Elastodynamics based Modelling of Acoustic Emission for Earlier
Bearing Damage Detection
Anurag Bhattacharyya
1
, Krishnan Thyagarajan
2
, Jin Yan
3
, Kevin Pintong
4
, Qiushu Chen
5
, Joseph Lee
6
, Peter Kiesel
7
, and Kai
Goebel
8
1,8
Intelligent Systems Laboratory, Future Concepts Division, SRI International, Palo Alto, CA, 94304, USA
anurag.bhattacharyya@sri.com
kai.goebel@sri.com
2, 3, 4, 5, 6, 7
Hardware Research Technology Labs, Future Concepts Division, SRI International, Palo Alto, CA, 94304, USA
krishnan.thyagarajan@sri.com
ABSTRACT
It is crucial for many applications to detect bearing damage as
early as possible to allow for scheduling of maintenance with
lead times that minimize operational disruption. State of the
practice is the detection of spalling but damage initiates prior
to spalling as subsurface and surface cracks. Such damage
is much harder to detect and to model. This study proposes
a unique application of the nanofrictional Prandtl-Tomlinson
model to predict macroscopic acoustic emission (AE) sig-
nals that occur at cracked interfaces under relative motion.
The study integrates large deformation modelling of struc-
tures with elastodynamic simulations to investigate early AE
signals generated under different bearing rotational speeds.
Experimental studies are carried out to measure acoustic vi-
brations from metal-metal surface friction using fiber optic
sensors and compared to those predicted by the model. Broad
agreement of results highlights the validity of this framework.
1. INTRODUCTIO N
It is forecast that by 2026, the digital twin market will grow to
a size of more than $ 48 billion (Abraham et al., 2022). With
the advent of the industry 4.0, development of digital twins
has been under a considerable amount of focus to avoid faults
or breakdowns that can affect the operational output and qual-
ity. Through digital twinning, manufacturers can both tweak
designs and monitor assets so they can predict when any parts
might need replacing. As a result, there has been an in-
creased shift from reactive maintenance to proactive mainte-
nance. One of the most important machinery components is
the bearing which is critical to operation of virtually all rotat-
ing equipment. The current state-of-the-art in damage detec-
Anurag Bhattacharyya et al. This is an open-access article distributed un-
der the terms of the Creative Commons Attribution 3.0 United States Li-
cense, which permits unrestricted use, distribution, and reproduction in any
medium, provided the original author and source are credited.
tion of bearings leaves a gap between the initiation of fatigue
cracks and the eventual detection of crack networks. Closing
this gap has the potential to save costs related to maintenance,
labor, and downtime.
Studies have shown that due to repeated loading, high-cycle
fatigue initiates subsurface cracks in the bearing inner-race.
These subsurface cracks eventually form a crack network
leading to spalling. The current state-of-the-art techniques
can detect AE signals generated mostly after the formation of
crack network. Currently used techniques to determine bear-
ing failure rely on monitoring acceleration, velocity, or dis-
placement. Low frequency eddy current measurements (0-6
kHz) rely on displacement detection, laser doppler vibrom-
eters (2 Hz – 12 kHz) measure velocity differences, while
AE detection (50-300 kHz), shock pulse, accelerometers, and
spike energy detection (20 kHz – 350 kHz) detect accelera-
tion changes due to bearing damage. It is posited that early
signs crack initiation are manifested at higher frequencies rel-
ative to signals generated by spalling. Therefore, one requires
the capacity to measure across a larger frequency spectrum.
Currently, most deployed detection platforms interrogate into
small brackets of frequency and there is no single system that
covers the entire range.
While AE signal detection has been suggested for bearings,
the majority that work focuses on damage resulting from ar-
tificially seeded defects (Cockerill et al., 2016)(Al-Balushi,
Addali, Charnley, & Mba, 2010). However, these represent
an already advanced stage of damage. Both Data-driven and
Physics-informed methods have investigated bearing dam-
age detection (Fuentes, Dwyer-Joyce, Marshall, Wheals, &
Cross, 2020)(Lu et al., 2023). Some investigations have
looked into computational simulation of AE signals focused
on crack propagation(Faisal Haider & Giurgiutiu, 2019).
Studies have also been performed to determine the generation
1
