ANALYTICAL MODELING AND PERFORMANCE PREDICTION

OF REMANUFACTURED GEARBOX COMPONENTS

1
Raja V. Pulikollu, 1Nathan Bolander, 2Sandeep Vijayakar, 3Matthew D. Spies
1
Sentient Science Corporation, 672 Delaware Ave, Buffalo, NY 14209
2
Advanced Numerical Solutions LLC (ANSOL), 3962 Brown Park Drive, Suite C
Hilliard OH 43026 USA
3
US Army Aviation Development Directorate – AATD, Ft. Eustis, Virginia

Keywords: Rotorcraft, Wind Turbines, Gearbox, Remanufacturing, Fatigue models,

Mixed-elastohydrodynamic lubrication model, Multi-body Dynamics

Abstract

Gearbox components operate in extreme environments, often leading to premature

removal or overhaul. Though worn or damaged, these components still have the ability to
function given the appropriate remanufacturing processes are deployed. Doing so reduces
a significant amount of resources (time, materials, energy, manpower) otherwise required
to produce a replacement part. Unfortunately, current design and analysis approaches
require extensive testing and evaluation to validate the effectiveness and safety of a
component that has been used in the field then processed outside of original OEM
specification. To test all possible combination of component coupled with various levels
of potential damage repaired through various options of processing would be an
expensive and time consuming feat, thus prohibiting a broad deployment of
remanufacturing processes across industry. However, such evaluation and validation can
occur through Integrated Computational Materials Engineering (ICME) modeling and
simulation. Sentient developed a microstructure-based component life prediction (CLP)
tool to quantify and assist gearbox components remanufacturing process. This was
achieved by modeling the design-manufacturing-microstructure-property relationship.
The CLP tool assists in remanufacturing of high value, high demand rotorcraft,
automotive and wind turbine gears and bearings. This paper summarizes the CLP models
development, and validation efforts by comparing the simulation results with rotorcraft
spiral bevel gear physical test data. CLP analyzes gear components and systems for safety,
longevity, reliability and cost by predicting (1) New gearbox component performance,
and optimal time-to-remanufacture (2) Qualification of used gearbox components for
remanufacturing process (3) Predicting the remanufactured component performance.

1. Introduction

The manufacture of gears and bearings is a very energy and resource intensive
process. This is true especially for high value added components used in precision
engineered gearboxes used in energy, automotive, and aerospace applications. The
process begins with refining the metal ore through smelting, forming billets, machining
and forging of components, heat treating, and ends with surface finishing. As shown in
Figure 1, most of the energy is spent in the early stages, not in the later stages. Finishing
accounts for only 5% of the energy consumed in the manufacturing process. Repairs and
refurbishing of components which rely on re-applying such later stage processes will
significantly reduce energy consumption. The result is that refurbishing a component
through applying only a finishing process realizes a 95% reduction in manufacturing
energy consumption.

For instance, high value and high demand spiral bevel gears are widely used in
the tail rotor drive trains of most rotorcraft. The loads associated with the tail rotor drive
train are generally much more variable than
those in the main rotor drive train primarily
resulting from maneuvers. These variable load
operating conditions may cause severe or slight
surface damage to the gears. The slight surface
damaged gears are rejected during inspection
and order is placed to obtain new parts, which is
expensive ($40,000 - $80,000) and time
consuming (6-9 months lead time).

An effort was conducted by the US Figure 1 – Breakdown of energy costs

Army ADD-AATD to investigate an emerging associated with the manufacture of
advanced superfinishing process to repair and gears and bearings
put back the rejected parts in service [1][2].
Army performed series of overload tests on the remanufactured rotorcraft drive train
AMS 6265 (AISI 9310) spiral bevel gears to evaluate the endurance limit. These tests
were performed at the NAWCAD Propulsion System Evaluation Facility (PSEF),
Patuxent River, MD. The testing was conducted in the Helicopter Drive Systems (HeDS)
Test Facility, a primary element of PSEF, under the direction of the Army ADD-AATD.
However, physical testing is time consuming, expensive, and the results are somewhat
limited to a particular configuration of the gearbox. The ability to conduct virtual life
testing using Integrated Computational Materials Science Engineering (ICME) modeling
and simulation techniques employing probabilistic methods will have great potential to
reduce the time and cost associated with the testing of complex gear systems.

Sentient’s microstructure-based virtual modeling and performance simulation method,

DigitalClone can accelerate research and deployment of the remanufacturing techniques
by reducing physical test costs. Further advantage of the DigitalClone analyzed in this
paper include ensuring that the components are removed prior to catastrophic damage,
enabling such a cost savings due to remanufacturability and avoiding service/operational
downtime issues.

2. Microstructure-Based DigitalClone Life Prediction Technology

Sentient developed DigitalClone component life prediction (CLP) models to predict

the fatigue life of the remanufactured (repaired/refurbished) component, thus quantifying
the benefits of such processing and demonstrating the feasibility of such a tool. This was
achieved by modeling the gear design-manufacturing-microstructure-property
relationship, as shown in Figure 2.
Figure 2: Microstructure-based DigitalClone life prediction approach for remanufacturing

DigitalClone integrates a 3-D finite element (FE) model, mixed-elastohydrodynamic

(EHL) model, microstructure model, probabilistic methods, and fatigue damage model in
the simulation process. Figure 2 shows ANSOL’s Calyx multi-body dynamic analysis
(global FEA) model, representative of a full scale gear mesh (Step 1) subjected to service
operating conditions. It is computationally intensive to perform fatigue damage analysis
on global models, as a complete microstructure needs to be modeled. Hence, a
representative volume element (RVE) for the gear was defined (see the microstructure
model shown in Step 2 of Figure 2) to simulate fatigue damage. This sub-model
represents the high stress regions (critical sites for crack initiation) in the global FE
model. Surface pressure and traction profiles (step 3) generated by the mixed-EHL solver
are used to load the model segment in a dynamic solution. Mixed-EHL model can predict
the influence of surface finish (ground finish vs. superfinish) on asperity interactions and
pressure profiles. This approach can be used to perform micro-stress analysis at the grain
level and update micro-stress fields as fatigue damage accumulates and propagates (Step
4). This modeling approach is capable of estimating fatigue damage on a cycle-by-cycle
basis to provide accurate lifing analysis. Total fatigue life is determined at the end of
simulation, when a component fails (Step 5). A new RVE is generated for each fatigue
life simulation to capture the variability in material microstructure (grain size, orientation,
and inclusions), material micro properties (hardness), surface finish (roughness statistics),
and residual stresses (carburization process) using probabilistic methods. This will
provide the fatigue life distribution (Step 6). The DigitalClone model predicts crack
initiation and propagation life by modeling the gear component material microstructure
and variations in the fatigue lives that are vital for the design and analysis of a gearbox.
 3. DigitalClone Demonstration

3. 1 Remanufactured Gear Life Prediction

To qualify remanufactured AMS 6265 spiral bevel gear set (Figure 3), Army
ADD-AATD conducted series of physical tests and evaluated the effects of short to
moderate duration overloads on the bevel pinion performance [1] [2]. Sentient simulated
the physical test loading conditions using DigitalClone technology. This evaluation
consisted of virtual simulation of the actual dynamic overload testing of bevel gears.
DigitalClone modeled the gear mesh and predict stresses. DigitalClone then used
microstructure based probabilistic life prediction methodology (Figure 2) to conduct
multiple virtual tests to predict the probability of tooth contact and bending fatigue
failures (POFC and POFB) over a wide range of loads. Weibull plots shown in Figure 3
summarize the results of the virtual testing that correlated well with Army data [1] [2].
The simulations were suspended when the pinion exceeded 1.0E+10 revolutions (run-out).
DigitalClone predicted that there is a very low probability of a pure bending fatigue tooth
failure occurring at loads levels up to 800 HP. At some load between 800 and 950 HP the
probability of a bending fatigue failure becomes more likely. The POFB for 950 HP is
2.50% and for 1030 HP it is 15.2% considering the number of pinion failures at 
1.0E+10 revolutions. In the case of contact fatigue, the simulation results indicated that
the high contact pressure at the edge-of-contact in pinion (Figure 3) lead to micro-pits
formation and eventually resulted in tooth fracture, in agreement with Army ADD-AATD
physical test observations. The damage is likely the result of the lack of adequate tip
relief at the very high overload conditions.

3.2 Improve Gearbox Performance and Trade Analysis

DigitalClone technology can also be used to simulate the life of gear and bearings
under a wide variety of surface finish, residual stresses and other scenarios that would
otherwise be too expensive to physically test. This would assist in improving gearbox
performance and performing trade analysis.

Figure 4a shows the influence of surface finish on surface pressure/traction profile

and gear fatigue life. Sentient’s mixed-EHL model accurately predicted the influence of
surface finishing methods (Superfinish, Root Mean Square Height Rq: 3.0 µin and
Ground finish, Rq: 21.63 µin) on traction profiles. It is evident from Figure 4(a) that
improving the surface profile through a superfinishing process reduced asperity
interaction, thereby improved the fatigue resistance over ground finish gears. The contact
fatigue life of superfinish gear was about 1.81 times greater than the ground finish gears.
Figure 4(b) compares surface fatigue life predictions of carburized gear (surface
compressive stress of 145 Ksi), and carburized and shot peened gear (surface
compressive stress of 174 Ksi). The mean fatigue life of shot peened gear is 2.0 times
higher than carburized gear due to higher surface and subsurface compressive stresses.
This analysis demonstrates that DigitalClone can be used as the alternative to physical
testing.
Figure 3: DigitalClone predictions of percentage of remanufactured AMS 6265 pinions failed
vs. pinion revolutions for four different Horse Powers. DigitalClone predicted run-outs (no
fatigue damage) up to 800 HP, in agreement with Army endurance test results [1] [2].

4. Summary
Sentient’s DigitalClone technology assists in remanufacturing of high value, high
demand rotorcraft and wind turbine gearbox components. The benefits of this technology
includes (1) Reduce physical testing using virtual testing and provide cost savings, (2)
Decrease the energy, material resources, and costs associated with manufacturing, (3)
Ensure product performance is maintained/improved as a result of the process
modification or enhancement.

Figure 4: DigitalClone simulates the effects of surface finish, residual stresses on fatigue
life that would otherwise be too expensive to physically test

5. References
1.E. C. Ames, Repair of High Value/High Demand Spiral Bevel Gears by Superfinishing,
American Helicopter Society 67th Annual Forum, Virginia Beach, VA, May 3-5, 2011
2. L. Haney and E. C. Ames, MH-60K Tail Rotor Drive System High Load Endurance
Test, Report No: NAWCADPAX/TR-2012/75, April 2012