Tractor Trailer Aerodynamics

MISSING IMAGE

Material Information

Title:
Tractor Trailer Aerodynamics
Physical Description:
1 online resource (51 p.)
Language:
english
Creator:
Manthri, Vikram
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Master's ( M.S.)
Degree Grantor:
University of Florida
Degree Disciplines:
Mechanical Engineering, Mechanical and Aerospace Engineering
Committee Chair:
INGLEY,HERBERT A,III
Committee Co-Chair:
ROY,SUBRATA

Subjects

Subjects / Keywords:
aerodynamics -- cfd -- tractor -- trailer -- truck
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
Genre:
Mechanical Engineering thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Heavy Truck’s petroleum consumption is around 12% of the total US petroleum usage. At highway speeds, 65% of this energy is consumed by the trucks in overcoming the aerodynamic drag. So, reducing the aerodynamic drag will not only save money but will also make US less dependent on foreign oil. Present available devices like boattails, side skirts, side and roof extenders, gap fillers, splitter plates etc do not reduce the drag significantly. In addition, many of these devices have operation and maintenance problems. So, there is a need for the development of new devices. In order to see the effect of boundary layer manipulating devices like Plasma Actuators at the blunt base of the Trailer, accurate prediction of the wake is necessary as the pressure drag is the major contributor of the aerodynamic drag. In this thesis, a turbulence model study has been performed on GTS to predict the wake structure. It was found that RANS models are not good enough for predicting the wake structure. Furthermore, DES model was also not able to predict the wake similar to NASA experiment but as the simulation was carried out on a coarse mesh, it should be further studied in order to reach to a conclusion.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Vikram Manthri.
Thesis:
Thesis (M.S.)--University of Florida, 2013.
Local:
Adviser: INGLEY,HERBERT A,III.
Local:
Co-adviser: ROY,SUBRATA.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2013
System ID:
UFE0046060:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 TRACTOR TRAILER AERODYNAMICS By VIKRAM MANTHRI A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2013

PAGE 2

2 2013 Vikram Manthri

PAGE 3

3 Dedicated to my family for their love and support.

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank Dr. H.A. Ingley and Dr. Subrata Roy for giving m e the opportunity to work with them and to participate in the research. Their constant guidance support and motivation was invaluable. In addition, I would like to thank all the members of the Applied Physics Research group and my friend Karthik for their academic help and suggestion. Most important of all, I would like to express my gratefulness to my parents, sister and brother for their constant support and blessings.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ....... 4 LIST OF TABLES ................................ ................................ ................................ .................. 7 LIST OF FIGURES ................................ ................................ ................................ ............... 8 LIST OF ABBREVIATIONS ................................ ................................ ................................ .. 9 LIST OF SYMBOLS ................................ ................................ ................................ ............ 10 ABSTRACT ................................ ................................ ................................ ......................... 11 CHAPTER 1 INTRODUCTION: PROBLEM DESCRIPTION ................................ ........................... 12 2 PRESSURE DRAG ON A TRACTOR TRAILER ................................ ........................ 15 3 NASA EXPERIMENT AND FURTHER RESEARCH ................................ ................. 19 NASA Experiment and GTS Model ................................ ................................ ............. 19 Further Research ................................ ................................ ................................ ......... 19 4 PROCEDURE AND CFD CODE ................................ ................................ ................. 21 Procedure ................................ ................................ ................................ ..................... 21 CFD Code ................................ ................................ ................................ ..................... 21 5 EMPTY TUNNEL SIMULATION ................................ ................................ .................. 23 Geometry ................................ ................................ ................................ ...................... 23 Meshi ng ................................ ................................ ................................ ........................ 23 Physics ................................ ................................ ................................ ......................... 23 Results ................................ ................................ ................................ .......................... 24 6 GTS SIMULATION WITH RANS ................................ ................................ ................. 34 Mesh and Physics ................................ ................................ ................................ ........ 34 Results ................................ ................................ ................................ .......................... 34 7 GTS SIMULATION WITH DES ................................ ................................ .................... 42 Mesh and Physics ................................ ................................ ................................ ........ 42 Results ................................ ................................ ................................ .......................... 42

PAGE 6

6 8 CONCLUSION ................................ ................................ ................................ .............. 48 LIST OF REFERENCES ................................ ................................ ................................ .... 49 BIOGRAPHICAL SKETCH ................................ ................................ ................................ 51

PAGE 7

7 LIST OF TABLES Table page 5 1 Results of simulations for obtaining outlet pressure conditions. ............................... 28 5 2 Results of the boundary layer correction simulation. ................................ ................ 29 6 1 RANS simulation results ................................ ................................ ............................. 36 7 1 DES simulation results. ................................ ................................ ............................... 44

PAGE 8

8 LIST OF FIGURES Figure page 1 1 US petroleum consumption ................................ ................................ ........................ 14 1 2 Effect of shape on skin frictio n and form drag ................................ ........................... 14 2 1 Various areas of a tractor trailer contributing to the pressure drag .......................... 18 5 1 Wind tunnel geometry with boundary conditions. ................................ ...................... 26 5 2 Mesh with 30 inflation layers. ................................ ................................ ..................... 26 5 3 Mesh with 45 inflation layers for boundary layer correction simulation. ................... 27 5 4 Mesh with advance function settings. ................................ ................................ ........ 27 5 5 X velocity Isoplot ................................ ................................ ................................ ......... 30 5 6 Pressure Isoplot ................................ ................................ ................................ ......... 31 5 7 Pressure contour ................................ ................................ ................................ ......... 32 5 8 Velocity contours at the symmetry plane ................................ ................................ 33 6 1 X Velocit y contours at the symmetry plane ................................ ................................ 37 6 2 Pressure contours at the symmetry plane ................................ ................................ 38 6 3 Vortex structures at the base of the tr ailer with X velocity contour .......................... 39 6 4 Vortex structures at the base of the trailer with Y velocity contour .......................... 40 6 5 Cp at the cen terline of the truck. ................................ ................................ ................ 41 7 1 X velocity contour ................................ ................................ ................................ ........ 45 7 2 Pressure contour ................................ ................................ ................................ ......... 46 7 3 Vortex structures at the base of the trailer with X velocity contour .......................... 47

PAGE 9

9 LIST OF ABBREVIATIONS BSL Baseline CFD Computational Fluid Dynamics DES Detached Eddy Simulation DOT Department of Transportation FMG Full Multigrid initialization FHW A Federal Highway Administration GTS Ground Tran sportation System HOTR Higher Order Term Relaxation LES Large Eddy Simulation PIV Particle Image Velocimetry RANS Reynolds averaged Navier Stokes RF Relaxation Factor SA Spalart Allmaras SST Shear Stress Transport

PAGE 10

10 LIST OF SYMBOLS Cp Coefficient of Pressure Cd Drag Coefficient k Turbulent kinetic energy m metre Pa Pascal Re Reynolds number sec seconds u X component of vel o c ity v Y component of velocity w Trailer width (32.38 cm) x X coordinate y Y coordinate

PAGE 11

11 Abstract of Thesis Presented to the Graduate Scho ol of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science TRACTOR TRAILER AERODYNAMICS By Vikram Manthri December 2013 Chair: H.A. Ingley Major: Mechanical Engineering p etroleum consumption is around 12% of the total US petroleum usage. At highway speeds, 65% of this energy is consumed by the trucks in overcoming the aerodynamic drag. So, reducing the aerodynamic drag will not only save money but will also make US less de pendent on foreign oil. Present available devices like boattails, side skirts, side and roof extenders, gap fillers, splitter plates etc do not reduce the drag significantly. In addition, many of these devices have operation and maintenance problems. So, t here is a need for the development of new device s In order to see the effect of boundary layer manipulati ng devices at the blunt base of the t railer, accurate prediction of the wake is necessary as the pressure drag is the major contributor of the aerodyn amic drag. In this thesis, a turbulence model study has been performed on GTS to predict the wake structure. It was found that RANS models are not good enough for predicting the wake structure. Furthermore, DES model was also not able to predict the wake s imilar to NASA experiment but as the simulation was carried out on a coarse mesh, it should be further studied in order to reach to a conclusion.

PAGE 12

12 CHAPTER 1 INTRODUCTION : PROBLEM DESCRIPTION The world today is facing serious energy c risis. The dwindling oil and gas resources along with the current state of less reliable renewable energy sources and the increasing gas prices pose a serious challenge on the world countries. According to US D OT [1] and F H W A [2] US consumes 18.84 million barrels of petroleum per day which is 21.6 % of the World petroleum consumption. At the same time US produces only 9.5% of the World petroleum. So, US has to depend on foreign countries for oil. In order to decrease the oil dependency, Energy reduction and conservation methods are the need of the hour. Transportation sector accounts for 67 % of total US petroleum usage. There are around 10.77 million heavy t rucks registered and they travel ~140 billion miles on interstate highways and freeways in a year. These Heavy Trucks consume about 18% of the transportation petroleum which is ~12% of th e total US petroleum usage. The divisi on of US petroleum use for the t ransportation sector can be seen Figure 1 1. Most o f the energy consumed by these t rucks is used to overcome the aerodynamic d rag and the rolling f riction. But the power consumed to o vercome the aerodynamic d rag increases rapidly with increase in the speed of the vehicle as compared to the rolling f riction For a typical highway speed, aerodynamic d rag accounts for 65% of the total power c onsumption [3] According to Wikipedia [4] t he aerodynamic d rag can be categorized in to form drag and skin f riction. The form d rag is due to the size and shape of the object. Bodies with larger fron tal area tend to experience more form d rag than ones with less Skin f riction is due to the friction between the fluid and the surface of the body. Fig ure 1 2

PAGE 13

13 shows the variation of form drag and skin f riction for bodies with different frontal area. It can be seen that as frontal area increases, the body will experience more form drag than skin f riction. For vertical flat plate, the form drag is almost 100 % while the skin f riction is negligible. The problem with t ractor t railer is similar in nature. They h ave large frontal area, both the t ractor as well as the t railer. As a result, the drag experienced by the t ractor t railer is mostly due to the form d rag. Streamlining the body to reduce the frontal area is one of the solutions to reduce the form d rag. This had been employed by the car manufacturers for many years now. Cars have been streamlined for years for reducing the form d ra g [5] [6] Similarly the t ractor had been streamlined to reduce the form d rag in front of the vehicl e [7] but there is a problem with the t railers, the cargo space. For any given dimensions the box shape has the highest volume. As a result, for being able to carry as much as cargo as possible, the containers behind the t ruck cannot be streamlined

PAGE 14

14 Figure 1 1 US petroleum consumption Figure 1 2 Effect of shape on s kin friction and form d rag (Wikipedia)

PAGE 15

15 CHAPTER 2 PRESSURE DRAG ON A TRACTOR TRAILER The re are 4 areas on a tractor t railer which experience pressure drag, the fr ont portion of the tractor, undercarriage of the trailer, gap between tractor and trailer and the rear end of the t railer. The areas can be seen in the Figure 2 1 As the t ruck moves through the air, there is a large pressure drop on front of the t ractor as the air imparts momentum t o it in opposite direction of travel. The amount of pressure loss can be dec reased by the streamlining the t ractor. The gap between the tractor and t railer is another major concern particularly in the cross flow. The air gets entrapped in the gap creating low pressure vortices and imparting the momentum to the t ruck in its opposite direction of travel Various devices have been dev eloped to reduce the entrapment as well as to increase the pressure. Base bleeding [8] involves blowing the air through the base of the t ractor The blowing of air alters t he vortex structures and thus changing the pressure in the tractor t railer gap Cabin side extenders which are typically vertical plates installed on the t ractor base reduce the gap drag by blocking some of the flow entering into the tractor t railer gap. Rooftop aero shield deflectors, t railer mounted vortex stabilizer, gap fillers and sealers are some of the other de vices for reducing the gap drag [9] [10] Most of these devices except the base bleeding technique have maintenance an d operational issues due to which they have not been installed on the modern vehicles. The underc arriage of a tractor t railer hosts a huge number of devices viz. axles, wheels etc which acts as bluff bodies causing the pressure drop. Also, t here is a cro ss flow as well as the axial flow of air The complex geometries and the flow nature make the flow really complex and unsteady in nature Different types of skirts viz. side skirts,

PAGE 16

16 long wedge skirt, short wedge skirt etc [11] have been developed to prevent the crossflow of air entering into the undercarriage and thus reducing the drag. Various front fairings and rear f airings [12] [13] for rear axle have been developed to block the flow from the axle and reduce the drag As the air moves along the t railer, it separates fro m the blunt base of the t railer and forms 2 vortex structures. The low pressure created by this separatio n and the formation of the wake ca uses the drag on the tractor t railer. Some of the methods for reducing this base drag are boat tailing, base bleeding and b oundary layer manipulating devices. Boat tailing involves gradual reduction of the cross section at the rear end of the bluff body ( t railer in this case). This ch anges the way the strea mlines are bent after separating from the blunt base and when they approach the axis of wake causing the pressure increase. Boat tailing also decreases the size of the w ake Many devices have been developed and tested based on this concept. I nflatable boat tail [14] and base flaps [15] [16] [17] are the some of the devices which come under this category. Boundary layer manipulating devices are used for delaying or preventing the separation of boundary layer from the wall. We know that the flow separation occurs when we have a flow against an adverse pressure gradient and the fluid lo ses its momentum in the boundary layer compared to the wall. These devices tend to impart an extra momentum to the fluid in the boundary layer and thus delaying the flow separation. V ortex generators [18] and plasma a ctuators are some of the devices which come in this category.

PAGE 17

17 M ost of the available devices like boat tails, skirts, side extenders, gap fillers and vortex generators do not reduc e the drag significantly. In addition, many of these devices also have operation and m aintenance problems. As a result, there is a nee d for new drag reduction device In this study aerodynamics of a tractor t railer had been discussed and a study of t urbulence models had been per formed. This study can be used further to test the effectiveness of ne w devices to manipulate the wake behind the trailer.

PAGE 18

18 Figure 2 1 Various areas of a tractor trailer contributing to the pressure d rag

PAGE 19

19 CHAPTER 3 NASA EXPERIMENT AND FURTHER RESEARCH NASA Experiment and GTS Model NASA Ames Research Center conducted a study [19] on 1/8 th scale GTS model in their 7 X 10 ft wind tunnel for obtaining experimental data which can be used for the purpose of CFD validation. GTS i s an aerodyn amically simplified model of a tractor t railer. It is a cab over engine design with no tractor t railer gap. It is a 1/8 th scale model of a class 8 tractor t railer with no wheels and is resting on the four cylindrical posts. The experiment incl uded measuring body axis drag, surface pressures, surface hot film anemometry, oil film in terferometry, and 3 D PIV. They studied the effect of boat tails on the tractor t railer drag The wind averaged drag coefficients with and without boat tails were fou nd to be 0.225 and 0.277.Reynolds number study (0.3 2 million) was also carried out and t hey found that that the drag coefficient varied significantly below Re 1 million which involved less variation in base pressure coefficient and more in fr ont of the model. These results were used for validating the CFD simulation results. Further Research K. Salari et al. [20] found that the converging section of the wind tunnel was enough to capture the incoming boundary layer. So, he excluded the settling chamber of the wind tunnel. H e performed simulations using SA and k eps ilon turbulence model. As NASA experiment did not provide data at the outlet, he found the outlet pressure by iterating it until he matched the values of the experiment at the test section. He found that the

PAGE 20

20 pressure coefficien ts predicted at the front, re ar bottom of the truck showed good agreement with the experimental results while at the base of the truck, it sho w ed different trend. He also concluded that the size of the wake behind the trailer is sensitive to the turbulence model. The drag coefficient s that he predicted with k epsilon and SA turbulence model were 0.318 and 0.418 with 21% and 59% error respectively from the experimental value. He mentioned that he faced convergence issues with k epsilon turbulence model. Also, SA turbulence model predic ted the drag coefficient with 44% error for coarse mesh while for medium mesh it was 59 % error. This clearly shows that there is a problem with the SA turbulence model in drag prediction, as generally the drag prediction improve s with refinement. C.J. Ro y et al. [21] found that at R e 2 million, RANS Menter BSL k omega turbulence model (almost identical to Menter k omega SST model) provided good drag estimate while it predicted poorly at low Re 22000. He also found out that the surface pressure predicted is close to everywhere except in the base region. The simulation predicted a symmetrical pair of counter rotating vortices in the vertical central line plane i n the wake while the experimental data showed asymmetric pair. He also suggested that RANS model is not good enough to predict the effect of drag reduction devices which alter the near wake structure. The drag coefficients that he predicted with Menter k w and SA turbulence model were 0.298 and 0.413

PAGE 21

21 CHAPTER 4 PROCEDURE AND CFD CODE Procedure In NASA experiment, the pressure coef ficient had been calculated using a reference pressure at a reference point on the side wall of the wind tunnel. The referen ce point is located at x/w =4.47, y/w=2.59 and z/w= 4.7 where w is the t railer width and the origin is located in front of the GTS at mid plane. The pressure coefficient [21] is defined as Therefore i n order to validate the results with the experimental results NASA Ames 7X10 ft wind tunnel was modeled and the same reference point was used for calculation of pressure coefficient. Also, static pressure was not measured at the outflow of the tunnel in the NASA experiment For obtaining this boundary condition, empty wind tunnel simulations were performed first. I n these simulations, the outflow static pressure value was iterated until the static pressure and Mach number at the reference point reaches the experimental value. Then the boundary layer profile at the entrance to the test section of the wind tunnel was matched. After obtaining the outflow c onditions, GTS simulations with RANS and DES turbulence mode l s were performed. CFD Code For carrying out CFD simulations, Ansys workbench a commercial CFD package, was used. Modeling was done using Designmodeler, Meshing was done using

PAGE 22

22 Ansys Meshing and simulation was performed using Ansys Fluent, all of which ar e integrated into Ansys workbench. Ansys Fluent is a finite volume solver. T he post processing was done in Tecplot360 and CFD post

PAGE 23

23 CHAPTER 5 EMPTY TUNNEL SIMULATION Geometry The geometry of the NASA 7X10 ft e mpty w ind t unnel was obtained from C.J Roy and was modeled using Ansys Designmodeler. It was shown by K. Salari that only a part of the wind tunnel was enough for simulating the flow The wind tunnel has a 15 ft long test section that is 7 ft in height and 10 ft in width Half of the wind tu nnel was modeled about the symmetry plane. The inlet was taken as pressure inlet and outlet as pressure outlet. The bottom wall (road) was taken as no slip wall and the top and right walls were modeled as slip walls. The domain has been extended by a unit length at the inlet and outlet due to convergence issues. The modeled wind tunnel can be seen in Figure 5 1. Meshing Unstructured grid was used for meshin g. The meshing was produced by p atch conforming tetrahedron algorithm along with curvature and proximi ty advanced sizing functions. Inflation layers (boundary layers ) were added on the road. The first inflation layer thickness was given as 3e 6 m (estimated from turbulent flow over a flat plate). The mesh for obtaining outlet pressure simulation included 3 0 inflation layers with 20 % growth rate. Later on, in the boundary layer correction simulation 45 inflation layers with 20% growth rate were included in the mesh. Both meshes can be seen in Figure 5 2 Figure 5 3 and Figure 5 4 Physics The flow was assum ed to be as compressible and the air as an ideal gas. Density based solver was used and the steady state solution was obtained. For

PAGE 24

24 turbulence, RA NS k omega SST model was selected. The flow was initialized with standard initialization and then FMG initiali zation was performed with 5 cycles and a RF of 0.75. Pseudo Transient with a time step of 0.001 sec and HOTR with a RF of 0.25 were also activated. For the first 100 iterations, first order upwi nd scheme was used for discretis ation and then second order up wind scheme was activated. Results Table 5 1 shows the results for obtaining the o utlet pressure and the mesh independent study. Four meshes, coarse mesh with 0.24 m illion cells Medium mesh with 0.44 million cells, Refine1 mesh with 1 million cells and R efine2 mesh with 1.7 million cells have been studied. The simulations involved iterating the outlet pressure until the test section conditions of static pressure = 97582 Pa, M ach number =0 .27 were obtained for each mesh. The residuals were reduced until 10 e 6 and the reference point values were monitored for confirming the steady state solution. The outlet pressure did not vary much (4 Pa) as we refined from coarse mesh to Medium mesh but the outlet pressure did reduce by 350 Pa as we refined the mesh from Medium to Refine1 and then by 200 Pa as we refined the mesh further from Refine 1 to Refine 2. Medium mesh with 0.44 million cells was selected as Mesh independent solution due to limitation in computational resources. In order to match the boundary layer at section with that of experiment, Medium mesh was then modified with additional inflation layers and then flow was simulated. The outlet pressure was again iterated until the test section values were matched with the experiment. It can be said that that boundary layer profile is independent of the outlet pressure as the outlet pressure did not change much. The results from the boundary layer corre ction simulation can be seen in Table 5 2

PAGE 25

25 It can be seen from the x velocity isoplot ( Figure 5 5 ), pressure isoplot (Figure 5 6 ) and t he pressure contour ( Figure 5 7 ) that the velocity and pressure are uniform in the test section of wind tunnel. V velocity contour and W velocity contours (Figure 5 8 ) clearly show that the velocity in y and z directions is negligible in the test section.

PAGE 26

26 Figure 5 1. Wind tunnel geometry with b oundary conditions. Figure 5 2. Mesh with 30 inflation layers.

PAGE 27

27 Figure 5 3. Mesh with 45 inflation layers for boundary layer correction simulation Figure 5 4. Mesh with advance function settings.

PAGE 28

28 Table 5 1 Results of simulations for obtaining outlet pressure conditions. Mesh case inlet stagnation pressure (in Pa) inlet stagnation temperature (in K) test static pressure (in Pa) test velocity (in m/s) Mach no Outlet pressure (in Pa) y+(road ) Pseu do time step (in sec) Nodes Elements /cells coarse case1 87245 240638 102653.2 282 97555 90.224 0.27006 101002.5 0to3 0.001 coarse case2 87245 240638 102653.2 282 97580 89.994 0.26936 101010.6 0 to 3 0.001 coarse case 3 87245 240638 102 653.2 282 97582 89.977 0.26931 101011.2 0 to 3 0.001 medium case1 150781 439975 102653.2 282 97568 90.124 0.26975 101011.2 0to3 0.001 medium case2 150781 439975 102653.2 282 97583 89.99 0.26935 101015.5 0 to 3 0.001 medium case 3 150781 439975 102653.2 282 97582 89.997 0.26937 101015.2 0 to 3 0.001 Refine 1 case1 330510 1014008 102653.2 282 98286 83.475 0.2496 101015.2 0 to 3 0.001 Refine 1 case2 330510 1014008 102653.2 282 97717 88.841 0.26586 100802.5 0 to 3 0.001 Refine 1 case 3 330510 1014008 10 2653.2 282 97582 90.0067 0.26958 100752 0 to 3 0.001 Refine 2 Case1 529967 1696828 102653.2 282 98024 85.97 0.25715 100752 0 to 3 0.001 Refine 2 case2 529967 1696828 102653.2 282 97628 89.635 0.26827 100587 0 to 3 0.001 Refine 2 case 3 529967 1696828 102653.2 282 97582 89.995 0.26935 100567.8 0 to3 0.001

PAGE 29

29 Table 5 2 Results of the boundary layer correction simulation. Mesh Case Nodes Elements/cells inlet stagnation pressure (in Pa) inlet stagnation temperature (in K) test static pressure (in Pa) test velocity (in m/s) Mach no Outlet pressure (in Pa) y+(road) Pseu do time step (in sec) Medium case 1 207121 543661 102653.2 282 97557 90.234 0.27009 101015.2 0 to 0.35 0 .001 Medium case 2 207121 543661 102653.2 282 97580 90.024 0.2945 101022.4 0 to 0.35 0.001 Medium case 3 207121 543661 102653.2 282 97582 90.007 0.2694 101023 0 to 0.35 0.001

PAGE 30

30 Figure 5 5 X velocity Isoplot

PAGE 31

31 Figure 5 6 Pressure Isopl ot p =97582 Pa

PAGE 32

32 Figure 5 7 Pressure contour

PAGE 33

33 A) B) C) Figure 5 8 Velocity contours at the symmetry plane. A) W velocity B) V velocity C) U velocit y

PAGE 34

34 C HAPTER 6 GTS SIMULATION WITH RANS Mesh and Physics Medium mesh has been used for the simulation. Inflation layers have been added to the truck walls. The first layer thickness was 3e 6 m and 35 layers with 20 % growth rate were added. The mesh was refined further at the point of contact between the supports and the road to reduce the sk ewness of the mesh. The outflow pressure obtained from boundary layer correction simulation was used. The flow was assumed to be compressible and air was assumes as ideal gas. Density based solver was used to obtain the steady state solution. RANS k w SST turbulence model was used. The flow was initialized with standard initialization from inlet and then FMG initialization was carried out. First order upwind scheme was used for first 100 iterations and then second order scheme was turned on. The steady sta te solution was obtained by reducing the residuals to less than 10e 6 and was also confirming by monitoring the pressure and velocity values at the reference point. Results The test section static pressure was 97575 Pa where as the empty tunnel simulatio n predicted 97582 Pa. So the re wa s not much change in the pressure with the addition of GTS into the wind tunnel. Mach number at the test section was 0.269 which is same a s experimental value. The drag at 0 degree yaw angle was calculated to be 0.35. The results can be seen in Table 6 1. The x velocity contour in Figure 6 1 shows that there is uniform velocity in the test section. The velocity is low at the base of the truck and has couple of high velocity

PAGE 35

35 regions in front of the truck. The pressure conto ur in Figure 6 2 shows the high pressure region in front of the truck with low pressure region at the rear. There are also a couple of low pressure regions, as the air flows from front to top and bottom surface s of the truck, creating separation Vertical stream wise cut at the centerline of the base of the truck, near the wake region, along with the u, v velocity contours and streamlines c an be seen in Figures 6 3 and 6 4 NASA experiment predicted a large counter clockwise rotating vortex which is cente red at x/w= 8 y/w=0.4. Also a clockwise vortex is suggested at the top right corner of PIV window. RANS simulations predicted a symmetric pair of vortices centered at x/w = 8. Figure 6 4 shows the vertical stream wise cut with v velocity contour. Also, t he experiment predicts a high velocity region n earby the vortex centered at x/w =8.7, y/w=0.6. But RANS did not predict the high velocity region. Figure 6 5 shows the pressure coefficient Cp which is calculated with reference pressure, along the centerline of the truck. The Cp value could not be matched exactly with that of the experiment due to coarse r nature of the mesh. As we can see that RANS failed to predict the wake structure behind the trailer. It predicted a symmetrical vortex structures where as the experiment predicted an asymmetric al vortex structures.

PAGE 36

36 Table 6 1 RANS s imulation results Nodes Elements/cells inlet stagnation pressure (in Pa) inlet stagnation temp (in K) test static pressure (in Pa) test velocity (in m/s) Mach no Ou tlet pressure (in Pa) drag (full) iteration Pseu do time step (in sec) 501866 1282669 102653.2 282 97575 90.081 0.26962 101023 0.305 6703 0.001

PAGE 37

37 Figure 6 1 X Velocity contours at the symmetry plane

PAGE 38

38 Figure 6 2 Pressure c ontours at the symmetry plane

PAGE 39

39 c) Figure 6 3 Vortex structures at the base of the t railer with X velocity contour

PAGE 40

40 Figure 6 4 Vortex structures at the base of the t railer with Y velocity contour

PAGE 41

41 Figu re 6 5 Cp at the ce nterline of the truck.

PAGE 42

42 CHAPTER 7 GTS SIMULATION WITH DES Mesh and Physics The m edium mesh which has been used for the GTS RANS simulation was used The outflow pressure obtained from the boundary layer correction RANS simulation was a pplied The flow was assumed to be compress ible and air was assumed as an ideal gas. DES with k omega SST turbulence model was used. The flow was initialized with standard initialization from inlet. First order upwi nd scheme was used for first few iterati on s and then the second order scheme was turned on. Time step was started with 0.001 sec and was ramped upto 3sec. Each time step had 200 iterations. Courant number was also ramped up from 1 to 3. Results The simulation was run for 247 sec and the monitore d values wer e oscillating between 2 bounds. The residuals were in the order of 10e 2 but the net mass flux was 3% of the inlet mass flux. So, the solution was assumed to have reached to the steady state. Figure 7 1 shows the results of the simulation. As t he outlet pressure obtained from the RANS simulation was used, the pressure and the Mach number predicted at the reference point were 98362.5 Pa and 0.24894. But the test conditions in the experiment were 97582 Pa and 0.27. A drag coefficient of 0.24 was c alculated. The lower velocity at the test section compared to the experimental value also contributed to the under prediction of the drag coefficient.

PAGE 43

43 Figure 7 2 and Figure 7 3 shows the x velocity contour and pressure at the symmetry plane. Along with the wake behind the Trailer, there is a flow separation at the top of the truck creating a low pressure region. Figure 7 4 shows the vortex structures at the base of the truck. There are 2 vortices, the top rotating in counter clockwise and the bottom on e clockwise. The t op vortex structure is large and is centered near the base of the truck. C.J.Roy et al. DES simulation [22] predicted that the size and nature of the vortex structures is highly dependent on the mesh. His simulation with coarse mesh with 3.8 million cells predicted two asymmetrical vortex structures while the refine mesh with 13.2 million cells predicted the symmetrical vortex structures. Importantly, none of the simulat ion predicted the wake similar to the experiment. This shows us that DES simulation has to be further studied.

PAGE 44

44 Table 7 1 DES simulation results. Nodes Elements /cells inlet stag nation pressure (in Pa) inlet stag nation temp erature (in K) test static pres sure (in Pa ) test vel ocity (in m/s) Mach no Outlet pres sure (in Pa) drag (full) y+(road) 501866 1282669 102653.2 282 98318 84.744 0.25338 101023 0.24 0 to 0.5 98407 81.787 0.2445 98362.5 83.2655 0.24894

PAGE 45

45 Figure 7 1 X velocity contour

PAGE 46

46 Figure 7 2 Pressure contour

PAGE 47

47 Figure 7 3 Vortex structures at the base of the t railer with X velocity contour

PAGE 48

48 CHAPTER 8 CONCLUSION RANS simulations showed c learly that they are not good enough for predicting the vortices structure behind the t railer. The results of the DES simulation were also not in accordance with the experimental results. But as the simulation was carried out on a coarse mesh due to comput ational resource limitation, it should be further investigated before moving onto LES turbulence model. If the DES/LES simulations were able to predict the wake similar to that of the experiment then one can analyze the effect of boundary layer manipulatin g devices like Plasma actuators on the wake.

PAGE 49

49 LIST OF REFERENCES [1] S. C Davis and et.al, Transportation Energy 2012. [2] www.fhwa.dot.gov [3] W. D. Pointer, F. Browand, M. Hammache, T. Y. H su, M. Rubel, P. Chatalain, R. Effort to Reduce Truck Aerodynamic Drag Joint Experiments and Computations AIAA Fluid Dynamics Conference 2004. [4] www.wiki pedia.org [5] Annual Review of Fluid Mechanics vol. 38, no. 1, pp. 27 63, Jan. 2006. [6] The MIT Press vol. 7 no. 4, pp. 309 317, 1974. [7] K. R. Cooper, Applied and Computational Mechanics 2012. [8] 2009. [9] Scale Wind Tunnel Tests of Second Generation Aerodynamic Fuel Saving Devices for Tractor [10] Experiment al Results for a Generic Tractor Trailer in the Ames Research Center 7 by 10 Foot and 12 NASA/TM July 2006. [11] AERODYNAMIC AID FOR THE UNDERBODY OF A TRAILER WITHIN A TRACTOR BBAA VI International Colloquim on Bluff Body Aerodynamics pp. 20 24, 2008. [12] SAE International, 2004 [13] J. Leuschen and K. Scale Wind Tunnel Tests of Aerodynamic Drag Springer, Applied and Computational Mechanics, Volume 41 2009.

PAGE 50

50 [14] Tools for As Springer, Applied and Computational Mechanics, Volume 41 2009 [15] La wrence Livermore National Laboratory reports, 2003 [16] Lawrence Livermore National Laboratory reports, 2010 [17] D. G. Hy ams, K. Sreenivas, R. Pankajakshan, D. Stephen Nichols, W. Roger Computers & Fluids vol. 41, no. 1, pp. 27 40, Feb. 2011. [18] J. L Scale Wind Tunnel Tests of Production and Prototype Second Generation Aerodynamic Drag SAE International, 2006 [19] e Ground Model Transport ation (GTS ) in the NASA Ames 7 by 10 NASA NTRS, 2001 [20] Trailer Lawrence Livermore National Laboratory reports, September, 2003. [21] C. J. Roy, J. Payne, a nd M. McWherter Journal of Fluids Engineering vol. 128, no. 5, p. 1083, 2006. [22] Sprin ger, Applied and Computational Mechanics, Volume 41 2009.

PAGE 51

51 BIOGRAPHICAL SKETCH Vikra m Manthri was born in Andhra Pradesh, India. He completed his Bachelor of Technology in Mechanical Engineering from National Institute of Technology (NIT) Kurukshetra, Haryana, India in June 2008. Later on, he worked in Reliance Infrastructure Limited as Assistant Manager until June 2011. In August 2011, he joined Unive rsity of Florida to pursue his m aster ngineering. He got th e opportunity to work with Dr. Subrata Roy and Dr. H.A. Ingley in s pring 2013 and then worked in Applied Physics Research Group. After completing his master s Vikram plans to contribute to the engineering industry.