Software predicts fatigue life from strain data

Edited by the Engineeringtalk editorial team Oct 14, 2009

Caterpillar is using Safe Technology's Safe4fatigue to predict fatigue life from measured strain data.

Caterpillar uses a wide range of techniques to develop powerful, fuel-efficient and reliable engines.

The genesis of the new-design process at Caterpillar is heavily computer-aided, with an integrated CAD and FEA process.

In 2002, Caterpillar selected Sheffield-based Safe Technology's Fe-safe software for fatigue life prediction and it is now an integral part of its overall design process.

In addition, Caterpillar employs a number of other methods to improve and support existing designs in the field.

An example of this involved a piston experiencing a few field failures somewhat earlier than anticipated.

The original fatigue prediction for this component had only involved hand calculations.

This was a good opportunity to use a test-based process and Caterpillar engineers used the advanced fatigue analysis and signal processing module in Fe-safe, Safe4fatigue, measuring strains to predict fatigue life.

Lab engineers arranged instruments around the piston skirt with rosette gauges at three locations.

Signals were measured corresponding to the three components of strain.

The test loading of a piston is complex, due to the various stages of the internal combustion process.

A load cycle consists of two complete revolutions of the crankshaft.

As the piston is loaded and unloaded, each strain gauge produces three signals.

The signals are then processed by Safe4fatigue to create valid strain histories.

The measured strain approach has the advantage that strains directly reflect the physical hardware and actual loading; whereas the accuracy of the finite element approach entails careful application of loads and consideration of the mesh density near the areas of high strain.

Material properties add complexity to accurate fatigue assessment and prediction.

The piston skirt is made of cast aluminum, which exhibits behaviour somewhere in between brittle and ductile behaviour.

In order to consider situations for both material types, Caterpillar assessed the fatigue life in Safe4fatigue using both maximum principal stress algorithm (based on Nasa MSFC-388 S-N curve data) and Brown-Miller method (based on material data from an ASM book).

The former algorithm is appropriate for brittle material while the latter (Brown-Miller) is more appropriate for ductile material.

The effect of mean stress correction was also investigated using three different methods: Goodman correction, the traditional hand calculation method and a mean stress correction curve for cast AISI material.

In addition, the effect of surface finish was also considered in the Safe4fatigue calculations.

The minimum life was predicted by Safe4fatigue at the notch area of the piston skirt.

This correlated with field observations.

Using the maximum principal stress method, life was predicted at approximately 1.3 million cycles; the Brown-Miller method predicted around 1.6 million cycles until failure.

In areas of high compressive stress, which is the case at the notch, the effect of mean stress was considerable.

The hand calculation, which used a default mean stress correction curve, had predicted a fatigue life at the notch of more than 2.5 times the life predicted by Safe4fatigue, which used a more accurate mean stress correction.

Considering the effect of surface finish also proved vital.

Fatigue life decreased about 60 per cent between 'polished' and 'as-cast' surface finishes (the polished surface has the higher fatigue life).

Overall, Caterpillar found that Safe4fatigue accurately predicted fatigue life from strain gauge data, and proved to be a useful complement to Fe-safe and its other predictive and test methods for product design.