)vectors behind truck with surface area equivalent rectangular flap at 501.
Fig. 24. Airflow (a) velocity magnitude and (b)vectors behind truck with length equivalent triangular flap at 501.
A. Altaf et al. / J. Wind Eng. Ind. Aerodyn. 134 (2014) 30–43 39
5.5. Perforated flaps
Perforations on the flaps were also tested with two differentperforation diameters. The results are shown in Table 7 below.
It was found that the elliptic flap gave the best drag reductiononce again of 8.86% proving that the novelty of the shape itself is
better than other flap designs. Also it was found that biggerperforation gave higher drag reduction compared to the smallerone, in all three cases.
Fig. 25. Airflow (a) velocity magnitude and (b)vectors behind truck with surface area equivalent triangular flap at 501.
Table 5CD for different flap configurations.
Flap Drag coefficient Drag reduction (%)
Baseline drag 0.646 –
Elliptical (Le¼0.12 m) 0.574 11.11Rectangular flap equivalent length (Lr¼0.12 m) 0.605 6.38Rectangular flap equivalent surface area (Lr¼0.094 m) 0.602 6.85Triangular flap equivalent length (Lt¼0.12) 0.605 6.43Triangular flap equivalent surface area 0.616 4.73
Fig. 26. Presentation grids behind truck.
A. Altaf et al. / J. Wind Eng. Ind. Aerodyn. 134 (2014) 30–4340
Fig. 27. Pressure distribution and velocity vectors on presentation grid behind truck with different flaps. (a) Baseline model, (b) Elliptic flap at 501, (c) length equivalentrectangular flap at 501, (d) surface area equivalent rectangular flap at 501, (e) length equivalent triangular flap (f) surface area equivalent triangular flap.
Fig. 28. Summation of pressure over 44 presentation grids.
A. Altaf et al. / J. Wind Eng. Ind. Aerodyn. 134 (2014) 30–43 41
6. Conclusion
The shape, sizing and design of flaps for drag reduction of bluffbodies is a poorly studied area in heavy vehicle aerodynamics. Inthis paper elliptically shaped flaps are suggested as an add-on toreduce drag for trucks. A TGX MAN long-haul truck is used as abaseline model to establish a baseline drag coefficient value of0.646. The design of an elliptical flap is presented and equivalentlysized rectangular and triangular flaps are also outlined. Gridindependence is established by calculating baseline drag coefficientusing different mesh coarseness. Simulations of the flaps at differ-ent mounting angles suggest an optimal mounting angle of 501 forall three flap designs. The size of the semi-minor axis of the ellipticflap is varied at Le¼0.16 m, 0.14 m, 0.12 m, 0.08 m and 0.06 m.The results for the elliptic flap are found to produce a tapered andmuch smaller wake. A maximum drag reduction of 11.1% is seen inthe case of the elliptic flap, which is a significant increase from thedrag reduction of 6.3%, 6.8%, 6.4% and 4.7% of similarly sized
rectangular and triangular flaps. Also, the effect of perforated flapson the drag reduction was studied and once again elliptic flapshowed better drag reducing capability as compared to rectangularand triangular flaps.
References
Ahmed, S.R., Ramm, G., Faitin, G., 1984. Some Salient Features of the Time –
Averaged Ground Vehicle Wake (SAE-TP-840300). Society of AutomotiveEngineers, Inc., Warrendale, PA.
Beaudoin, J.-F., Aider, J.-L., 2008. Drag and lift reduction of a 3D bluff body using flaps.Exp. Fluids 44 (4), 491–501. http://dx.doi.org/10.1007/s00348-007-0392-1.
Bruneau, C.H., Creusé, E., Depeyras, D., Gillie ́ron, P., Mortazavi, I., 2012. Active andpassive flow control around simplified ground vehicles. J. Fluid Mech. 5 (1),89–93.
Fourrié, G., Keirsbulck, L., Labraga, L., Gilliéron, P., 2010. Bluff-body drag reductionusing a deflector. Exp. Fluids 50 (2), 385–395. http://dx.doi.org/10.1007/s00348-010-0937-6.
Table 6Representative energy of wakes.
R R R ðPressurePGÞ:dv
Baseline model �5.873Eþ06Elliptic flap 2.333Eþ04Length equivalent rectangular flap �2.210Eþ06Surface area equivalent rectangular flap �3.0861Eþ06Length equivalent triangular flap �2.323Eþ06Length equivalent triangular flap �1.810Eþ06
Fig. 29. (a) Baseline wake presentation grid values, (b) elliptic flap wake presentation grid values, and (c) difference of (a) and (b). (For interpretation of the references tocolor in this figure, the reader is referred to the web version of this article.)
Table 7Drag reduction of perforated flaps.
Perforation diameter (m) CD Drag reduction (%)
Elliptic 0.02 0.588 8.860.01 0.599 7.24
Rectangular 0.02 0.611 5.370.01 0.617 4.36
Triangular 0.02 0.612 5.260.01 0.618 4.31
A. Altaf et al. / J. Wind Eng. Ind. Aerodyn. 134 (2014) 30–4342
Ha, J., Jeong, S., Obayashi, S., 2011. Drag reduction of a pickup truck by a reardownward flap. Int. J. Automot. Technol. 12 (3), 369–374. http://dx.doi.org/10.1007/s12239-011-0043-7.
Hinterberger, C., Garcia-Villalba, M., Rodi, W., 2004. Large eddy simulation of flowaround the Ahmed body. The aerodynamics of heavy vehicles: trucks, buses,and trains.
Hsu, F.-H., Davis, R.L., 2010. Drag reduction of tractor-trailers using optimized add-on devices. J. Fluids Eng. 132 (8), 084504. http://dx.doi.org/10.1115/1.4001587.
Lee, T., Ko, L.S., 2008. PIV investigation of flowfield behind perforated Gurney-typeflaps. Exp. Fluids 46 (6), 1005–1019. http://dx.doi.org/10.1007/s00348-008-0606-1.
Leuschen, J., Cooper, K.R., 2009. Summary of full-scale wind tunnel tests ofaerodynamic drag-reducing devices for tractor-trailersThe Aerodynamics ofHeavy Vehicles II: Trucks, Buses, and TrainsSpringer, Berlin Heidelberg, pp.451–462.
MAN Truck and Buses UK Ltd., 2011. Euro 5 Spec Sheets – TGX 6�4 tractor ChassisSpecification.
Ministry of Transport Malaysia. (2010). Transport Statistics Malaysia. Retrieved May29, 2013, from ⟨http://www.mot.gov.my/my/Publication/Official/Statistik Pengangkutan Malaysia 2010.pdf⟩.
Mohamed-Kassim, Z., Filippone, A., 2010. Fuel savings on a heavy vehicle viaaerodynamic drag reduction. Transp. Res. Part D: Transp. Environ. 15 (5),275–284. http://dx.doi.org/10.1016/j.trd.2010.02.010.
Nayeri, C.N., Haff, J., Greenblatt, D., Loefdahl, L., Paschereit, C.O., 2009. Dragreduction on a generic tractor–trailer using active flow control in combinationwith solid flapsThe Aerodynamics of Heavy Vehicles II: Trucks, Buses, andTrainsSpringer, Berlin Heidelberg, pp. 179–191. http://dx.doi.org/10.1007/978-3-540-85070-0_14.
A. Altaf et al. / J. Wind Eng. Ind. Aerodyn. 134 (2014) 30–43 43