Effect of pavement surface and structural conditions on rolling resistance and fuel consumption:

Evaluation of empirical and mechanistic models

Karim Chatti, Ph.D.Department of Civil and Environmental Engineering

Michigan State University

ISAP Technical Committee: Pavement Field Evaluation (TC-PFE)ISAP Day at TRB (January 7 2018)

AcknowledgmentsDGMENT OF SPONSORSHIP

SPONSORSHIP1. NCHRP1-45 Project2. University of California Pavement Research Center (John Harvey)

California Department of Transportation 3. USDOT UTC Center for Highway Pavement Preservation

RESEARCH COLLABORATIONS1. Tom Papagiannakis, Jorje Prozzi and Jolanda Prozzi, Stefan Romanoschi

MSU postdoctoral, graduate, and undergraduate students2. John Harvey and group (UCPRC)

Erdem Coleri (Oregon SU)Argavan Louhghalam (U. Mass)

3. Roozbeh Dargazany, Nizar Lajnef (MSU), Imad Al-Qadi and Hasan Ozer (UIUC)

INSTITUTIONAL COLLABORATIONS1. Texas DOT, Michigan DOT, NCAT2. UC Davis, Oregon State U., U Mass.3. UIUCCLA

Factors Affecting Fuel Consumption

• Vehicle

• Pavement (Rolling Resistance)

3

Thermodynamic efficiency of the engineAerodynamicsWeightTechnological characteristics of the tire:

GeometrySurface characteristics**Structural behavior of the pavement*

• Inflation Pressure• Temperature• Design, materials, dimensions

4Chatti & Zaabar (2012)

5

Field tests experimental matrix

6

Loading conditions

Light truck Articulated truck

6,000 lbs of concrete blocks 45,000 lbs of steel sheets

7

Effect of Surface Characteristics (ANCOVA)

Vehicle class

Speed (km/h)

Summer Winter

Significance (p-value)* Number of Data Points

Significance (p-value)* Number of Data Points IRI MPD Grade lack of fit IRI MPD Grade lack of fit

Medium car

56 0.00 0.54 0.00 0.77 136 0.00 0.81 0.00 0.61 136 72 0.00 0.22 0.00 0.78 146 0.00 0.13 0.00 0.56 146 88 0.00 0.90 0.00 0.75 136 0.00 0.84 0.00 0.78 136

Van 56 0.00 0.35 0.00 0.96 136 0.00 0.83 0.00 0.79 136 72 0.00 0.38 0.00 0.68 146 0.00 0.21 0.00 0.97 146 88 0.00 0.29 0.00 0.77 136 0.00 0.22 0.00 0.78 136

SUV 56 0.00 0.20 0.00 0.75 136 0.00 0.61 0.00 0.86 136 72 0.00 0.40 0.00 0.86 146 0.00 0.83 0.00 0.91 146 88 0.00 0.70 0.00 0.71 136 0.00 0.35 0.00 0.88 136

Light truck 56 0.00 0.50 0.00 0.79 136 0.00 0.96 0.00 0.75 136 72 0.00 0.60 0.00 0.75 146 0.00 0.40 0.00 0.86 146 88 0.00 0.10 0.00 0.77 136 0.00 0.72 0.00 0.71 136

Articulated truck

56 0.00 0.03 0.00 0.79 137 No tests were conducted in winter 72 0.00 0.06 0.00 0.97 146

88 0.00 0.07 0.00 0.78 137

8

Effect of pavement type on fuel consumptionSummer Winter

Sig. Not Sig. Sig. Not Sig.Passenger Car √ √VAN √ √SUV √ √Light Truck √* √† √

Articulated Truck √* √† N/A

* Trucks driven over AC at 56 km/h (35 mph) consume more than trucks driven over PCC in summer conditions

† Not significant at 72 and 88 km/h (45 and 55 mph)

9

Articulated Truck - Summer Condition (30oC)

10

Light Truck – Summer Condition (30oC)

11

NCHRP 720 (HDM 4) Model

Aerodynamic forces

Rolling resistance

Gradient forces

Curvature forces

Inertial forces

( )1000

ircgatr

FFFFFP

++++=ν

2*****5.0 υρ AFCDCDmultFa =

gGRMFg **=

−

= −3

22

10**

***

,0maxCsNw

egMR

M

Fc

υ

( )( )2*13*12*1*11**2 υbMbCRNwbFCLIMCRFr ++=

aaaaMFi *2arctan*10* 3

+=υ

Tractive power

[ ]DEFaIRIaTdspaaKcrCR *3*2*1022 +++= Surface factor

12

13

Step 1: All the PCC site data were used to calibrate the model for PCC pavements (a3*DEF=0). Step 2: The DEF values were backcalculated for all the AC site data points. Step 3: The new DEF values were used to estimate a relationship between deflection, speed and

temperature for trucks. Step 4: All backcalculated deflection parameter (DEF) data were regressed versus speed. Step 5: A linear function was assumed for the temperature adjustment factor. This assumption is

based on the AASHTO1993 overlay design method.

Calibration of the HDM 4 model

00.050.1

0.150.2

0.250.3

0.350.4

0.45

0 20 40 60 80

Def

lect

ion

(mm

)

Speed (km/h)

All dataAT positive dataAll data medianAT medianAT all data

00.20.40.60.8

11.21.4

-10 0 10 20 30 40

Adj

ustm

ent f

acto

r

Temperature ( ͦC)

0

0.1

0.2

0.3

0.4

0.5

0.6

0 20 40 60 80 100 120

Def

lect

ion

(mm

)

Speed (km/h)

-1001020304050

0

2

4

6

8

10

12

14

16

0 20 40 60 80 100

Diff

eren

ce in

fuel

con

sum

ptio

n (%

)

speed (km/h)

15 ͦ C30 ͦ C

Effect of roughness on fuel consumptionHDM 4 versus regression data

Before calibration After calibration14

Effect of texture on fuel consumptionHDM 4 versus regression data

15

EFFECT OF PAVEMENT SURFACE ROUGHNESS ON ROLLING RESISTANCE AND FUEL ECONOMY:

Evaluation of different rolling resistance models

16With Imen Zaabar, Ph.D.

Initial attempts to infuse mechanistic analyses in NCHRP 720 HDM4 model (Zaabar & Chatti 2014)

17

Original NCHRP 720 HDM4

Modified NCHRP 720 HDM4 using DLC

Modified NCHRP 720 HDM4 using DLI

Recent work (Louhghalam et al. 2015)

18

Figure 1. Calibrated Louhghalam et al. model: Roughness-induced change in fuel consumption as function of IRI at V = 70 and 100 km/h for a medium car (Louhghalam et al. 2015)

Vehicle parameters per axle (Louhghalam et al. 2015)

Recent work (Kim et al. 2017)

19

Mechanistic ApproachMichigan State Univ.

20

10-3

10-2

10-1

100

101

102

Wavenumber (cycle/m)

10-14

10-12

10-10

10-8

10-6

10-4

10-2

Elev

atio

n sp

ectra

l den

sity

(m2

. m

/cyc

le)

IRI = 1 m/km

IRI = 3.5 m/km

Example of simulated surface profiles

21

IRI=3.5 m/kmIRI=1 m/km

0 20 40 60 80 100 120 140 160

Distance (m)

-20

-15

-10

-5

0

5

10

15

20

Elev

atio

n (m

m)

0 20 40 60 80 100 120 140 160

Distance (m)

-20

-15

-10

-5

0

5

10

15

20

Elev

atio

n (m

m)

10-3

10-2

10-1

100

101

102

Wavenumber (cycle/m)

10-14

10-12

10-10

10-8

10-6

10-4

10-2

Elev

atio

n sp

ectra

l den

sity

(m2

. m

/cyc

le)

IRI = 1 m/km

IRI = 3.5 m/km

Vehicle Simulation: ¼ car model

Passenger car: - vehicle parameters for ½ of an axle- 4 wheels

Articulated truck: - vehicle parameters for ½ of an axle with dual wheels

(Maxle=8.8t)- 18 wheels

22

Vehicle Simulation: ½ car model

23

Passenger car Heavy truck

Vehicle Simulation: full car model

Calculation of Fuel ConsumptionEnergy dissipated:

- ξb = Calorific value of fuel X engine efficiency= 10.5 MJ/L for passenger car= 16 MJ/L for articulated truck

24

Total fuel consumption:

Passenger car: ( ) ( )( )

0 0* * IRI iIRI i IRI i

b

DFuel Fuel Nw Fuel Fuel Nw

ξ= + ∆ = +

Articulated truck: ( ) ( ) ( )( )

0 0/ 4.4 * * IRI iIRI i IRI i

b

DFuel Fuel M Fuel Fuel N

ξ= + ∆ = +

- Fuel0 = total fuel consumption assuming no rolling resistance from the NCHRP 720 model

Results for Passenger Car

25

1. The mechanistic approach is able to predict (without calibration) the field-observed effect of roughness up until 3.5 m/km.

2. Difference could be caused by the full contact at all times hypothesis between the tires and the surface, which could be violated at higher levels of roughness.

Results for Passenger Car

26

1.00

1.02

1.04

1.06

1.08

1.10

1.12

1.14

1.16

1.18

1.20

1.22

0 5 10 15 20 25 30 35 40No

rmal

ized

fuel

cons

umpt

ion

Speed m/s

1.1

2.3

3.4

4.6

5.7

IRI (m/km)

3. Effect of roughness increases then decreases as the vehicle speed increases, with the maximum being at about 20 m/s

20 9.8 0.647

Vf HzDπ π

= = =×

Stimulus frequency

Unbalanced wheel state

10.7 NWHHz f≈ =

Figure. Effect of speed on fuel consumption at different roughness levels

Results for Articulated Truck

27

1. Reasonable agreement (without any calibration)

2. There are other sources of energy dissipation in the suspension such as:• Friction between various

complex components of a truck suspension.

• Non-linearity in the suspension, etc.

EFFECT OF PAVEMENT SURFACE TEXTURE ON ROLLING RESISTANCE AND FUEL ECONOMY:

Evaluation of different rolling resistance models

28

NCHRP 720 report Texture Model

29

Relationships between pavement surface MPD and a) Adjustment factor for fuel consumption, b) Adjustment factor for rolling resistance, and c) Rolling resistance force at v=80km/h

The rolling resistance adjustment factor is defined as AMPD = x / AMPD = 0.5 , at the speed of 80km/h, temperature of 30°C, IRI of 1 m/km and grade of 0%

MIRIAM Project (2012)Fr= Cr × m × 9.81

Cr = CR0 + IRI × v × Cr1 + MPD × Cr2, with Cr0 = Cr00 + CrTemp (5-T)m = vehicle mass (kg) and Cr0, Cr1, and Cr2 = rolling resistance parameters.

The effect of texture as reported in the MIRIAM study is significantly higher than that reported in the NCHRP 720 report.

For example, at a speed of 90 km/h, alignment standard of 1, increasing a unit value of MPD (1 mm) will cause: • 2.8% increase in the total fuel consumption for a car, • 3.4% for a truck, and • 5.3% for a truck with a trailer. • These values are much higher for the rolling resistance force. 30

31

Boere Study (2009)

Boere Study (2009)

• The tire-pavement surface texture interaction is accounted for by applying a nonlinear contact stiffness to the road.

• Numerical results obtained from the model were compared to measurements using a trailer on test tracks in the Netherlands.

• Measurements are reported in terms of a rolling resistance coefficient (ratio between rolling resistance and axle forces).

• Root mean square (RMStex) of the surface profile is used as the surface texture measure.

• Regression equations for: Numerical model: CRR = 0.001 RMS + 0.0135Empirical model: CRR = 0.001 RMS + 0.0087

32

Experimental and numerical relationships between rolling resistance and texture RMS at v=80km/h on 30 test tracks (Boere, 2009)

Comparison of NCHRP 720 Model with Numerical & Experimental Results by Boere• Rolling resistance Adjustment factor for a light truck (4.1t) at v=80 km/h:

33

• NCHRP 720 results and the experimental results reported by Boere are very close. • Boere’s numerical model predicts the effect of texture on rolling resistance very well.• The difference in the slope shown here is due to the effect of the intercept, since the

model was not calibrated

MPD = 1.729 RMS + 0.019 (Avaik et al. 2013)

Comparison of NCHRP 720 and MIRIAM Models

• Rolling resistance adjustment factors for:a) car (1.5t):

b) heavy truck with trailer (30t):

• The effect of texture is significantly higherin the MIRIAM model. 34

Summary Comparison of RR Texture Models• The increase in rolling resistance with a 1mm increase in MPD for the different

models are summarized below:

• NCHRP 720 predictions agree with the independent rolling resistance measurements reported by Boere, suggesting that the predictions from the MIRIAM model are unreasonably high.

35

Car Light truck Articulated truck

NCHRP-720 3.23% 6.08% 6.08%

MIRIAM 17% - 25%

Boere-experiment - 6% -

Boere-numerical - 4% -

The objective of this work is to investigate the effect of pavement surface texture and roughness on tire rolling resistance using mechanistic approach.

Different components of tire rolling resistance: • Tire bending• Tread slippage• Tread deformation

Scope of the study:• Modeling of surface texture• Characterizing the rubber material• Influence of pavement micro-texture on

rubber block deformation and tread slip• Influence of pavement macro- and mega- texture

on tire deformation (ongoing work)• Influence of pavement roughness on tire and

vehicle suspension system36

Objective and scope

Pavement texture

Rubber material characterization

Micro-scale modeling

Ongoing and future work

On-going Work: Mechanistic Modeling of the Effect of pavement texture on tire rolling resistance

Bending Hysteresis

Tread Slip Hysteresis

Tread Deformation

Grip

AdhesionHysteresis

Shear

Slip Speed

Rolling Resistance 45

IMPACT OF PAVEMENT STRUCTURAL RESPONSE ON ROLLING RESISTANCE AND FUEL ECONOMY

37Danilo Balzarini (Ph.D. candidate)

38

SRR can be defined as the energy required for a rollingwheel to move uphill.The positive slope is caused by the delayed deformation of the pavement.

Structural Rolling Resistance (SRR)

Chupin et al. (2013)

39

Flexible Pavement Response: ViscoWave II-M

-4 -2 0 2 4 6Space[m]

0

50

100

150

200

250

Z D

ispl

acem

ent[m

icro

nmet

ers]

Vehicle Energy Loss – Flexible pavement

Assumptions: Non-dissipative tires Constant vehicle speed

1

ni

diss i ii i

wW p Sx=

∆=

∆∑

40

Traffic direction

dissRR

b

WFuelξ

=

100 100dissRRexcess

C b C

WFuelFuelFuel Fuelξ

= ⋅ = ⋅⋅

ξb is the calorific value of the fuel.

HMA SectionsCaltrans-UCPRC Study

41

Dynamic back-calculation of Relaxation modulus Et, complex modulus E*, and shift factor from FWD time histories (VISCOWAVE-II GP)

Mechanistic PredictionsEffect of Structural Capacity and Season

42

Mechanistic PredictionsEffect of Speed, Temperature and Vehicle Weight

43

HMA vs. PCC: Mechanistic Predictions

44

Rigid Pavement Response - DYNASLAB (Chatti, 1992)

Any Questions?45

Thank you!