1

W:\Artikel\AGMA Fall Meeting\2005\ToothFormCalculation-uk-4.doc

Hombrechtikon (Switzerland), 8 may 2005

To AGMA Headquarters, Alexandria VA 22314

Draft of Paper for 2005 AGMA FALL TECHNICAL MEETING

Subject:
Noise/Vibration

Title:
Tooth Meshing Stiffness Optimisation based on Gear Tooth Form Determination for a Production
Process Using Different Tools

Authors:
Dr. Ing. Ulrich Kissling
KISSsoft AG
Frauwis 1
CH-8634 Switzerland
Tel: +41 55 264 20 30
Fax: +41 55 264 20 33
E-Mail: [email protected]

Dipl.-Ing. Markus Raabe

KISSsoft AG
Frauwis 1
CH-8634 Switzerland
Tel: +41 55 264 20 30
Fax: +41 55 264 20 33
E-Mail: [email protected]

Dr. Michael Fish

KISSsoft AG
Frauwis 1
CH-8634 Switzerland
Tel: +41 55 264 20 30
Fax: +41 55 264 20 33
E-Mail: [email protected]
 2

Autors: Ulrich Kissling, Markus Raabe, Michael Fish

Tooth Meshing Stiffness Optimisation based on Gear Tooth Form Determination for a
Production Process Using Different Tools

Introduction
The methods used in gear production are in constant development. In recent years form grinding (an
alternative to the classic meshing grinding) has become the trend. Another example is a method used
mostly in automotive industry: to improve the working life of tools and in order to get tooth form with
higher root strength, gears are produced using up to three different pre-cutters plus a final honing or
grinding process. One of the latest tendencies in the development of optimized gears is to apply a
special wave-form-like profile modification during the finishing process for the reduction of
transmission error.
These production methods require the development of appropriate calculation methods. In this paper
the calculation of the resulting tooth form is described when several tools are used. Then, based on
this tooth form, the effective meshing stiffness (under load) and the stress calculation are discussed.

The variation of the tooth meshing stiffness during operation induces a deviation in the rotation-angle
of the output gear from the nominal transmission ratio (transmission error) causing vibrations and
noise. The meshing stiffness variation can be improved through optimization of the gear geometry
(transverse contact ratio εα and overlap ratioεβ), but the type of profile modification is also very
important for the stiffness under load.
The current calculation method for the tooth resistance following either AGMA2001 [1] or ISO6336 [2]
is based on the assumption of a tooth form produced by one tool in a meshing process. The method
includes, when using a tool with protuberance, also a production process with a pre-cutter (with stock
allowance for finishing) and final grinding or honing process.
This implies that the formulas in AGMA or ISO resistance calculation methods can not be applied with
gears produced by form grinding or other non-conventional methods. The problem is that for the
calculation of the tooth root stresses some values such as tooth thickness and root rounding must be
known. The calculation method assumes that the tooth form is not exactly known, and therefore
presents formulas which permit calculation of the tooth form just in the considered section of the tooth.
These formulas assume production through a meshing process. But in principle, if the tooth form is
given, the tooth can be calculated by directly using the formulas proposed by the standards.
Therefore, if the tooth form calculation is integrated into the resistance calculation software, AGMA or
ISO standards can be used for any production method.

Tooth form calculation with different tools

In most of the available gear calculation software it is possible to calculate the tooth form when using a
standard tool (hob, generating cutter or gear type cutter). Normally it is also possible to introduce a
grinding allowance and so simulate a 2-step production process (cutting, then grinding). Answering
frequent requests of users of the widely recognized gear calculation software KISSsoft [3], it was
decided to implement a new approach for tooth form calculation in the software.
An unlimited number of tools such as hob, cutting tool, gear-type cutter, grinding disk (generating or
form grinding), and honing wheel can be defined in any sequence desired. The tooth shape can be
visualised after every step, and different shapes can be superimposed to show the material removal
from one step to another. The manufacturing process from tool to gear is also visualised, and sliding
and rolling velocity vectors are indicated (to optimise the tool life).
Fig. 1 shows the different stages in getting to the definitive tooth form when using two cutters and a
final rectifying process. For the gear with a pressure angle αn 20°, a pre-cutter with αn 25° is used.
This leads to an increase of the tool service life and creates a better rounding of the tooth root. As the
flank form should be identical to the final design, the base diameter db of the gear has to remain
identical. Therefore the module of the tool must be increased by COS(20°)/COS(25°) for the pre-
cutter.
 3

Fig. 1: Tooth form generated with a 3-tool-production cycle

1: Pre-cutter mn=1.0368, αn = 25° (green)
2: Protuberance-Cutter mn=1.0, αn = 20.0° (brown)
3: Final rectifying process (blue)

Some additional features of the tooth form calculation are:

- Tool service life: For the improvement of the tool service life (no. of gears cut until the tool has
to be re-sharpened or replaced) the display of the specific sliding on the tool cutting flank is
very important (fig. 2). There are many factors influencing the tool service life. One of them is
the local specific sliding on the tool. A high negative specific sliding on the tool implies that a
short section on the cutting edge of the tool produces a large section on the gear. This means
that this part of the tool is highly utilized and consequently subject to high wear. As shown in
fig.2, a pre-cutting-tool with higher pressure angle (as in fig.1) has a significantly reduced
specific sliding.

- Grinding notch: For the root strength it is important to know if through the grinding process a
so called “grinding notch” (ISO6336-3, factor Ysg) results. Therefore during the calculation of
the grinding process, the notch has to be recognized. Such a notch can reduce considerably
the safety factor for bending stress. In KISSsoft, the corresponding data is automatically
transferred to the resistance calculation and the notch factor is included in the results.
 4

Sliding (Pre-Cutter 25°)

1.5

0.5
Specific Sliding
Tooth Form
0
0 0.5 1 1.5 2
-0.5

-1

-1.5

Sliding (Final Cutter, Protuberance, 20°)

1.5

0.5

Specific Sliding
0
Tooth Form
-0.5 0 0.5 1 1.5 2
-0.5

-1

-1.5

Fig. 2: Drawing of a cutter-tooth for pre-cutting (pressure angle 25°) and finish-cutting (20°) process
with indication of the specific sliding (between cutter and gear) during manufacture. High
positive specific sliding (in this case on the dedendum) indicates higher wear risk. The sliding
on the 20° cutter is twice as high as on the 25° per-cutter.
 5

Vibrations caused by tooth mesh stiffness variation

For many applications today, the noise level is very critical, and should be as low as possible. Noise is
generated by transmission errors, which produces an incremental change of the velocity on the gear
about the nominal value. This effect induces an instantaneous acceleration / deceleration into the
transmission chain and the result is vibrations. The transmission error is produced through the
variation of the stiffness during a mesh cycle. It is also well known that fabrication errors generate a
certain transmission error. Improvement of the gear quality helps to improve the situation, but even
with a gear set of highest quality, we have transmission errors due to the stiffness variation.
For this purpose, the software calculates the single tooth stiffness of any tooth form and determines
the real contact path under load. From this information the real transmission error of the gear pair is
derived. It is possible to study very quickly the effect of a proposed profile modification on the
behaviour of a gear stage under different loads.

The calculation of the tooth pair stiffness under load permits a comparison of different gear geometries
to find the best solution. Fig. 4 shows a typical case of a standard gear set (tooth form in fig. 3) with
transverse contact ratio (εα) = 1.67 (z: 25:76, DP 4.2333, module 6.0). The transmission error is low
(stiffness is high) during 67% of the cycle, when 2 pairs of teeth are in contact, and otherwise high,
when only 1 tooth pair is in contact. (Remark: A theoretical transverse value of 1.67 signifies that
during 67% of the time 2 pairs of teeth are in contact, and during 33% only 1 pair.) The gear without
profile modification experiences a sudden “jump” from low to high transmission deviation as shown in
fig. 4. A well designed profile modification would reduce the rate (or steepness) of this jump producing
a smoother change in the stiffness, but would not change the level of the low and high stiffness phase.
So profile modification is useful but not the solution to the problem. The best method to reduce the
variation of the stiffness is to use a deep tooth profile with, theoretically, a transverse contact ratio of
2.0. In this case, as 2 pairs of teeth are always in contact, the “jump” is eliminated.
Best results will be achieved when the third tooth coming into contact is unloaded. This is possible
when an appropriate profile modification is applied. In this case, the theoretical contact ratio has to be
higher than 2.0 according to the length of profile correction planned.

Fig.3: Representation of the gear pair used for the calculation of the transmission error
 6

0
-0.001 -8 -4 0 4 8
-0.002
-0.003 T (0%)
Delta phi2 (°)

-0.004 T (25%)
-0.005 T (50%)
-0.006 T (75%)
-0.007 T (100%)
-0.008
-0.009
-0.01
phi1 (°)

0
-0.001 -8 -4 0 4 8
-0.002
-0.003 T (0%)
Delta phi2 (°)

-0.004 T (25%)
-0.005 T (50%)
-0.006 T (75%)
-0.007 T (100%)
-0.008
-0.009
-0.01
phi1 (°)

0
-0.001 -8 -4 0 4 8
-0.002
-0.003 T (0%)
Delta phi2 (°)

-0.004 T (25%)
-0.005 T (50%)
-0.006 T (75%)
-0.007 T (100%)
-0.008
-0.009
-0.01
phi1 (°)

Fig.4: Transmission error calculated considering the stiffness under load for 0, 25, 50, 75 and 100%
of the nominal power.
Top: Gear set with unmodified profile.
Middle: Gear with small symmetric profile modification (13 µ).
Bottom: Gear with normal symmetric profile modification (26 µ).
 7

The calculation of the transmission error due to tooth mesh stiffness

The stiffness of a single tooth is composed of four important effects. These are:
- Tooth bending
- Shear deformation of the tooth
- Hertzian compression in the contact
- Tilting of the tooth in the gear body
Equations for the calculation of these effects were developed by Peterson [4] for involute gears. The
method can also be adapted to gears with profile modification and to non-involute gears. The stiffness
of two teeth in contact is called ‘Combined tooth stiffness of one pair of teeth’ (c´) and the total
stiffness of all teeth in contact is the ‘Mesh stiffness’ (cγ).
The calculation of the behaviour of the meshing stiffness during a contact cycle is important. For
unloaded involute gears with no profile modification, the path of contact is a straight line, but for real
gears under load the effective path of contact is complicated to find. Due to bending of the teeth the
real transverse contact ratio εα increases. Therefore the contact path has to be calculated step by
step; and for every single step, the effective number of teeth in contact has to be determined. As the
tooth stiffness depends on the load, and the load on a single tooth (with constant torque on the pinion)
depends on the number of teeth in contact, the solution must be found through iteration (Fig. 6). The
calculation is time-consuming because the stiffness itself depends on the applied normal force (due to
the Hertzian compression); therefore every single step on the contact path has to be found by double
iteration.
The result is quite impressive. Fig 4 shows a typical example of a gear set without (top), and with
different degrees of profile modification (middle and bottom). The technique of displaying transmission
error under varying load such as 25, 50, 75 and 100% of the nominal torque load, and defined as far
back as 1958 by Harris[5], is a very helpful illustration for evaluating the behaviour of a proposed
profile modification. Most gearboxes do not always run with a steady torque, so the performance of the
gear set should be optimal within a certain

0.004

0.002
T (25%)
0
Delta phi2 (°)

T (50%)
-8 -4 0 4 8
-0.002 T (75%)
T (100%)
-0.004
T (0%)
-0.006

-0.008
phi1 (°)

Fig.5: Gear set with small symmetric profile modification (13µm) as in Fig. 4, but with an additional
sinus-wave form modification in the single contact region of the tooth flank.
Left: The modified tooth form (here exaggerated) with additional thickness 10µm.
Right: Resulting transmission error under different loads

torque range. The transmission error of the unmodified gear in fig. 4 also shows an often discussed
feature of gears: due to the flexure of the teeth, the transverse contact ratio (εα) increases with higher
torque form 1.67 to approximately 1.78 (50% load) and 1.81 (at 100% load).
Any small modification on the tooth profile has an important influence on the transmission error curve
as the comparison of the different gear sets in fig. 4 shows. A new tendency in the development of
optimized gears is to apply a special profile modification during the finishing process for the reduction
of transmission error. When increasing the thickness of the tooth between the limits of the single
 8

contact diameters (the section of the tooth flank with one tooth pair in contact), the value of the
transmission error in this section can be reduced. To demonstrate this, fig. 5 presents the transmission
error showing the effect of a sinus-wave like modification. It is evident that such a modification
produces higher transmission errors in low load conditions, but in the nominal torque range the change
of the transmission error is significantly reduced. The behaviour of the transmission error can be
further improved when using other modification forms. The sinus-wave was chosen arbitrarily to
demonstrate the principle.

It is well known, that fabrication errors (such as pitch deviation) have an important influence on the
transmission error of real gears. These effects can be analysed by applying an error to the tooth form.
For example, the distance between two flanks can be increased by the amount of the pitch deviation
error before calculating the transmission error.
 9

Start

Define Nominal Torque (assume constant)

Set φ1 for pinion (Loop over 1 pitch in

Nsteps steps of ∆φ1 )

Set φ2 based on transmission ratio.

Estimate ∆φ 2

Calculate distance, δ, (overlap) for each

valid section, ‘j’, on each tooth,’i’.

Calculate Torque (tangential

component of force) based on overlap,
stiffness function, contact position.

Compare theoretical torque with torque produced

Calculate corrective by contact.
measure ∆φ 2

No Inside limit?

Yes

Store ∆φ2

Yes Next
position?

No

Finish

Fig.6: Calculation of the transmission error of a gear pair under load

 10

Gear Stress Analysis for Optimised Involute Tooth Forms

A reliable algorithm was developed for the strength calculation of optimised gears. All normal
calculation procedures determine the stress in the tooth root via a simplified model of the real
conditions. According to ISO 6336, the critical cross section in the tooth root has to be found at the
contact point of the 30° tangent in the root contour. According to AGMA2001 (with AGMA908-B89),
the Lewis parabola is fitted to the tooth form, where the point of contact of the parabola with the tooth
root rounding determines the critical cross section. Depending on the actual shape of the tooth root
rounding, a more or less greater error is implied. In the publication of B. Obsieger [6] some years ago,
a substantially improved calculation method was proposed. Based on the actual tooth form, the tooth
form factor (ISO:YF, AGMA:Y) and the stress correction factor (ISO:YS, AGMA:Kf) are calculated at
each point in the tooth root area and subsequently the location of the maximum of the product
(ISO:YF*YS, AGMA: Y/Kf) is determined (Fig. 7).

Figure 7: Derivation of the tooth root stress at any location of the root between YF(r) and YS(r)

This calculation procedure is integrated into the KISSsoft software. The critical tooth root cross section
can be determined based either on the force application on the tip (AGMA: Loaded at tip), or the force
application at the single point of action (AGMA: Loaded at PSTC). The strength calculation according
to ISO or AGMA is then completely performed using this specific data. The course of the stresses can
also be graphically shown (fig. 8). This method gives good results when compared with FEM
calculations. The effect of a grinding notch is also included in this procedure. For a resistance
 11

verification following either AGMA or ISO, the point of maximum stress is identified, and the calculation
is executed at this specific point. As the ISO standard explicitly stipulates a method A, for an improved
algorithm based on the same basic philosophy, this type of calculation is 100% in accordance.

The computation of the Hertzian pressure along the tooth flank is also calculated based on the actual
tooth form. For each point of action the corresponding radii of curvature are determined for both gears
and starting from this the pressure is computed.
With the same data, the calculation of the sliding speed is possible as well as the computation of the
local contact temperature, the efficiency, and the safety factor for scoring of any gear pair.

Fig. 8: Tooth root stresses (and Hertzian pressure), calculated on the base of the actual tooth form
and the meshing conditions.
 12

Gear Stress Analysis over the Path of Contact

The procedure for the calculation of the transmission error (fig. 6) also provides the normal force and
the point of contact on the tooth during the meshing cycle. This permits an even more precise
analysis. Fig. 8 shows the Hertzian pressure and the maximal tooth stress calculated with the actual
normal force.

Fig. 9: Left: Normal force; Right: Hertzian stress and tooth root stress during a meshing cycle. (Same
gears as in fig. 4 at 100% load.)
 13

As the standards (AGMA, ISO) are based on the calculation of just one situation, i.e. the application of
the normal force in the point of single tooth contact, the more realistic calculation of the normal force
with the corresponding stresses in every position is far more precise. However, the result may be quite
different from the results of the procedure in the standards.

Conclusion
Using advanced software for the simulation of the manufacturing process with one or more different
tools and production methods, the effective tooth form is calculated. Based on this data, the
calculation of the root resistance and transmission error under load is performed. An enlargement of
the root resistance calculation is proposed, so either by AGMA or ISO standards can be applied to any
tooth form, for example also for form-ground gears. For the development of low-noise gear sets, the
calculation of the transmission error under different loads is very important. The software permits the
study of the effects of different profile modifications on the behaviour of the gear set under load, and to
find the best solution for the required torque range.

[1]: AGMA2001-C95, Fundamental Rating Factors and Calculation Methods for Involute Spur and
Helical Gear Teeth, AGMA Standard, 1995.
[2]: ISO6336, Part 1-3, Calculation of Load Capacity of Spur and Helical Gears, 1996.
[3]: Gear and Gearbox Design (with KISSsoft), Gear Technology, January 2004.
[4]: Peterson, D.: Auswirkung der Lastverteilung auf die Zahnfusstragfähigkeit von Stirnradpaarungen,
Dissertation, Braunschweig, Germany, 1989.
[5]: Harris, S.L.: Dynamic Loads on the Teeth of Spur Gears, Proc. Inst. Mech. Eng, vol 172, (1958) S.
87-112
[6]: Obsieger, B.: Zahnformfaktoren von Aussen- und Innenverzahnungen, Konstruktion 32 (1980) S.
443-447.