# Dedicated to the 80th Anniversary of Prof.

Radoš Řezníček

**Influence of soil and tire parameters on traction
**

M. Schreiber1, H.D. Kutzbach2

1 2

AGCO GmbH, Product Marketing Manager (MP), Marktoberdorf, Germany University of Hohenheim, Department of Agricultural Engineering, Stuttgart, Germany

Abstract: The drawbar pull, travel reduction (slip), and rolling resistance are the main criteria to describe the traction behaviour of off road vehicles. Besides the engine performance, the drawbar pull is influenced by the traction conditions such as soil and the tire parameters. These traction conditions have to be described by a limited number of parameters which can be easily determined. Empirical equations were used to analyse roughly 850 traction curves measured and published by Steinkampf. As a result, the important parameters to describe the traction conditions are three tire parameters (radius, width, inflation pressure) and five soil parameters (soil cover, upper soil strength, lower soil strength, clay content, moisture content). These parameters with relative values between 0 and 100% are used to establish the equations for the traction prediction. Main steps to achieve this goal are the extension of the traction slip equation by a linear term of slip, and the description of this curve by 4 meaningful characteristic coefficients: the x- and y-coordinates of the κ-maximum (σκmax, κmax), the y-axis intercept ρe, and the gradient of κ at zero slip (κ’(0)). Keywords: tire; traction; wheel-slip

The tire and soil parameters influence the traction performance of farm tractor tires (Schreiber 2006). In 1956, Bekker laid the foundation for scientific investigation of soil-wheel interaction mechanism and extended his model in the following years (Bekker 1956, 1960). Numerous attempts followed to quantify the soil-traction device interaction in order to set up models for the traction prediction (Upadhyaya & Wulfsohn 1990). Wismer and Luth (1973) used the Cone Index CI as the only soil parameter and considered the tire width d and tire diameter b in the wheel numeric Cl × d × b Cn = –––––––––– Fz

where: Fz – wheel load

**σ = –––––––– = 1 – –––––––– vth ω × rdyn
**

where: MT – input torque rdyn – relevant dynamic tire radius vth – speed without slip v – actual speed of the tire

vth – v

v

(4)

(1)

and established the following equation for the gross traction ratio µ with the travel reduction (slip) σ µ = 0.75 (1 – e–0.3 Cn × σ)

where: MT µ = –––––––––– rdyn Fz RES. AGR. ENG., 54, 2008 (2): 43–49

(2)

For obtaining the basic information, as well as for the validation of the traction prediction equations, field measurements are unavoidable of traction and rolling resistance of agricultural tires under different traction conditions. Different single wheel testing devices have been developed for these measurements (Steinkampf 1974; McAllistar 1979; Armbruster & Kutzbach 1989; Du Plessis 1989; Upadhyaya et al. 1993; Smulevich et al. 1994). Upadhyaya et al. (1989) conducted extensive field tests using a single wheel tire tester and found that the traction test results always fitted the equations of the following type for the net traction ratio with good correlations. χ = a(1– e–c × σ)

where: κ – net tractrion ratio σ – slip

(5)

(3)

43

κ = a1 – b1 × e–c1 × σ – d1 × σ ρ = a2 + b2 × σ (10) (11)
After the coefficients σpull. While a2 and b2 in Eq.8 Traction performance ratio 0. As shown by Schreiber & Kutzbach (2007). coefficients a1 to d1 in Eq. because Eqs (8) and (9) fit the measurements very well and a large number of measurements providing a good description of the tire and soil parameters are available. coefficients a1 to d1.
Converting the coefficients
For the description of the dependence of the rolling resistance ratio and net traction ratio on the slip. They are based on Steinkampf ’s Eqs (8) and (9). (4)). b2 is the gradient of the rolling resistance ratio ρ’). (11) are concrete values (a2 is the rolling resistance ratio at zero-slip ρ(0).and y-coordinates of the local maximum (σ κmax and κmax).Dedicated to the 80th Anniversary of Prof. ENG. Net traction ratio and rolling resistance ratio over slip
The third important value for the tractive behaviour. which is important as the measurements show a maximum of the net traction ratio at less than 100% slip. The net traction ratio κ and the rolling resistance ratio ρ are calculated as shown in Eqs (6) and (7). like how large is the net traction ratio in the maximum. These characteristic values are the x. κ’(0) and ρ(0) give clear information about the tire behaviour. 2008 (2): 43–49
. or how fast the slip increases with increasing tractive forces. Schreiber & Kutzbach 2007). which means in relation to the local maximum of the curve. For the description of describe the traction behaviour of tires. They were chosen to define exactly the same curve (Figure 1) as the 4 coefficients a1 to d1. κmax. Eqs (10) and (11) provide good results. To calculate the values for different slip and displaying them as in Figure 1.6 0. the κ-equation being extended by the term – d1 × σ (Scheiber 2006.4 0.2 0
Slip σ (%)
60
Figure 1. ρe and κ’(0) have been calculated (Eqs (20)–(26)). κmax. this can be done in 4 steps: (1) The absolute gradient κ’(0) has to be converted into the standardized gradient κsta’(0). like the maximum shear stress and characteristics of the contact patch. they have to be converted into coefficients a1 to d1 in Eq.
MATERIAL AND METHODS New proposal for the function of net traction ratio
parameters. κ = a– b × ec × σ ρ=a+b×σ (8) (9)
The coefficients a. With this linear component. The mathematical way to calculate these coefficients from the characteristic values is shown in the next chapter. This approach is used as the base of the presented equation for traction. a local maximum can be displayed. at which slip this maximum occurs. AGR. The values σκmax. 54.
RES. a2 and b2 must be calculated using the tire and soil
44
(2) The backup-parameter p has to be solved numerically by equation (13). Radoš Řezníček
He used different semi-empirical methods and linear regressions to determine the empirical coefficients a and c with the measured parameters. the y-axis-intercept ρe (external rolling resistance) and the gradient of κ at zero slip (κ’(0)).. which are defined by 3 and 2 coefficients (Steinkampf & Jahns 1986). as shown by Schreiber & Kutzbach (2007). (10). especially for the tractive efficiency.2 0. (10) are not that demonstrative. the curves for the tire behaviour are already defined.0 –ρe 20 40 κ'(0)
(σκmax|κmax)
κ
ρ
–0. is the slip σ (Eq. FX κ = ––––– Fz FR ρ= ––––– Fz (6) (7)
0. the tractive force Fx and the rolling resistance force FR are often based on the tire load Fz. b and c describe the tire behaviour. The same equation can be defined by different 4 coefficients. The recpective curves are shown in Figure 1. The dependence of the rolling resistance ratio ρ and the net traction ratio κ on the slip can be described by Steinkampf ’s empirical Eqs (8) and (9). σκ max κ'sta (0) = κ' (0) ––––––––– κmax + ρe (12)
To predict these curves.

They are displayed in Table 1. satisfactory values can be calculated also by the approximated inverse function. all the parameters having relative values between 0 and 1 (0–100%). in which the tests of the database were divided. 2008 (2): 43–49
Figure 3. the values can be fitted linearly.6 0. the model function input parameters were chosen differing from the database parameters. all 850 values are used without looking for ceteris paribus conditions. tire load. inflation pressure. …).73 1. ploughed. culti1.2 0.194
(14)
vated.6 0. parameters a1 to d1 can be determined and used in Eq. This provides a large database to fit the empirical functions and calculate the specific parameters from Figure 1. soil surface. the maximum net traction ratio is displayed as a function of the clay content (Figure 2). grass land. Steinkampf & Jahns (1986) observed the following tire-characteristics in his tests: tire labelling. many
1. (10) to calculate the curves for the net traction ratio. moisture content and pore volume specify the traction tests.26 × cc + 0. …).48 40 60 Clay content cc (%) 80 100
To figure out the influence of the soil and tire parameters on the specific coefficients. the linear approximation gives a good result. As shown.17 × x + 0.Dedicated to the 80th Anniversary of Prof.4 0. As the soil parameters.2 0. however. κmax + ρe b1 = ––––––––––––––– (15) 1 – p (1 – ln (p)) ln (p) c1 = – ––––––– σκ max (4) Calculate a1 and d1 using Eqs (17) and (18).
RESULTS Soil and tire parameters to calculate the coefficients
(3) The next step is to calculate b1 and c1 using Eqs (15) and (16).4 0.683 ––––––––––––––––––– –0. p≈e
ln (κ'sta (0)) – 0. AGR.
Determination of the coefficient
To estimate these new. running direction.and soil-conditions. preceding crop. the first step was to calculate them for 850 curves measured (Schreiber & Kutzbach 2006).6
Maximum net traction ratio κmax
Figure 2. 54.0 0.0 0 20 κmax = 0. lug height. but for different ceteris paribus conditions this gradient can differ. For an improved practical use. The vertical alignment of the points results from the soil classes. characteristic parameters for different tire. Another possibility to evaluate the relations is to use only tests with ceteris paribus conditions. the large number of values compensates this effect and the trend can be evaluated by the gradient of the linear fitting.0 0. Radoš Řezníček
(p – 1) × ln (p) κ'sta(0) = –––––––––––––––––– 1 + p × (ln (p) – 1)
(13)
The inverse function cannot be calculated explicitly.8 0. and if there are more than 10 available. However. which becomes obvious in the diffusion of the values for one clay content.4 Inflation pressure (bar) 1. as for example the inflation pressure and the maximum net traction ratio κmax in Figure 3.8 0. tillage conditions (not tilled. and natural cover (Stubble. As an example. In this case.8 κmax = –0. To compensate this effect. their correlation is investigated by Steinkampf ’s measurements. driving velocity. d1 = p × b1 × c1 a1 = b1 – ρe (16)
(17)
(18)
With these equations. the gradient is calculated for these 32 values.2 1. Maximum net traction ratio over clay content for 850 measured curves RES. and the dynamic rolling radius. ENG.0 1.6 0.0 0. the soil type. Maximum net traction ratio and inflation pressure for 32 measured curves
Maximum net traction ratio κmax
45
..

width. as the rolling resistance results mainly from the tire deformation and the friction in the contact patch. New input parameters for the traction equations with their minimum.8 0. For example. but if it was considered linear.1 k2 oisture + 0.2 0. support this observation. which increases with the increasing radius. the gradients cannot be used directly. Thus. the three tire parameters are combined into one new parameter.8 1. ENG. The factor of the influence changes. Therefore. B kclay kmoisture kradius kwidth kpressure
46
RES.2 0. parameters ktire and kstrength are strongly interdependent. (20). and decreasing inflation pressure. This one tire parameter adequately represents the influence on the tractive behaviour compared to 5 soil parameters. The remaining parameters shown in Table 1 are considered linear.0 0. 54.5 bar stubble field grassland compacted soil compacted soil pure clay wet (30% moisture) 90 cm (big tire) 80 cm 2 bar Model function’s input parameters kcover kstrenth. A kstrenth.6 kmoisture 0. 2008 (2): 43–49
.0
Figure 4. It is a fact that. the minimum and the maximum values were chosen for each coefficient and the trend values were used to calculate the influence of the tire and soil conditions in relation. because the resulting range of values can scatter too much or too little. To consider the correlations in the modelling equations. For a hard and dry stubble field. 1989). AGR.6 κmax 0.4 0.Dedicated to the 80th Anniversary of Prof.and maximum-conditions Value of the parameter Natural cover (roots) Upper soil strength Lower soil strength Clay content Soil moisture content Tire radius Tire width Inflation pressure tilled very loose soil very loose soil pure sand dry (5% moisture) 50 cm (small tire) 25 cm 0. which does not consider any soil conditions. but a larger contact patch is always good for a higher net traction ratio. Even if this is only an inexact value. which is calculated as follows: kradius + kwidth + 1 – kpressure ktire = ––––––––––––––––––––––––– 3 (19)
1. which is even more unrealistic as a too low influence. for the net traction ratio. However. It can be shown that for loose soil. a smaller contact patch is of advantage. The trend for the rolling resistance ratio is influenced contrarily by the parameters ktire for loose and hard soil surfaces. the optimum occurred for extremely wet conditions. shown in Figure 4. a large contact patch is advantageous because of less sinkage and
soil deformation. (20). The disadvantage is that the influence of this parameter is rather too low.54 m 0. there exists one exception. For further simplification. but for almost all of them the trend remains the same. for higher and lower moisture contents the maximum value is lower. as shown in Eq. More interdependences between the parameters are not considered in this model because they cannot be expected and could not be shown by the analysis of Steinkampf ’s data. and that keeps the equations simple. Maximum net traction ratio and inflation pressure for 32 measured curves
This new parameter ktire is a benchmark for the size of the contact patch. a moisture content between 15% and 20% is the optimum. the correlations between the tire behaviour and the modelling functions can be shown.
Table 1. parameter kmoisture is considered quadratic and not linear. Thus.12kmoisture + 0. Steinkampf ’s measurements. because changes in soil conditions influence the tractive performance much more than the changes in the tire dimensions (Upadhyaya et al. but contrary behaviour is not reasonable. parameter ktire influences the value of the rolling resistance ratio positively on hard soil and negatively on loose soil.4 0.0 0 κmax = –0. Radoš Řezníček
correlations have to be calculated for one parameter and the mean value of all these correlations can be chosen to characterise the dependency.. This effect appears accounted linear in Eq. All parameters have an interdependent influence on the tire behaviour.

If no significant relation could be found between the input parameters and the characteristic coefficients.11kstrenth. however. it is considered as a constant value. Radoš Řezníček
1.8kcover + 1.2 0.015 + 0. To figure out the quality of these equations. characteristic coefficients are shown.Dedicated to the 80th Anniversary of Prof.0 0 20 net traction ratio κ rolling resistance ratio ρ 40 60 all values k = 1 all values k = 0. as calculated by Eqs (12)
47
. AGR. They already show the best traction conditions on the grass land. ENG. calculated by Eqs (20) to (26).B –6kclay+8 kmoisture σκmax =––––––––––––––––––––––––––––––––––– 100 (21) κ'(0) = 5 + 2.18 – 0.A + 0.09 (–4kmoisture2 + 4kmoisture) + + 0. Figure 5. In the last part of the table.05 kstrenth.8 Net traction ratio κ Rolling resistance ratio ρ 0. Simulated curves for extreme k-values
Figure 6. on a cultivated field.13kcover + 0.
Proposed equation for traction
influence of this value is not that important for the tire behaviour. ρe = ρ – ρi
Examples for net traction ratio and rolling resistance ratio
(26)
The resulting prediction equations comprising the influence of soil and tire parameters are as follows: κmax = 0.31 + 0. like all k-values = 1 or all k-values = 0. and the analysis of further measurements is needed to optimise this model for the traction prediction.07kclay + 0.0 ρ 0 20 Slip σ (%) 40 60 greenland stubble field κ cultivated ploughed
Slip σ (%)
Figure 5.A – 0. The curves show an extreme behaviour. that even for extreme assumptions concerning the k-values. the results are not unrealistic. they could be considered in the prediction equations. but even now the results are good and display a realistic tire behaviour. if there are more interdependences or nonlinear relations. the parameters derived from Eqs (10) and (11) are shown. In the second part of the table the new.01 × ktire The rolling resistance ratio is calculated by: ρ(σ =0) = 0.13 ktire (20) 55–18kcover–12kstrenth.0 Net traction ratio κ Rolling resistance ratio ρ 0. This.A + kstreth.6 0. It is obvious that. a net traction ratio higher than 1 or comparable absurd results are avoided by this model.06kstrenth.013 (25) The external rolling resistance ratio ρe is the main part of the total rolling resistance ratio.B – 1) × 0. ρi = 0.03 ktire (24) The gradient of this rolling resistance ratio can be approximated linearly with the slip. The values differ in a small range for all measurements and the
RES. these are not included in the equation for that value. there is no biological cover and the upper soil is very loose..25 0.3kstrenth.75 0. the tire and soil parameters are shown. ρgradient = 0.B + 10.B – (kstrenth. Four simulated curves for different tire-soil-behaviour
For a further optimisation of the model.A – 8kstrenth. Table 2 shows the respective values.A (22)
The internal rolling resistance ratio ρi increases with bigger tires and lower inflation pressure. In the first third.02 kcover – 0. these were used for defined tire and soil conditions to calculate the net traction ratio and rolling resistance ratio. 54. The interpretation of the new database is not yet finished. 2008 (2): 43–49
(23)
One advantage of this model is. more tests have to be performed and. They can be easily assumed for any soil condition.5 all values k = 0 0.5 0.6 0. As an example. as they can be compared easily and allow an interpretation of the tire behaviour.4 0.09kstrenth. The results are shown in Figure 6. four different soil/ field – conditions are used to show the results of the model.

8–14. Ann Arbor. McAllistar M.
References Armbruster K.0 0.775504 6.0844 0.5858 0.0000 0. Bd.7 0.M.0130 0.304972 0.. Another point is a high rolling resistance ratio for cultivated and ploughed soils.7073 0.0130 0. which holds the tire and a high net traction results.1446 0.991432 0.0641 0.0203 0. This can happen by analysing more different traction tests and calibrating the numeric values of the equations.090685 0.3340 5. All these model curves agree well with the tire behaviour in the field. 11th Int.5 0.854023 6. Radoš Řezníček
Table 2. 1755–1761. The University of Michigan Press.658403 –8.5520 5.. but the results are realistic for further calculations and they describe standard tire behaviour for special conditions.0000 0. If the tests show more interdependent or nonlinear behaviour. (1960): Off-the-road Locomotion.6 0.G. A Tire and soil parameters kstrenth. Examples for 4 different soil conditions (used tire: 540/65R30) Stubble field kcover kstrenth.5 0.0 0.5 0.8 0.839570 1.0203 0. After the grass cover is sheared off at slip values between 20% and 30%. Bekker M.D. (1979): A rig for measuring the forces on a towed wheel.1243 0.0203 0. B kclay kmoisture ktire κmax σκmax Characteristic κ. The proposed empirical equations give a good opportunity to estimate the tire behaviour for any kind of agricultural tires used on farmland.0 0.4820 5.0130
to (18). the equations can be extended to take this into account.0844 0. Ann Arbor.757757 0. Bd. Research and Developments in Terramechanics.0130 0.5 0. In: Proc. AGR. The University of Michigan Press.712710 6.4-35/15-35 tractor tyre.5 0.0130 Ploughed 0.0130 Grass land 0. Wageningen. 24: 259–265. ISTVS.611145 0.1092 0.0203 0. (1956): Theory of Land Locomotion.267193 0. Du Plessis H.575806 0. even if they have to be further optimised.6 0. Dublin.
CONCLUSIONS
advantage compared to other models is that the results can be calculated by some equations without the need of complex models. and the results are realistic and representative for the tractive behaviour in the field. 2008 (2): 43–49
The results of the proposed equations are good.0517 0.Dedicated to the 80th Anniversary of Prof. Kutzbach H.25 0.6 0.6 0. ENG.1166 0.5 0.8 0.904692 0.G.1166 0. 4th European Conf.6 0. The input parameters are easy to measure or to estimate for standard soils. RES.1 0. (1989): The combined lateral and longitudinal forces on a 18.7667 0. 1.2580 5. The Mechanics of Vehicle Mobility.and ρcoefficients κ’(0) ρi ρe ρ(0) ρ’ a1 Coefficients of equations (11) and (12) b1 c1 d1 a2 b2 0.194184 0. The main
48
. (1989): Development of a single wheel tester for measurements on driven angled wheels. Journal of Agricultural Engineering Research. 54.0963 0.700072 7.5133 0.5 0.L. 3.9 0.6 0. These values are unsuitable for interpretation but needed to display the curves.5 0. which is essential for further vehicle modelling. Agricultural Engineering. The curves do not fit all values measured. This can be explained by the grass cover.956358 –3.1446 0.6 0. which is hardly possible.0720 0.0720 0. For example a high net traction ratio at low slip values on grass land. In: Proc. Bekker M. Congr.0130 6. the net traction decreases.0130 Cultivated 0.2293 0.

4. Steinkampf H.J. šířka. Jahns G. (1986): Betriebseigenschaften von Ackerschlepperreifen bei unterschiedlichen Einsatzbedingungen. (2007): Comparison of different zero-slip definitions and a proposal to standardize tire traction performance. a gradientu κ při nulovém prokluzu (κ’(0)). Landbauforschung Völkenrode. 2008
Abstrakt Schreiber M.. October 3–6.K. Sonderheft 80. Klíčová slova: pneumatika. Journal of Terramechanics. In: Int. 2006. Eng. pevnost horní části půdy. Johann-Georg-Fendt-Str. (1989): Traction prediction equations for radial ply tyres. které se dají jednoduše určit. obsah jílu.D. 26: 149–175. 12th Int.: + 49 8342 77 270. Kutzbach H. Journal of Terramechanics. Shaker Verlag. pohon. e-mail: Matthias. (1994): A new field single wheel tester. Agr. Product Marketing (MP).K. (1990): Review of traction prediction equations.D.
Tahová síla.de
RES. (1974). Journal of Terramechanics.. a popis této rovnice 4 významnými charakteristickými koeficienty: souřadnicemi x a y u κ-maxima (σκmax.D. Luth H.K. Radoš Řezníček
Schreiber M. Landbauforschung Völkenrode. pevnost spodní části půdy. 1994.. Journal of Terramechanics. Upadhyaya S. ISTVS.. Res. K analýze 850 trakčních křivek byly užity empirické rovnice.D. [Ph. In: 10th European Conf. August 29 – September 1. Kromě motoru je tahová síla ovlivněna vlastnostmi půdy a pneumatik. Sonderheft 27. AgEng. 2008 (2): 43–49
49
. Thesis. Upadhyaya S. tří parametrů charakterizujících pneumatiku (poloměr. ENG. [Ph. Mehlschau J. Schreiber M.. Upadhyaya S.] TU Braunschweig.. 30: 1–20. 510–511. Jubbal G. Wulfsohn D. Matthias Schreiber. Ermittlung von Reifenkennlinien und Gerätezugleistungen für Ackerschlepper.] Universität Hohenheim. úseku na ose y ρe. (1993): An instrumented device to obtain traction related parameters. 54: 43–49. prokluz a valivý odpor jsou hlavními kriterii pohonu terénních vozidel. Wulfsohn D. Hlavními kroky k dosažení tohoto cíle jsou: rozšíření rovnice prokluzu o lineární člen prokluzu. Wolf D.. 10: 49–61. Ronai D. půdní vlhkost).. (2006): A traction prediction model for agricultural tyres.. (2008): Vliv půdy a pneumatik na pohon. Paper 94-D-034..Schreiber@xfendt. Received for publication December 27.Dedicated to the 80th Anniversary of Prof. 44: 75–79.. Tyto parametry s relativními hodnotami mezi 0 a 100 % jsou užity k určení rovnic pro předpověď pohonu. Kutzbach H. Chicago. prokluz
Corresponding author:
Dr. Wulfson D. Výsledkem analýzy bylo určení několika důležitých parametrů. Winter Meeting of the ASAE. (2006): Kraftstoffverbrauch beim Einsatz von Ackerschleppern im besonderen Hinblick auf CO2-Emissionen. In: Proc. které byly naměřeny a publikovány Steinkampfem.. 2007 Accepted after corrections February 20. Numerische Beschreibung der Betriebseigenschaften von Ackerschlepperreifen. AGR. Wismer R.. Tyto vlastnosti musí být popsány omezeným počtem parametrů. Steinkampf H. Thesis. κmax). Schreiber M. (1973): Off-road traction prediction for wheeled vehicles. Budapest. 87616 Marktoberdorf. Congr.D.. Smulevich I. Kutzbach H.D. AGCO GmbH. tlak nahuštění) a pět půdních parametrů (půdní povrch. 54. Germany tel.