Return to Session Directory

Author’s Name Name of the Paper Session

DYNAMIC POSITIONING CONFERENCE

October 15-16, 2013

THRUSTER SESSION

Gear failures: Lessons learned

By Timo Rauti
Rolls-Royce Oy Ab
 Timo Rauti Thruster Session Gear Failures: Lessons learned

Abstract
In general gear failures are very rare events. The basic dimensioning of the gears is very trivial and
follows commonly used standards. Whenever a gear breaks it will result a total overhaul of the thruster.
This is extremely unfavourable for the business.

The purpose of this presentation is to present the findings from the investigations of the experienced
failures and present new procedures and design changes that make the thrusters even more robust.

MTS DP Conference - Houston October 15-16, 2013 Page 1

 Timo Rauti Thruster Session Gear Failures: Lessons learned

Introduction
An azimuth thruster is a configuration of a fixed pitch or controllable pitch propeller put in a housing that
can be rotated in any horizontal direction. The power is transmitted to the propeller using at least one gear
in the construction, which is located in the underwater housing. This kind of one geared thruster solution
from Rolls-Royce Oy Ab is mainly used in offshore and the thruster units are called UUC. Thrusters that
have two gears are called US type thruster and those are mainly used in commercial applications.

Figure 1: Rolls-Royce Oy Ab UUC type thruster

Rolls-Royce Oy Ab has also products specially made for operating in heavy ice conditions, even in ice
breaking operations. These units are called ARC units and their design philosophy has completely
originated from ice management demands.

Figure 2: ARC type unit at Rolls-Royce Oy Ab factory

MTS DP Conference - Houston October 15-16, 2013 Page 2

 Timo Rauti Thruster Session Gear Failures: Lessons learned

Gears in Azimuth thrusters

The gears that are used in Azimuth thrusters are spiral bevel gear type. It means that the gears are in
90deg angle to each other. The gears flanks do not have a constant helical angle as it grows from the inner
diameter to the outer diameter.

Figure 3: Spiral bevel gear [4].

Figure 4: Spiral angle relation [4].

Since the spiral angle is not constant on the flank and the gear has many other special features compared
to for example more regular spur gear. The production of the gear is complex and demands high skills and
precision from its maker.

MTS DP Conference - Houston October 15-16, 2013 Page 3

 Timo Rauti Thruster Session Gear Failures: Lessons learned

Characteristics of a spiral bevel gear

In order to understand some of the issues handled in this document few terms have to be introduced.

• Concave and convex flank, tip and root, heel and toe

A tooth has two flanks; concave flank and convex flank. The outer radius of the flank is called heel and
the inner radius of the flank is called toe. Tip and root are self-explanatory in the figure 5.

Figure 5: Description of tooth terms [2]

The pinion (the gear with fewer teeth) will be driving the wheel (also called ring gear) with its concave
flank. The wheel’s convex flank is then driven. This is the normal rotational direction of the gears. The
gears can also be driven with opposite direction but this is not normal practice. When the gears are run at
normal direction the gear contact for a flank will start from the toe and will build up towards the heel.

• Backlash

The term backlash means the distance from the pinion convex flank to the wheel concave flank. For the
spiral bevel gears used in Azimuth thrusters there always has to be some backlash. The gear designer
should choose the backlash so that this will be true during the operation of the gears other vice there
would be also contact on the reverse flanks which is not acceptable for gears in Azimuth thrusters.

MTS DP Conference - Houston October 15-16, 2013 Page 4

 Timo Rauti Thruster Session Gear Failures: Lessons learned

Figure 6: General representation of the backlash (spur gear)

There are actually infinite different ways to measure backlash. One can measure it from the pinion or
from the wheel and from different radius of them both. For a spiral bevel gear also the direction of the
measurement can vary and is important since the flank helical angle changes. Therefore normal backlash
for a spiral bevel gear will be different to tangential backlash measured from the same radius.

A standard way to define the backlash is to give a normal backlash at outer diameter of the wheel. If other
definitions are needed those can be calculated from this value.

• Contact pattern

As already mentioned the gear contact will start from the heel and it will end at the toe. As the gear
rotates a momentary contact will move along the flank sweeping it. A contact pattern is then an imprint of
these moments on the flank. Rolls-Royce Oy Ab provides a torque test on each pair of gears that it
delivers and the contact pattern is best visualised during that procedure. Contact patterns can also be
simulated using special software e.g. Becall.

Figure 7: Contact patterns at the full torque test

Special paint is put on the flanks and gears are rotated using design torque test.

The size and the location of the contact pattern will change with different loadings. The location is
changed because the gear box will deflect and change the relative position of the gears. The size is of the
pattern related to the magnitude of the load.

What is important is that the contact pattern should stay on the flank and it should be centred even under
the maximum allowable load (not biased towards the root or the tip).

• Mounting distance

Mounting distances (MD) for pinion and wheel are unique for both and it determines their position related
to each other. MD for a pinion is the distance from any position (one should be chosen) from the pinion to
the axial centreline of the wheel. MD for the wheel is the distance from any position from the wheel to the
pinion axial centreline. When the last phase of the gear manufacturing process is made only then is the
accurate MD distance are known and the gear maker will stamp these values on the gears. In the Figure 8
an example of the distances are presented.

MTS DP Conference - Houston October 15-16, 2013 Page 5

 Timo Rauti Thruster Session Gear Failures: Lessons learned

Figure 8: One example how to define the mounting distances, red for the pinion and green for the wheel.

The red line is the MD for pinion and the green line is the MD for the wheel. The pair of gears should be
set into the housing so that these distances are true. If so then the gears are in correct position related to
each other. This means that also contact pattern and backlash will be as planned.

• Mounting distance relation to the backlash

One should recognize how the change of mounting distance of pinion or wheel changes the backlash. In
Figure 9 this is illustrated for one random gear geometry. If both mounting distances are set up correctly
we are in the origin (blue dot) of the chart and the backlash value is 0.55mm as wanted. If the wheel is
moved towards the pinion (minus direction) app. 0.25mm (green dot) our backlash has reduced to a value
of 0.35mm (green line). If we again start from origin we would have the same effect if we moved the
pinion app. 0.5mm (red dot) towards the wheel. Our backlash would again be 0.35mm. In reality the
backlash will change due to combination of pinion and wheel movement.

MTS DP Conference - Houston October 15-16, 2013 Page 6

 Timo Rauti Thruster Session Gear Failures: Lessons learned

0,8
pinion MD
change

0,6

0,4

0,2
Backlash 0.55mm
wheel MD (nominal)
change Backlash 0.75mm
0
-0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8
Backlash 0.35mm

-0,2

-0,4

-0,6

-0,8

-1

Figure 9: MD relation to backlash for random gear geometry.

Since there is eventually a line for a zero backlash it is vital to know how much movement for a pinion
and for a wheel is allowed before the backlash reduces to zero or close to that.

As it can be seen the movement of the wheel has more effect on the backlash. Therefore the backlash is
set up by adjusting the wheel location.

MTS DP Conference - Houston October 15-16, 2013 Page 7

 Timo Rauti Thruster Session Gear Failures: Lessons learned

Gears in correct position

A pair of gears is in correct position when two of the points in Figure 10 shall come true.

Backlash

Gear in correct
MD pinion
position MD wheel
2/4

Contact
pattern
(no load /
laod)

Figure 10 Gear location check methods.

A short classification of the gear failure types

Gear failures can be classified into numerous different modes. AGMA 1010 E95 presents the following
table of different types of failure modes.

MTS DP Conference - Houston October 15-16, 2013 Page 8

Timo Rauti Thruster Session Gear Failures: Lessons learned

Table 1. Nomenclature of gear failure modes [1]

MTS DP Conference - Houston October 15-16, 2013 Page 9

 Timo Rauti Thruster Session Gear Failures: Lessons learned

Figure 11: A damaged gear

In this context we just take it as a fact that the failures that have occurred in Azimuth thrusters have been
of the type pitting, scuffing or subsurface fatigue (or combination of these). One common factor for these
failures is that an overload will cause them all.

Findings
Here is a listing of the findings of the few gear failures that have affected our thrusters. All the issues
listed here do not apply to all the cases. Corrective actions have been based on these findings.

 Basic dimensioning of the gears is adequate

 No common factor can be found related to material properties and material forming properties
(hardness, case hardening, forging rate etc.)
 Backlash during operation has been zero or close to zero
 Backlash set up at the factory has not been according to the required value
 Geometry error on the flank of the wheel

MTS DP Conference - Houston October 15-16, 2013 Page 10

 Timo Rauti Thruster Session Gear Failures: Lessons learned

Improvements made
In the following the improvements are described.

• Deflection calculation and gear flank specification

When the UUC thrusters were designed many years ago the correct location for the zero load contact
pattern was defined by trial and error and tested in a full torque tester. Still today it is sufficient if this
method is used. However if some corrections or better control is needed the deflections of the
environment of the gears has to be put in to numbers.

The relative movement of a pair of gears can be determined for example with the following Gleason
parameters.

Figure 12: Gleason parameters [4]

P is the axial movement of the pinion. G is the axial movement of the wheel. E is the hypoid error
between the pinion and the wheel. S is the angle change between the pinion and the wheel. With these
four parameters the relative movement between the gears can be fully determined.

In order to determine these values finite element model of the thrusters has to be made. The process of the
calculation is not described here, what is important is that based on the calculation gear flank
microstructure can be now determined by the gear maker.

MTS DP Conference - Houston October 15-16, 2013 Page 11

 Timo Rauti Thruster Session Gear Failures: Lessons learned

Figure 13: Finite element model of a thruster

Gear maker will now provide the flank micro geometry with the no load contact pattern and a simulation
of the full load contact pattern. If everything is correct the pattern at the full torque test should be like the
simulated one.

Figure 14: Zero load contact pattern and the simulated full load pattern

From the Figure 14 can be seen how the initial zero load contact pattern is set to toe on the flank and
because of the deflections of the environment the full load pattern will be in the centre of the flank. The
end result would not be good if the initial location of the contact pattern would be in the centre of the
flank.

MTS DP Conference - Houston October 15-16, 2013 Page 12

 Timo Rauti Thruster Session Gear Failures: Lessons learned

• Relative movement due to temperature differences

When the thruster is running it will generate heat inside the gear box. Main sources of heat are oil
churning, gear contact and bearings. Oil is circulated and fresh cleaned oil is coming to the lower housing.
Since the lower housing of a thrusters is directly in contact with water it will be cooled more than the
power transmission parts inside the gear box. This has the following consequence that the pinion will be
warmer than the lower housing. Therefore the pinion will have relative movement towards the wheel
(pinion MD will reduce).

The phenomenon described above is a known fact. What is new is that since we are able to put the
thruster deflections in to numbers we can also have more control on this issue as well. As all the
deflections of the lower housing has been calculated as mentioned in the previous chapter the relative
temperature difference can been taken as a design parameter ns well. Relative temperature difference is
added to the P deflection parameter of the Gleason parameter.

Figure 15 Simulation of hot and cold situation

MTS DP Conference - Houston October 15-16, 2013 Page 13

 Timo Rauti Thruster Session Gear Failures: Lessons learned

When the gear manufacturer is designing the gear two situations are simulated; so called hot and cold.
Also at the full torque test these two situations are tested. A cold situation is when all the temperatures are
the same. A hot condition is when the pinion is moved towards the wheel in order to simulate the
temperature difference. The purpose of this test is to make sure that the contact pattern will stay on the
flank in both situations.

• Setting up the gears and backlash

Is has been common to set up the pinion and wheel by checking the zero torque contact pattern location
and the backlash. This is sufficient way to have the gears in correct location respect to each other. A plane
has been added to the lower housing in order to have the pinion at the correct position inside the gearbox.

Figure 16: Added plane on the lower housing (red)

In the Figure 16 the red distance in now known since the plane is added. The blue distance is the MD of
the pinion given by the gear maker. We now measure the green distance and set the pinion so that blue +
green = red. When the backlash is set up for its correct value the wheel will then also be at the correct
location.

During the backlash setup there has been deviations in the assembly process and in design process as
well. In order to avoid any misunderstandings when talking and measuring the backlash a term backlash
R100 has been taken into use. There is now one consistent term for the backlash. What it means is that
there is a certain backlash value to be achieved when measured from the pinion at 100mm radius. It is
convenient that the measurement can be taken from e.g. 350mm radius and the result just has to be
divided by 3.5 to get the correct value.

It has been also stressed at the factory that when doing the backlash measurement from the pinion the
wheel shaft (propeller shaft) has to be locked firmly. Special tools have been made for this purpose.

MTS DP Conference - Houston October 15-16, 2013 Page 14

 Timo Rauti Thruster Session Gear Failures: Lessons learned

• Geometry error

It has been concluded that even a minor defect in the gear flank geometry may cause significant increase
in contact pressure thus contributing to the observed failures. Actions have been taken to assure better
flank geometry, especially in heavy ring gears.

Conclusion
Although it is important to investigate and control loads coming from outside to a thrusters the most
important thing is to have everything in order inside the thrusters. A pair of gears in any environment in
any application can destroy itself if for example installed incorrectly.

A gear failure seldom occurs due to one reason only. Usually the unfavourable events pile up eventually
which results a failure. A drastic event that a spiral bevel gear in an Azimuth thruster can face is when the
backlash goes to zero. There will be cross contact on the teeth resulting high loads eventually breaking the
gears.

As a conclusion, the improvements described in this presentation will help us to even further improve the
track records of large bevel gears.

References

1. AGMA 1010 E95 Standard

2. Bevel Gears for Thrusters, Peter Häger, Klingelnberg Söhme GmbH, 2001 DP Conference

3. Drive Line Analysis for Tooth Contact Optimization of High Power Spiral Bevel Gears, Jesse
Rontu, Gabor Szanti, Eero Mäsä, ATA Gears ltd, AGMA Technical paper

4. http://en.wikipedia.org/wiki/Spiral_bevel_gear

MTS DP Conference - Houston October 15-16, 2013 Page 15