Case Study

Customer:   KOMATSU

Synopsis:     KOMATSU was looking to improve its ability to predict failures in construction and mining equipment.

After analyzing pleiotropically industrial big data collected from their machines and equipment sensors, we detected hidden anomalies and degradations in moving parts by using defined parameters and original algorithms.

Analyze Big Data of Mining Equipment
Detect Signs of Failures from Sensor Signal
Aim the Maximum Efficiency of Mining Operations

KOMATSU is the top IoT leading company in Japanese industry.
The brand of KOMATSU is established by distinct user service
based on equipment operation data. KOMATSU has top market
share in Japanese construction equipment, and, even in global
market, KOMATSU keeps second position after US-based
Caterpillar. This year, KOMATSU has gone into high gear on the
diagnostics project in a long-term growth business of mining
equipment. It realizes prediction of unexpected failures via
equipment’s sensor signals emitted in the field of mining.
This is the solution provided by collaboration of ISID and
US-based Predictronics, called Intelligent Maintenance.

But Failure Occurs
— The Limit of Condition Monitoring

Copper mine spreads out in the highlands of South America,
Chile. Surface of the mountain has been scraped off. 7 meters
high, max. load capacity of 327 tons, such huge dump trucks
incessantly run back and forth through the path. Most of these
mines including the open-pit copper mines are open 365 days a
year. Similar to a blast furnace and a chemical plant, mining is
continued day and night. If a failure stopped equipment, the
downtime shall directly affect to the business loss.

“The productivity is critically reduced if a huge dump truck is just
stuck in the middle of the way” said Director Hisashi Asada, a
leader of Business Innovation Promotion Division at KOMATSU.
He is responsible for the data science organization, and working
to maximize efficiency of mining operation using sensor
information from the equipment.
Mr. Asada explains “We provide not only regular maintenance,
but we also provide condition monitoring to detect anomalies by
machine’s vital signs (health index) to reduce life cycle cost of
mine construction equipment. For example, there is blow-by gas
pressure in the diesel engine (pressure in the crank case), and this
value rises if wear of the piston ring begins. We detect anomaly
or degradation from that.”

However, Mr. Asada pointed out this approach is not perfect.
“Vital sign health monitoring is only effective when you know a
part is about to break and where the sensor signal is clear.
Since mining equipment is used various ways in the field, there
shall be many noises along the sensor data, so then it is difficult
to acquire stable data. Unfortunately, not all the failure can be
found from the vital sign. You can keep the vehicle healthy by increasing the number of regular inspections and cycle of
maintenance. However if the dump truck frequently stopped to
exchange parts, maintenance cost increases and productivity
decreases. As the result, customers become unhappy. We’ve
been looking for a new prescription to reduce the down time”.

Data Mines Anomaly
— The Advanced Data Science revealed its high potential.

While Mr. Asada was struggling to find a breakthrough to
reduction of downtime within R&D at KOMATSU, he asked
several IT vendors to examine benchmarking on each method in
autumn 2015.
Mr. Asada recalled, “I felt improvement of downtime reduction by
condition monitoring reached plateau. Although knowing its
difficulty, I stepped into the world of predicting failure which
required advanced data analysis”.

Our mission is to provide absolute No.1 Dantotsu service
to our users in ahead of any other competitors,
and to let customers recognize KOMATSU as essential.
In such case, ISID is reliable.

Analysis of ISID excelled other competitors
in terms of data processing and detection accuracy
as well as feasibility, but we were mostly impressed
by their distinguished amount of experiences.

KOMATSU asked the analysis and anomaly detection to different
vendors by providing the past failure data, in which its source or
background of values were masked.
“We wanted to see if statistical and mathematical approaches
overcome the problem derived only by the solution with purely
numerical values, rather than solving structural and equipment
mechanisms.” Mr. Asada expressed his honest intension.

ISID joined this benchmarking and analyzed using Intelligent
Maintenance methods built upon Predictronics extensive past
experiences. The Intelligent Maintenance analytic method is to
analyze pleiotropically industrial big data collected from machines
and equipment sensors. ISID and Predictronics performed to
detect hidden anomalies and degradations in moving parts by
using defined several parameters and original algorithms.
Mr. Asada told, “Analysis of ISID exceled other competitors in
terms of data processing and detection accuracy as well as
feasibility, but we were mostly impressed by their distinguished
amount of experiences. Their report included some comparison of
the past failed analysis results. These evidences showed their
methods and techniques were developed by actual experiences.
Failure is an option for technology achievements in R&D”.

Based on this evaluation, KOMATSU began the validation project
of Intelligent Maintenance with ISID at Chile mine on March 2016.
Mr. Asada adds, “We opened masked information at
benchmarking, and now we verifying validity and effectiveness of
Intelligent Maintenance together. A delighted vision, which had
never been seen, for future failure detection, which had never
been achieved, has positively accepted by site workers”.

“Data Driven”
— The Key to Dantotsu Service on Era of IoT

“It takes some more time to see actual result.” said with this
preamble, Mr. Asada still suggests the benefit of applying
predicting failure based on Intelligent Maintenance technique to
KOMATSU business. “Construction and mining equipment
business is not ended when products delivered to customers.
It always comes along with services such as spare parts
supplying, parts repairing, and operational improvements.
Of course, competitors will attempt to enter here so it is
necessary to differentiate from them. KOMATSU advantage is
that we can expand data driven services in the world of ICT, and
take actions based on data. This failure prediction technique
becomes one of the tools.”

KOMATSU is known its brand as IoT manufacturing company by
selling data driven user service and solution such as “KOMTRAX,”
“KOMTRAX Plus”, “Autonomous Haulage System”, “smart
construction”, etc.
Mr. Asada recalls his feeling, “Glad to start together with ISID in
the era of IoT. Our mission is to provide absolute No.1 Dantotsu
service to our users in ahead of any other competitors, and to let
customers recognize KOMATSU as essential. In such case, ISID
is reliable. Their technology and wisdom based on their
experience will prove their strength in such rapid competition.”