Paint application is touted to be one of the most complex and demanding activities. Any defects in the painting process can result in poor customer experience and impact the company brand. DataRPM shares ideas on how automakers can improve their output quality, reduce defects and improve the operational efficiencies by applying analytics in the right way across the paint shop.
3. Why are we discussing this today
v The paint shop corresponds to nearly third of the time taken in the automotive build process.
v Automation and Manual Processes:The process of paint application involves a large number of steps with a
high degree of automation along with manual processes and inspections.
v As per research, 40% of the cars that exit a paint shop are likely to undergo some sort of rework (either on a
part or in total).
v Environmental Impact: Paint is one of the highest waste generating process with a major environmental
impact.
v Capital Intensive: Setting up and changes to the paint shop is a capital intensive process (almost a third of the
total cost).
v Automation is generating large amount of data which is unused for asset analysis
3
4. Henry Ford Once Famously Said
4
“Any customer can have a car painted any colour that he wants so long as it is black.”
5. Customer Preferences Have Changed Over Time
Car Colors have always reflected the mindset of the era and hence have a huge impact on the customer purchase mindset
Customer tastes have transformed over time and car manufacturers have to match up to their requirements with no tolerance in
drop of quality
5
All Blacks 1920’s
Wacky 1960’s
Apple Effect of 2000’s
8. Observations
• Entire Paint Shop Process is a 7+ hour multi-step process with a number of manual and
automated steps
• Defects in the paint process can occur due to:
• External Influencers like temperature, dirt, particles, paint quality, mix etc
• Production Line Influencers like fault in the robotic arms
• Manual Influencers like lindt particles, hair, PVC particles from workers
• There are multiple inspections done on the quality of the paint of the car both manually and
through automated techniques to diagnose defects
• In case a defect is observed, it is recorded in the system and the car is sent back for rework
• On average, each manufacturer could spend >£76,000 in re-work costs for every 10,000
cars produced.
***Based on 63 million manufactured cars produced globally in 2012***
8
10. Applying Preventive Maintenance
10
Man Machine Method Material
Identify the Possible Cause Across the Entire Process
Conduct an
Asset Level
Analysis for
each Robotic
Arm
Predict Working
States of all the
assets
independently
12. Process Level Analysis - Root Cause Analysis
12
Apply Data Science Techniques which work on large amount of data, use machine learning to learn and correlate multiple data
patterns and finally create asset level predictive models are the need of the hour.
Analyze Root Cause of the
Errors
• Simultaneously observe the changes in the Vital X’s across the entire
production line to differentiate between normal and abnormal
working conditions
• Using Associative Mining Rules, determine correlations between
defects and observed behavior in order to determine candidates
of causality
• This may be linked to occurrence of a certain operator, a particular time of day, line temperature, number of units
processed before the defect, a particular part of the car, robotic arm settings etc.
• The factors need to be observed independently and in combination to determine the effect of
interaction across the candidate X’s
13. Asset Level Analysis – Engineering Features on Sensor Data
13
• From the previous analysis, there are likely to be certain machines and processes which will get highlighted as Vital
X candidates.
• Owing to high degree of automation, the machine arms are fitted with machine sensor data. These data points need to
be captured and analyzed for potential deviations.
• Building a Predictive Maintenance Pipeline will allow you to ensure that deviations in those machine and processes can be
captured early in the cycle and do not lead to quality disruptions.
Apply Feature Engineering
to enhance machine data
Time to Frequency Domain
Mean, Skewness, Kurtosis
• Feature Engineering is essential to understand the events that
preceded the sensor value at a given time as well as deviations in the
readings.
• These feature are critical to differentiate normal working conditions to
anomalies
Choice of features will vary from asset to asset. Hence, it is imperative that multiple features are created and
the algorithms determine which features need to be selected for the final model.
14. Asset Level Analysis – Determining the Anomaly State
14
Detect Anomalies in
Machine Data
Analyze Root Cause of
Within Machine Error
Unsupervised Learning
Intrinsic Factors
Extrinsic Factors
• Based on the Features created in the previous step, we need to
differentiate between the normal and abnormal working conditions of
the machine.
• The right operating criteria can be influenced by manual rules but the
unsupervised learning algorithm will determine the various operating
states within the Machine
• Associated Conditions in these states will highlight various intrinsic and
extrinsic factors which are causing the machine to be in that state
• This will further help determine the root cause of deterioration of the
machine health
15. Asset Level Analysis - Preparing for the Future
15
Create a Prediction Model
for Every Machine
Predict possible
failures in advance
Reduce Downtime and
improve quality
• All this analysis culminates with an asset level prediction model
which determines the likely machine state in the near future.
• The model is regularly tweaked by a feedback loop which tunes it
based on the changing working conditions of the machine.
• Rather than post-mortem monitoring, the floor managers can be
better prepared to tweak processes if they observe any significant
deviation in the machine.
Recommend Actionables
Improve System
Settings
Adjust Temperature
Range
Increase Paint Density
• Further, basis the strength of the model variables, various
recommendations can be made to the line manager around
improving the process.
• These too are driven by machine observed patterns and basis the
asset conditions