Professional Documents
Culture Documents
Machine Tools
Abstract
Rolling-sliding contacts are tribological systems, which change their properties during oper-
ating. In addition to the load, the material properties and the sliding kinematics, the perfor-
mance of tribological systems regarding wear, friction and contact fatigue is significantly
influenced by the properties of the lubricant as well as the surface roughness and structure.
The knowledge about the interaction between lubricant, its additives and the contacting parts
is fundamental for optimization of running behavior. In the first part of this paper, the influ-
ence of the lubricant with different amounts of additives is evaluated from the experimental
point of view.
In the second part of the paper it is focused on the changes of surface roughness during run-
ning-in. The evaluation and knowledge of surface roughness after completed running-in is
most important for an accurate design of rolling-sliding contacts. Therefore, a new method
for predicting the elastic-plastic deformation of single roughness peaks from a measured sur-
face topography is presented. The solving algorithm performs locally an iterative correction
of the geometry in order to reduce the local maximums of pressure at each roughness peak.
The vital advancement in this procedure is the output of the running-in surface topography,
which can be directly compared with testing results for the validation of the calculation ap-
proach. The calculation approach is validated for two different surface topographies result-
ing from grinding and shot peening, whereas in both cases, the method quantitatively con-
firms the deformation effects at the roughness peaks.
running-in can be divided into different categories. On the one hand there are geometrical
changes of the surfaces caused by the flattening of roughness peaks [LOHN15a, LÖPE15]
and furthermore there are changes of micro structure in the subsurface zone [GOHR82,
VOLG91]. On the other hand changes of the boundary layer can be observed in consequence
of tribo-induced chemical reactions [BREC15, LOHN15b]. In case of a shift of operating
conditions, there are usually no stationary contact conditions, which means that there is a
renewed running-in of the tribological system [MICH87, TOBI01, LÖPE15]. By definition
of the operating conditions during the running-in process, the properties of tribological sys-
tems can be optimized in terms of contact fatigue strength, friction and wear.
M1, n1
Continuous Running-In
Process during Loading
S
system change
I II III
Influence Factors: time / load cyles Influence Factors:
– Manufacturing process – Initial surface conditions
Running-
– Process parameters Fatigue Damage – Material properties
In
– Properties of work piece – Operating conditions
© WZL
Figure 2 shows the main influences on the running-in process of rolling-sliding contacts. In
addition to the kinematics and the lubrication, the main influence categories on the properties
of the rolling-sliding contact after running-in are the contact geometry, load and the material
properties of the subsurface zone. The radii of curvature of the contacting parts, waviness as
well as the surface roughness and structure are assigned to the category “contact geometry”.
The material properties can be quantified by the modulus of elasticity and the yield strength
of the subsurface zone, whereby the yield strength has a large number of influence factors
(micro structure, degree of deformation etc.). For the categories “kinematics” and “load” the
surface velocities in terms of value and direction, the normal force and the contact time are
subsumed, for example. The influence factor “lubricant” contains the properties of the lubri-
cant in regard of base oil viscosity and additives.
In this report the two categories of running-in “change of geometry” and “change of the
boundary layer” are focused. For the category “change of geometry” the basic relations of the
running-in behavior under rolling-sliding conditions are known from the phenomenological
15-3 Running-In in Gear Applications
point of view. Experiments have repeatedly shown that surface roughness reduces during run-
ning-in and that the final surface structure has a decisive influence on the performance of the
rolling-sliding contact during operating conditions in regard of load capacity and efficiency
[BÖRN76, GOHR82, VOLG91, BUGI09, MAYE13, LÖPE15]. Nevertheless, the surface
roughness after the running-in is generally not known in the design phase of gears since nei-
ther empirical models nor knowledge-based calculation approaches exist to predict the geo-
metrical changes during running-in. The resultant uncertainties in the design process are com-
pensated by extensive testing as well as oversizing. For the future, a generally applicable
prediction model could close this gap by quantifying the surface geometry of the contacting
parts after completion of running-in processes as a function of any initial surface state. In this
report, an approach for a prediction model is build up considering the main influence factors
load, contact geometry and material properties, while the kinematics and the lubricant are
neglected in the first step. This approach allows a stepwise development of knowledge for the
geometrical running-in process and builds the basis for a strong validation of the model.
M1, n1
Continuous Running-In
Process during Loading
Contact
Lubrication
Geometry Material
Load
R1 Properties
Kinematic
F
σ
v
S
ω
ε P
© WZL
For the category “change of the boundary layer” this report highlights the influence of differ-
ent lubricants and additives on the formation of boundary layers. Gear transmissions are usu-
ally run under mixed friction, i.e. the tooth flanks are not completely separated by an oil film.
Boundary layers, which develop on the surfaces during manufacturing and application, can
protect tooth flanks from metal-metal contacts. Knowing the influence of the lubricant on the
characteristics of the boundary layer, it is possible to define a tribological system during the
design phase, which performs best in regard of efficiency and load capacity.
WZL Gear Conference in the USA 15-4
2 Research Aim
The tribological system of two gear flanks changes consistently during operating conditions.
The strongest changes can be observed during running-in process. This report examines the
subject of running-in on two different levels. On the one hand the influence of the lubricant
and its additives is investigated in the chemical point of view. The manufacturing process,
especially the grinding feed rate and the metalworking fluid, cause manufacturing induced
boundary layers and influence the running behavior of the produced parts [BREC14,
BREC15]. In the later application, the lubricant and its additives determine the running be-
havior. Therefore, the chemical interaction between the surface and the transmission oil is
analyzed in the first part of this report. The aim is to determine the effect of the lubricant
composition on the running-in in regard of friction coefficient and roughness reduction, Fig-
ure 3. Besides the usage of a common gear box oil including an well-designed additive pack-
age, the potential of white oil without any chemical additives is evaluated in detail.
-50%
Ra - Value
a) S b)
P
before after
Additives
Research Aim 1 Research Aim 2
Determination of the Influence of Determination and Prediction of
different Lubricant Additives on Surface Flattening of Manufacturing-
Friction Coefficient and Roughness Induced Surface Structures within
during Running-In the Area of the Roughness Peaks
© WZL
On the other hand physical aspects of running-in are highlighted in the second part of this
report. The focus of this section is on the descriptions and the prediction of geometrical sur-
face changes during running-in, Figure 3. By knowing the running-in state of the surfaces
instead of the initial state after manufacturing, a more precise calculation of rolling-sliding
contacts is possible during design phase. Therefore, the second research aim is the determi-
nation and prediction of surface flattening of manufactured-induced surface structures in the
area of the roughness peaks. The core of the calculation approach is the half-space theory
according to BOUSSINESQ/LOVE. The elastic-plastic material model is based on the idea of a
limiting pressure, which borders the maximum pressure inside the contact area. The main
15-5 Running-In in Gear Applications
result of the calculation is the geometrically corrected surface topography reproducing the
running-in surface of the experiment. The predicted and the real geometry can be directly
compared with each other for validation. In detail, the prediction model is validated by inves-
tigations on the disk-on-disk test rig with ground and shot peened surface structures. Finally
the development status of the prediction model is scrutinized and further improvement steps
are derived.
WZL Disk-on-Disk Test Rigs Contact Fatigue Test Rig Friction Force Test Rig
1 Test Shaft (cylindrical)
2 Counter Shaft (crowned) 5 4 1
3 Pressure Piston
4 Force Lever FN
5 Housing
6 Counter Shaft Support n1
7 Feather Springs FR
8 Friction Force Strain Gauge
Contact Geometry
4
Test Shaft 5 n2
FDMS
Counter Shaft
8 7 6 2 3
FDMS n1
3 2 1 TOii n2
Point Contact Line Contact
Point Contact Line Contact
© WZL
plications. The white oil also has the viscosity class ISO VG 68 and is specified by the Ger-
man Pharmacopoeia 10 [HART91]. All used lubricants are provided by the company Fuchs
Europe Lubricants.
Gearbox Oil, with Additives Gearbox Oil, No Additives Mineral Oil (White Oil)
Industrial transmission oil Base oil of the industrial Medical mineral oil specified
transmission oil by DAB10 standard
DIN-Class: CLP
– Circular Lubrication (C) Aromates 7% High degree of refinement
– Additives against Retained sulpur 0.4 – 0.7% Contains no aromates or
corrosion and aging (L) sulphur
– Additives for lower
friction and wear (P)
© WZL
than the friction coefficient of the industrial gearbox oil. However, the different time profile
of the friction coefficient suggests that the surface and lubrication states of the three test var-
iants also differ from each other over the testing time.
The increase of contact pressure leads to a reduction of friction coefficient for both variants
of gearbox oil. This relationship can be explained by the pressure-viscosity behavior of the
lubricant at higher pressures, resulting in an improved surface separation and thus a lower
proportion of solid-state friction [LÖPE15]. On the other hand the lubrication by white oil
leads spontaneously to a scuffing damage. One possible explanation is that the high degree of
refining of the white oil reduces the lubrication potential despite the same viscosity class,
because neither additives nor unsaturated molecular groups, e.g. aromatics exist in the oil
[ASCH31]. This experimental results emphasize that the frictional behavior, the scuffing load
capacity and thus the surface state are to be investigated independently of each other.
Scuffing
0.06
0.05
Gearbox oil Gearbox oil, Mineral oil
+ additives no additives (white oil)
© WZL
stripes at the edges due to the contact conditions. The usage of the gearbox oil without addi-
tives leads to discolorations in the center of the contact area but no micropittings. In contrast,
the discolorations caused by the mineral oil (white oil) are outside of the contact area. The
contact are is shiny without discolorations or micropittings.
The low value of friction coefficient correlates with the values of surface roughness, which
were measured after the test run in the center of the contact area. For all three variants, there
is a reduction of the arithmetical mean roughness Ra. While the reduction in the Ra value is
17% compared to the initial state of the gearbox oil with additives, roughness is reduced by
57% for the gearbox oil without additives and 67% for the mineral oil (white oil).
Key 0.1
New condition 0
After test run Change of Roughness
© WZL
The results emphasize that the change of roughness and surface discolorations depends on the
amount of additives. Regarding the discolorations, it is likely that with increasing load and
temperature, a chemical change is associated for both lubricants with a low amount of addi-
tives (gearbox oil without additives and mineral oil – white oil), because both lubricants do
not contain additives for protection against oxidation and aging. Regarding the reduction of
surface roughness, a possible explanation is that the local thermal and chemical stresses are
higher for the lubricants without additives. In this case the smoothing of the surface roughness
is higher as the material properties of the test parts are the same at the beginning. The fact
that increased stresses in the subsurface zone during the running-in period can lead to a better
running behavior in the following operation conditions has already been shown by finite life
strength tests [BREC13] and friction force measurements [LOHN15a]. The occurrence of
scuffing for the white oil shows that the effect of friction reduction is not sufficient as a quality
criterion for a lubricant. A compromise to achieve a reduction in friction during running-in
WZL Gear Conference in the USA 15-10
and at the same time maintain a sufficient load capacity during operating would be a separa-
tion of the running-in oil and operating oil.
the basis of roughness parameters (Ra, Rz, Rpk, Rk, Rvk). This validation strategy ensures the
quantitative comparability of the primary profiles by the use of roughness parameters as well
as the qualitative comparability by the direct adjustment of single roughness peaks.
Aim
F
Approach
Comparison of Roughness
© WZL
of the characteristic roughness values Rpk, Rk and Rvk, which can be derived from the Abbott-
Firestone curve. The reduced peak height Rpk decreases the most and confirms the experience
of a smoothing of roughness peaks during running-in [LOHN15a, LÖPE15]. However, the
characteristic values of the core roughness Rk and the reduced valley depth Rvk are also re-
duced, which indicates a displacement of the material from the roughness peaks in the direc-
tion of the roughness valleys, Figure 9. The roughness valleys are practically filled up by the
material of the roughness peaks.
Roughness [µm]
Rbal 1/2 = ∞/166 mm -50% -61% -50% -40%
0,6
Testing Conditions
0,4
pH = 1100 MPa
n1 = 3000 1/min 0,2
s1 = 0% 0
No Lubrication Ra Rpk Rk Rvk
Variant 1: ARTS
- external plunge grinding Shot Peened Surface Structure
Roughness [µm]
-1,5% -0,5% -2,5% +9,5%
1,2
0,9
0,6
Variant 2: KGS
0,3
- shot peened
0
Ra Rpk Rk Rvk
For the shot peened surface structure of the variant KGS, the arithmetic mean roughness
changes by ΔRa = -1.5%, which is within the measurement accuracy for the time-discrete
comparison of surface profiles [YUSO16]. The increase of the reduced valley depth after the
running-in is not to be expected with otherwise nearly unchanged roughness values and can
relate with the higher measurement deviation for non-directional surface structures.
A possibility for a qualitative evaluation of the surface change is the direct matching of the
primary profiles before and after the run-in, Figure 10. The ground surface structure (variant
ARTS) emphasizes the reduction of the arithmetic mean roughness Ra in comparison to the
measurement before running-in. As a consequence of the normal force during testing, the
roughness peaks resulting from the grinding process develop to rounded crests, Figure 10 top.
In addition to the expected reduction of the roughness peaks (Rpk value, see Figure 9), the
reduction of the roughness values can also be recognized in the qualitative comparison. In
contrast to the amplitudes of the profile, the characteristic shape of the roughness peaks and
valleys does not change. The frequencies of the primary profile are maintained, especially
with a distinct plastic deformation, whereby only the amplitude of the oscillation decreases.
15-13 Running-In in Gear Applications
For the shot peened surface structure of the variant KGS, the qualitative change of the rough-
ness profile is less than the change of the ground surface variant ARTS. Thus, the analysis of
the roughness values in Figure 9 is qualitatively confirmed. In the region of the roughness
peaks, the change of the primary profile is more pronounced than in the area of the roughness
valleys. The highest roughness peaks being in contact during loading are deformed with an
enlarged radius of curvature, but only to a limited extent. In contrast, there is no change of
the roughness valleys.
4
[µm]
Variant 2: KGS 2
- shot peened
0
0 200 400 600 800 x1000
[µm] 1200
The minor changes of the shot peened surface structure compared to the ground surface can
have different causes. On the one hand the shot peened structure has a higher amount of long-
wave profile components with an already rounded top structure from the process itself (see
Figure 10). In this case, the normal force is absorbed by the long-wave structure of less rough-
ness peaks but more evenly distributed over a the contact area of a single roughness peak. The
maximum pressure resulting from the initial geometry are therefore lower than for the ground
surface structure.
Furthermore, the material properties of the subsurface zone of both variants differ. Shot peen-
ing effects a material hardening which is accompanied by an increase of hardness, a concen-
tration of dislocations and possibly a structural transformation in the shot peened regions
[KÖCH96]. For the case-hardened disk-on-disk test specimen of 16MnCr5, an increase of
surface hardness as well as a reduction of retained austenite can be expected due to the me-
chanically applied loads by shot peening [KÖCH96, SEKI07] because the retained austenite
is a metastable structural component at room temperature. The increased hardness and the
reduced retained austenite content cause an increase of yield strength [KÖCH96], which can
be another reason for a less pronounced geometric running-in of this variant.
WZL Gear Conference in the USA 15-14
The procedure for the elastic-plastic contact calculation is shown in Figure 11. Due to the
high non-linearity in the elastic-plastic contact calculation an iterative solving algorithm is
chosen. The starting point of the calculation is the surface topography of the contacting parts
in the initial state, e.g. after the final manufacturing step. In the first step of the iteration, the
elastic pressure distribution for the current contact geometry is calculated on the basis of the
elastic half-space according to BOUSSINESQ / LOVE [BOUS85, LOVE29]. Using the "method
of combined solutions" of BRECHER ET ALT. [BREC16], the complete primary profile of the
roughness measurement is used, so that the macro geometry and long-wave deviations are not
neglected. In particular, the transferability to any surface structures is ensured in this case for
the developed running-in prediction model.
In the second step of the iteration (Figure 11), a local correction of the surface topography
according to Eq. (5-1) is performed. At all contacting points i, where the purely elastically
calculated pressure exceeds the theoretical limiting pressure plim, material is removed from
the original surface topography over the correction value Δhcor(i), Eq. (5-2). The geometric
correction of the profile Δhcor(i) is always defined positive, because in case of a pressure over-
load only material is removed and not added. The amount of the geometric correction is de-
fined as a function of the pressure difference between the elastically calculated pressure and
the limiting pressure in order to achieve a better convergence behavior.
For a physically reasonable calculation, it has also to be checked whether all contact points
with a local geometry correction Δhcor(i) are still in contact at the end of the calculation. Due
to the high non-linearity and the corss influences of deformation, it is possible that the cor-
rection of geometry is locally too high for a previous iteration step. Thus, contact points do
not reach the required limiting pressure plim any longer. In this case, adding of material or the
reduction of Δhcor(i) is necessary, Eq. (5-3).
15-15 Running-In in Gear Applications
Initial Geometry
Elastic Contact Calculation
Half-space theory (BOUSSINESQ/L OVE)
„Method of Combined Solutions“
deformed F
h
Iteration:
• Plastic deformation Iterative
Initial geometry
finished? Solving
• Equilibrium of forces
fulfilled? Algorithm Correction of Contact Geometry
Pressure
𝑝 𝑖 > 𝑝𝑙𝑖𝑚
Running-In
Surface x
removing of
h x material (local)
© WZL
After the local correction of the contact geometry, the progress of the iterative solving algo-
rithm has to be tested for different characteristic values, Figure 11. For a successful elastic-
plastic contact calculation, the external force equilibrium has to be fulfilled and furthermore
the plastic deformation in the form of the geometric correction has to be completed. The ex-
ternal force equilibrium is checked analogously to the calculation procedure according to
HARTNETt by defining a limiting error, Eq. (5-4) [HART79]. If the current error exceeds the
limiting error εFN, the global displacement of the contacting parts u0 is adjusted.
𝑝(𝑖)−𝑝𝑙𝑖𝑚
(| |) > 𝜀𝑝𝑙𝑖𝑚 ∀𝑖𝑤𝑖𝑡ℎ∆ℎ𝑐𝑜𝑟 (𝑖) ≠ 0 (5-5)
𝑝𝑙𝑖𝑚 𝑎𝑟𝑖𝑡ℎ𝑚
Calculated profile height including
h µm p MPa Elastically calculated pressure
of the plastic deformation
Δhcor µm Geometrical correction of profile plim MPa Limiting pressure
hstart µm Profile height of the initial state FN,soll N Given normal force
Global displacement of contacting
FN,ist N Calculated normal force u0 µm
parts
εFN - Tolerated error of normal force i - Contact point
Tolerated error of the average deri-
εplim -
vation of contact pressure
WZL Gear Conference in the USA 15-16
The completion of the plastic deformation of the roughness peaks is quantitatively examined
by the average deviation of contact pressure from the limiting pressure plim for all corrected
contact points, Eq. (5-5). If the average pressure deviation exceeds the permissible error εplim,
the contact geometry is corrected again in the next iteration step according to the criteria of
Eq. (5-2) and (5-3). According to these two termination criteria, the iterative loops are run in
parallel until the resulting surface topography does not exceed the limiting pressure at the
roughness peaks in the case of purely elastic contact calculation and until the equilibrium of
forces is fulfilled.
Compared to existing elastic-plastic contact calculations on the basis of the half-space theory,
the main advantage of the new approach is the direct calculation of the plastically deformed
surface topography, so that a local matching between experimental and calculated primary
profiles is possible. It is not obligatory to use roughness values or tribological substitute var-
iables such as, the contact ratio under load, in order to validate the calculation. In addition,
the development of the iterative solution algorithm allows the extension of the calculation by
a strengthening material behavior. Furthermore, the advantage of this knowledge-based ap-
proach is the flexible and targeted transferability to any surface structure.
Figure 12 shows the iterative solution procedure for the calculation of the plastic deformation
during the running-in with the example of a ground surface. For different iterations steps, the
corrected contact geometry and the corresponding elastically calculated pressure distribution
are compared. In the first iteration step, the initial surface geometry is used. The pressure
distribution is inhomogeneous and exceeds the theoretical limiting pressure at many contact
points by more than factor three. Depending on the local pressure, the contact geometry is
successively corrected in the subsequent iteration steps.
Contact Geometry
Progress of Iterative Solving
R1/2 = 21/21 mm
Rbal 1/2 = ∞/166 mm 1. Iteration 10. Iteration 40. Iteration
Calculation Parameter 1,2
[µm]
Geometry
pH = 1100 MPa
E1/2 = 210000 MPa 0,8
dx = 1 µm
0,4
plim = 3000 MPa
Surface Structure 0 x x x
- external plunge grinding 0 0,5 [mm] 1 0 0,5 [mm] 1 0 0,5 [mm] 1
(tangential)
[GPa]
9
Pressure
6 plim
3
0 x x x
0 0,5 [mm] 1 0 0,5 [mm] 1 0 0,5 [mm] 1
© WZL
After the tenth iteration step, a smoothing of the highest roughness peaks can be already ob-
served. The maximum pressures in the elastic contact could be reduced by the adaptation of
the geometry. Nevertheless, the limiting pressure is still exceeded at the contact points. After
40 iterations, the termination criterion according Eq. (5-5) is fulfilled for the present surface
structure. The correction of the contact geometry is progressed in a way that the elastically
calculated pressure is below the limiting pressure for each contact point. Contact points which
do not reach the limit pressure at the end of the solution either do not enter the contact area
or are deformed purely elastically without geometric correction. Due to the elastic-plastic
calculation, the amount of contact points increases in comparison to the initial geometry.
The starting point of the validation strategy for the developed prediction model of the geo-
metric running-in is the contact geometry after the final production step or rather before the
start of the test run. The measurement data of the surface structure are the most important
input variable for the calculation. After performing the test run, the surface structure of the
contact area is measured again and compared with the calculation result. In this report, the
calculated and experimental surface structure after running-in is compared in two different
ways. On the one hand 2D roughness values are derived from the measurement data, which
can be used to analyze the similarity of the primary profiles. The usage of roughness values
is particularly suitable for a quick and quantifiable comparison of primary profiles, whereby
the influence of the filter parameters have to be considered for the extraction of the roughness
profiles. On the other hand the measured and the calculated primary profile after the running-
in are compared by a qualitative one-to-one adjustment. Although the quality of the prediction
cannot be assessed quantitatively in this approach, a direct comparison offers the possibility
to detect local deformation effects and to optimize the material models used.
In order to validate the prediction model, test results for two different surface structures are
measured before and after the test run. Figure 13 shows a comparison of the primary profiles
for a ground surface structure in the initial state, after the running-in and for the calculation
result. The qualitative smoothing of the roughness peaks during rolling in the test rig is con-
firmed by the calculation. Depending on the peak height and the distribution of the neighbor-
ing roughness peaks, the degree of smoothing for the individual peaks varies because the
cross-influences of deformation affect decisively the load distribution and thus the exceeding
of the limiting pressure. Due to the elastic and ideal-plastic material model, the corrected
roughness peaks are formed in a crest-shape (Figure 13). This phenomenon can also be seen
when neighboring roughness peaks converge into a single peak, since the material model does
not yet take into account any hardening effects depending on the degree of deformation. In
the region of the roughness valleys, the surface structure is not changed by the calculation.
The measured profile remains completely preserved in these areas because the computational
algorithm does not yet fulfil the volume constancy by distributing the material removed from
the roughness peaks.
WZL Gear Conference in the USA 15-18
The quantitative evaluation by means of 2D roughness values confirms the qualitative analy-
sis of the experimental and calculated primary profiles after the running-in, . Figure 13. The
value of the reduced peak height Rpk differs by less than 6% between the calculation and the
examination, whereas the reduced valley depth Rvk are virtually unchanged after the compu-
tation compared to the initial state. The increase of the Rvk value by just under 7% despite the
identical profile is due to the interactions in the evaluation of the Abbott-Firestone curve since
the change in the roughness profile shifts the turning point of the Abbott-Firestone curve and
thus also reduces the reduced valley depth Rvk.
pH = 1100 MPa 1
n1 = 3000 1/min
0,5
s1 = 0%
No Lubrication 0
Calculation Parameter 0 100 200 300 x [µm]
pH = 1100 MPa
Comparison of Roughness Values
E1/2 = 210000 MPa
dx = 1 µm
0,338
0,292
0,312
Roughness
0,216
0,187
0,6
0,220
0,175
0,679
plim = 3000 MPa 0,091
0,107
0,086
[µm]
0,560
0,4
Surface Strutcture
- external plunge grinding 0,2
(tangential) 0
Ra Rpk
Rpk Rk
Rk Rvk
Rvk
Initial State Experiment Calculation
© WZL
Due to the unchanged value for the reduced valley depth Rvk, the core roughness Rk and, in
particular, the arithmetic mean roughness Ra for the computed profile are calculated too high.
The extension of the calculation algorithm for the displacement of the material from the
roughness peaks to the valleys is expected to be more accurate. The use of the Ra and Rk
values for the assessment of the model quality is therefore only useful if the extension of the
model is performed by volume consistency. The boundary pressure plim for the ground surface
was quantified by the elastic contact calculation with the surface structure assumed. For elas-
tic contact, the upper ten percent of the pressures are between p max,10% = 3000 – 4000 MPa.
The choice of plim = 3000 MPa at the lower edge of the scattered band leads to the highest
agreement of the Rpk values. Contrary to the literature, the limit pressure is lower [BART01].
For the shot-peened surface structure, only small changes of the roughness could be deter-
mined after the experiment (Figure 14 and Figure 9). The smoothing of the roughness peaks
in the scale of nanometers can be seen in the region of the highest peaks for calculation result
and experiment (Figure 14). The calculation also confirms the experimental results for this
15-19 Running-In in Gear Applications
surface structure. The roughness valleys retain their original shape after manufacturing be-
cause the overall slight change in the profile peaks leads to a lower flow of material compared
to the ground surface. The roughness values remain nearly constant compared to the initial
state, but the tendencial reduction of the characteristic values are confirmed by the prediction
model. The limiting pressure for the beginning of the plastic deformation has been set higher
for the shot-peened surface structure since the production process of shot peening causes al-
ready a material strengthening by the action of mechanical energy. The order of magnitude
of the limit pressure plim = 4500 MPa is consistent with data from the literature based on hard-
ness measurements [BART01].
Testing Conditions
4
[µm]
pH = 1100 MPa
3 Initial State
n1 = 3000 1/min
2 Experiment
s1 = 0% 1
No Lubrication Calculation
0
Calculation Parameter 0 100 200 300 400 x500
[µm] 600
pH = 1100 MPa
Comparison of Roughness Values
E1/2 = 210000 MPa
dx = 1 µm 1,2
Roughness
0,453
0,449
0,492
0,381
0,379
0,376
0,414
0,412
0,390
1,170
plim = 4500 MPa
1,14
0,9
1,17
[µm]
The prediction model presented in this report for the geometric surface changes during the
running-in is an innovative approach for the calculation of the plastic deformation of individ-
ual roughness peaks. The central output variable of the calculation is the surface topography
including the plastic deformation, which allows a direct comparison of experiment and cal-
culation and thus a targeted validation. Because of the generally applicable approach, trans-
ferability to various surface structures resulting from the manufacturing process is possible.
The current development level of the solving algorithm for calculating the geometric surface
changes involves the local material removal at the roughness points on the basis of a limiting
pressure. In the future, the model has to be extended to fulfill the volume consistency during
plastic deformation in order to achieve a higher calculation accuracy. For this purpose, the
implementation of the material spreading after material remocal is necessary. For validation,
WZL Gear Conference in the USA 15-20
the calculated primary profiles can be compared and analyzed in the same way with the ex-
perimental primary profiles. In particular, the quality of the reduced valley depth Rvk can be
increased by using volume constancy because the roughness valleys are successively "filled"
by the spreading of the material. In this case, the accuracy of the prediction of the arithmetic
mean roughness is also increased in comparison with current calculation results.
A further central challenge in the transferability of the method to different, process-related
surface structures is the exact knowledge of the limiting pressure plim. The example of the
ground and shot-peened surface structure shows the highest correlation between experiment
and calculation for different limiting pressures. The critical parameters for the limiting pres-
sure are the edge hardness, the degree of deformation, the micro structure (amount of retained
austenite) and the residual stress state. For future research activities, the determination of the
correlation between the material parameters and the limiting pressure is a key target.
The prediction model is validated by investigations on the disk-on-disk test rig with ground
and shot peened surface structures. The validation focuses on the comparison of 2D roughness
values as well as on the direct positioning of the measured primary profiled before and after
running-in. Despite the different process chains, the prediction for both surfaces structures
proved to be accurate. As expected, the running-in effects observed during the experiments
are confirmed especially for the roughness peaks (Rpk value).
To increase the quality of the prediction model, the extension of the solving algorithm regard-
ing volume constancy during the plastic deformation is aimed in future work. By distributing
the removed material of the roughness peak into the roughness valleys, a more accurate pre-
diction of the Rvk value is expected. Moreover, further roughness values such as Ra und Rz
will be predicted more accurately. In addition, the quantification of the limiting pressure as
the used yield criterion is a major challenge. For a wide transferability of the method to other
surface structures, new methods have to be developed in order to determine the local material
properties within the first micrometers of the surface. The local material data are necessary
as input parameters for an enhanced calculation.
Acknowledgement
The authors gratefully acknowledge financial support by the German Re-
search Foundation (DFG) [BR2905/44-2] for the achievement of the project
results.
The authors gratefully acknowledge financial support by the WZL Gear Re-
search Circle for the achievement of the project results.
7 Bibliography
[ASCH31] Ascher, R.:
Die Schmiermittel. Art, Prüfung und Verwendung.
2. ed. Berlin: Springer, 1931
[LOHN15a] Lohner, T.; Mayer, J.; Michaelis, K.; Höhn, B.-R.; Stahl, K.:
On the running-in behavior of lubricated line contacts.
In: J. Eng. Tribol., 2015, pp. 1–12
[LOHN15b] Lohner, T.; Merz, J.; Mayer, J.; Michaelis, K.; Kopnarski, M.; Stahl, K.:
On the Effect of Plastic Deformation (PD) Additives in Lubricants.
In: Tribologie + Schmierungstechnik, Vol. 62, 2015, No. 2, pp. 13–24
[SEKI07] Seki, M.; Yoshida, A.; Ohue, Y.; Hongo, T.; Kawamura, T.; Shimoyama, I.:
Influence of Shot Peening on Surface Durability of Case-Hardened Steel
Gears.
In: J. Ad. Mech. Design, Sys. and Manu., Vol. 1, 2007, No. 4, pp. 518–529