www.mesa.org2008 North American Plant-to-Enterprise ConferenceSeptember 21-23, Orlando, FL
Metrics for Diagnostic Purposes
Steven KaplanGlobal MES Administrator
The following Strategic Initiatives of MESA International are associated with
this presentation:
Lean ManufacturingQuality & Regulatory Compliance
Real-Time Enterprise
Using Metrics to Diagnose Issues = Lower Costs
Cos
t Of
Res
olut
ion
DISCOVERY TIME
COST OFRESOLUTION
Low
High
TrendNCR
CAPA
Complaint
Recall
Immediate Extended
NCR: Nonconformance Report
CAPA: Corrective and Preventive Action
Study Background
• MESA Metrics sub-group: – Several companies, different disciplines and
manufacturing methods
• What was successful – Common metrics or current base line metrics – Unique metrics – Why utilized, drivers, how metric was developed
• Key: What these companies do with the metrics.
Contributing Companies
• Murata Power Solutions - electronics• National Starch - starch• RobMax BiWMetrics - robotics• Ablestik - adhesives• Teknikum Group Ltd. – rubber hoses• Camstar – software
Note: others participated in discussions & review
CorporateGoals Benchmark
Data
DataCollection
Manufacturingranked KPI's
Frequency ofReporting
AdditionalData Input
SupportSystems
Performance Managementin Manufacturing
Deviations
Root CauseAnalysis
Presentationof Data
CorrectiveActions TrendsCheck
Results
AlarmSystem
PatternDetection
PreventiveActions
Definitions
• A metric is a quantitative value measuring or assessing a given process (derived from wikipedia). Some people simply call metrics: Performance Indicators.
• A KPI is a specialized metric assessing performance to a corporate goal (thus the word Key in KPI).– Aggregate data from multiple systems– Set by departments to meet goal– Prioritized by importance
OEE: A Proven Aggregate Metric
Overall Equipment Effectiveness
OEE = Availability x Performance x Quality / 100
86.7% Availability x 93.0% Performance x
95.0% Quality = 76.6% OEE
Corporate
• Corporate goals are not only financial– Balanced Scorecard: customer; internal business
processes; learning and growth; financial • Promote a Non Penalizing culture• Processes must have owners at each area• Plant managers should own the process• Everyone needs to be looking at the same metrics
– Definition– Calculation– Input variables
Support Systems
• Defined as: Systems to collect, measure, evaluate and report data and corporate processes and process controls
• Should include– Web-based tools – Access systems– Access points– Calculation tables
Achieve greater results with: Machine, device or tool integrated visual detection / bar code system
Frequency of Reporting: Considerations
1. Real Time varies by plant: 1X/day, by usage2. Question: is manual input needed for data generation
or processing?3. Frequency ok as a criteria, if the data is truly also
evaluated and studied on the same timescale.4. The frequency of reporting can be driven higher, if data
is available – and evaluation will leverage this addition.
Benchmark Data
Categorize By:Departmental, Corporate, Industry, and the World
• Utilize single measurements that encompass key KPI's– Throughput - Score Carding – Averaging Metrics
• Use historic data and industry benchmarks to set goals• Set higher expectations as progress• Incorporate Lean Manufacturing Principles• Establish Six Sigma Methodology
– identifies areas of concern
Data Collection
• From production line / machine / man• Error messages by department• Quality assurance data: electronic &
manual• Transactional data from MES systems,
LIMS, ERP, etc.
Alarms
A method of signaling the occurrence of some undesirable event
• Email• Text Messages• Visual Alerts • Audio Alerts• Automated Actions
Review of OEE Analytics — Root Cause Analysis
A trigger is set to notify when downtime exceeds 30 minutes in an 8 hour period.
With the trigger, we could have the
notifications 2 days sooner.
This would eliminate
downtime by almost 2.5 hours
Note the downtime occurred 3 days without a trigger for action
Alarm Triggers: Ability to Proactively Prevent Problems
Setting The Goal
• Normalizing a driven goal to a percentage for manufacturing is an obstacle to overcome.
REDUCE SCRAP BY 10%
Presenting the Data
Presenting the Data
Visual Score Card KPIs
This takes all the KPI's in a plant and sets real expectations against them. With this, numerical scores are determined per KPI.
In a multi-plant environment these scores determine who's 'best', to generate improvement via competition.
• Why– To develop daily, forward-
looking metrics that directly correlate with OTS goals
• Demand Forecast• Production Efficiency• Schedule Adherence
• Expectations– Daily Dashboard is updated
each morning– Posted daily in each
manufacturing department– Reviewed once daily at
production planning meetings
Daily Dashboard Metrics
• Why– To develop daily, forward-
looking metrics that directly correlate with OTS goals
• Demand Forecast• Production Efficiency• Schedule Adherence
• Expectations– Daily Dashboard is updated
each morning– Posted daily in each
manufacturing department– Reviewed once daily at
production planning meetings
Daily Dashboard Metrics
Root Cause Analysis & Pattern Detection
Root Cause Analysis• Strong Portals with good
BI tools can drill down to cause of a trend
• Asking 5 times WHY?• Strong failure analysis
team, empowered to take corrective measures
Pattern Detection• Weekly, monthly and
quarterly multi-discipline team reviews
• Productivity by "shifts", with training implied for those with low efficiency
• Compare past data base info on deviations and root causes
Corrective Action
• Drive negative trends back to area of cause then corrective action
• Nominate responsible person• Give deadlines• Tiger team selects actions based on impact & speed
of implementation• Provide on-line help • Auto-trigger with manual intervention option to
confirm "corrective action and follow up" etc...• Review corrective actions in one plant in detail prior
to implementing in other plants
Check Results
– Compare the results achieved with benchmarks before the project was implemented.
– Review benchmarks in a global manner. – Revisit KPIs and metrics definitions to improve
accuracy and ensure they represent the process– Performance Management is a continuous cycle
The Deming Cycle
Plan
Act Do
Check
KPI’s In ActionThe 3 P’s
• Predictive– Automatically and manually analyze patterns and events in real
time– Continuously monitor manufacturing and quality information
• Proactive– Allow automatic and manual intervention and adjustment of
processes– Provide actionable Intelligence
• Preventive– Elevate information to knowledge– Best practices and processes based on process understanding– “Closing the loop”
• Right First Time (Viscosity):2008 Right First Time
88%
90%
92%
94%
96%
98%
100%
YTD Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec
China Japan Korea UK USA Global
Note: To improve RFT in USA, adjustment data is being tracked by product and batch, enabling the development of a paretoanalysis to determine the worst offending products.
Hybrid First Inspection
0100200300400500600
11/2/
2007
11/16
/2007
11/30
/2007
12/14
/2007
12/28
/2007
1/11/2
008
1/25/2
008
2/8/200
8
2/22/2
008
3/7/200
8
3/21/2
008
4/4/200
8
4/18/2
008
PPM
Ablestik
Murata-PS
Ablestik supplies Murata-PS with Epoxies & Pre-formsBoth companies participated in this study
Your KPIs Effect Your Customers’ Results
• Right First Time (Viscosity):2008 Right First Time
88%
90%
92%
94%
96%
98%
100%
YTD Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec
China Japan Korea UK USA Global
Hybrid First Inspection
0100200300400500600
11/2/
2007
11/16
/2007
11/30
/2007
12/14
/2007
12/28
/2007
1/11/2
008
1/25/2
008
2/8/200
8
2/22/2
008
3/7/200
8
3/21/2
008
4/4/200
8
4/18/2
008
PPM
Ablestik
Murata-PS
Your KPIs Effect Your Customers’ Results
Coincidence ?
Challenges
People & Process• Operation that finds a
fault is not always the operation causing it
• Queue & Lead times can prevent quick resolutions to a problem
• Set Up Times not measured
IT Systems• Multiple systems are very
fragmented• Separate quality control
system for corrective action• Some machines do not have
an access point to automatically collect data
• No Communication between the reporting systems
Comments
• Performance management = processes, not a tool or ability to see the KPI
• Actionable intelligence is the goal• Defining & extracting metrics is toughest
– Auto presentation of the data is the 'minor' part of the battle– Junk wrapped in a pretty facade is still junk
• Trends matter more than actual • Focus on the whole context not the metric• Companies missing aspects will fall short
Metrics for Diagnostic Purposes Working Group
• Company Profiles & Submissions: • Steve Kaplan, Murata Power Solutions• Jian Xu, National Starch• Nicolaus von Baillou, RobMax_BiWMetrics• Greg Agnew, Ablestik• Vesa vihavainen, Teknikum Group Ltd• Gilad Langer, Camstar
• Metrics Co-Chairs: Julie Fraser, Principal Analyst, CambashiJonathan Siudut, Exec. Software Proj Mgr, IBMSteven Kaplan, MES Administrator, Murata Power Solutions
• MESA Contact:Brandy Richardson
Next Up For Metrics
• Metrics for Diagnostic Purposes
– White Paper Soon to be available at: www.mesa.org– MESA Member Contact: Steven Kaplan, Murata Power Solutions
• Operational Metrics Ties to Financial Metrics & Outcomes
– MESA Member Contact: Darren Riley, Rockwell Automation
• Plant-Warehouse Metrics for End-to-End Execution Success
– MESA Member Contact: Julie Fraser, Cambashi