12-crs-0106 revised 8 feb 2013 data analytics in electronics manufacturing ieee nsw section stefan...
TRANSCRIPT
Data Analytics in Electronics Manufacturing
IEEE NSW Section
Stefan Mozar
2
Overview
The aim of this presentation is to show how Data
Analytics can be used to make improvements in
manufacturing, and the impact that engineering
has on the process.
Introduction
To illustrate the application of data analytics, an example will be use to illustrate an application.
Such an example is testing of electronic printed circuit board assemblies (PCA). Board testing is disruptive on the manufacturing flow.
Test engineers generally try and test as much as possible to verify a PCA is good.
Testing a PCA, typically takes much longer than the assembly process.
Thus PCAs are first completely assembled, and tested later.
3
Causes of Failure
Failure
Materials
AssemblyDesign
4
How Analytics can Help
Industry 4.0
Big Data or Cloud Computing will help predict the possibility to increase productivity, quality and flexibility within the manufacturing industry and thus to understand advantages within the competition.
5
Using Analytics
6
Forrester Wave TM
The Manufacturing Process
7
Results from Manufacturing
Field Data
Design Data
Reliability
Safety
Pilot Run
What are the Data Sources Available?
8
TestStrategy
Design Evaluation
Pilot Run
Reliability
Safety
Test Specifications
Screening with Simulation
Monte Carlo Simulation– It can predict production yield
– It can isolate design form process
– It provides a lot of data
– The confidence interval can be stated
– No data preprocessing required
9
Simulation Steps
1. Develop an equation to calculate tolerances
2. Identify tolerance for each component
3. Include random number generator
4. Run simulation to see if spread falls within range
5. Take further action if required.
10
Sample of Monte Carlo Analysis
11
The Next Step …
where
Process or design capability analysis tell us about robustness.
12
Additional Statistical Methods
A variety of statistical methods can be applied
Six Sigma Techniques
Optimization Models
13
Application to Failure Detection
14
Design Evaluation
Test Specifications
Safety
Reliability
Pilot Run
&
&
&
&
Redundancy Check
Revised Test Specification
Production Data
High Volume Process
Conclusion
This method is best suitable for high volume production
Be careful as simulation on its own can not detect all potential problems
Data Analytics is a game changer which is turning Research & Development work into a data centric discipline.
15