Six-Sigma and Reliability Dave Stewardson - ISRU with Froydis Berke - Matforsk Norway Soren Bisgaard - USA Poul Thyregod - Denmark Bo Bergman - Sweden
Pro-Enbis All joint authors - presenters- are members of: Pro-Enbis and ENBIS. This presentation is supported by Pro-Enbis a Thematic Network funded under the ‘Growth’ programme of the European Commission’s 5th Framework research programme - contract number G6RT-CT-2001-05059
Presenters: Introduction Dave Stewardson - ISRU and Froydis Berke Matforsk
Rational for Six-Sigma Improve processes Team - project based improvement Properly costed benefits Grow your own expertise Visible success Use of modern improvement tools
Rational for Modern Maintenance Preventative maintenance Condition monitoring Better planning Less machine downtime Operators monitor machine and process condition
Rationales Fit! Everyone involved Monitoring to help operators get better control over the process Publicise success Soren Bisgaard Leading Industrial Statistician
Maintenance and Reliability We can use six-sigma to crack maintenance problems Strategy is the same What is ‘reliability’ ?
Synopsis of Reliability Some Definitions 1) “The probability that the product continues to meet the specification”. 2) “The probability that an item will perform as required, under stated conditions, for a stated period of time”. 3) “The mean lifetime of a product”. 4) “The likelihood that a product will survive stated stresses”. 5) “The survival rate of something”. 6) “Resistance to failure”. 7) “How long we expect a thing to last”.
Related to: • Quality • Survival • Product Guarantees • Product Improvement • Process Control • Process Capability • Failure Modes Analysis • Problem Solving • Statistical Modelling • Quality Engineering • Preventative maintenance
Web-page example from Quality Digest By Thomas Pyzdek a consultant in Six Sigma. http://www.qualitydigest.com/june01/html/sixsigma.html
Web-Page Example II • Project was initiated by a group of senior leaders, • After receiving numerous customer complaints. • Pareto analysis on customer issues raised in the previous 12 months. • Solder problems were the No. 1 problem for customers.
Web-Page Example • A program manager chosen • Six Sigma team was formed • A Master Black Belt provided technical leadership. • The team began working through the design, measure, analyze, improve and control cycle. • Defined critical-to-quality measures, • Pareto analysis applied to the types of solder defects. • A wave solder team was formed included a process engineer, machine operator, an inspector and a touch-up solder operator.
Web-Page Example • A Black Belt providing training • The team identified and assigned various tasks, • data collection, • creating "as is" and "should be" process maps • Performed process audits.
Web-Page Example • Discovered: • ‘Touch-up’ was performed before any data were collected. • Because solder problems were routine, touch-up was considered part of the soldering process. • There were 24 full-time personnel and four full-time inspectors assigned to touch-up. • Most of the defects were touch-up defects, not wave solder defects. • The equipment desperately needed maintenance. • No preventive maintenance program was in place.
Web-Page Example Recommended several immediate changes: 1. Conduct inspection immediately after wave solder and before touch up. (Process Change! djs) 2. Use a control chart to analyze the results. 3. Perform a complete maintenance of the process.
Web-Page Example • Defects dropped by 50 percent within a month • Began DOEs • DOEs revealed that the majority of prior assumptions were false • sometimes the results were precisely the opposite of the accepted point of view. • Significant quality and cost savings resulted as the new knowledge was used to modify procedures.
Web-Page Example . Eventually defect rate in the area dropped by 1,000 percent over a period of 10 months. Productivity increased by 500 percent in terms of labor hours per board.
From: Using Designed Experiments and the analysis of Statistical Error to determine Change Points in Fatigue Crack Growth Rates. 1University of Newcastle, 2Corus Group UK, 3Instituto de Engenharia Mecanica e Gestao Industrial, Porto, Portugal, 4Centro Sviluppo Materiali, Italy, 5Voest-Alpine, Austria, 6Thyssen Krupp, Germany, 7Sogerail, France
Main Objective Determine the effects of stress ratio and relative humidity on the fatigue crack growth rates measured in grade 260 rail steel - Reliability Approximately 75% of the rails currently produced for use in Europe are 260 grade. Started just before Hatfield crash!
The Reliability Test Rail samples subjected to variable stress levels under a constant cycle Crack introduced into the sample Growth of crack measured over time against number of cracks Analysis of da/dN verses the stress intensity
Experimental Design Two stages, first considered a screening stage involving 2 Labs only. Design constrained by limit on material resource. Biggest problem - how to interpret the data?
Plan for second stage 1) Stress Ratio is important so fix it at a convenient value 2) Add Cyclic Frequency as a factor 3) Just monitor Relative Humidity and Temperature
Project Findings • Found most important factors • Can now set these at optimum • Found a good way to use the data • Can monitor the quality of rails • Better understanding of factors effecting reliability of rails
Conclusions were Experimental design helped to discover the important factors that effect these types of Reliability test. It is also possible to derive quality monitoring of the test data using charts of the Plot parameters; slope, error and intercept. Corus engineers now use these methods - training by ISRU
Test Equipment Condition Monitoring Ericcson (Sweden) Routine testing of electric components If test kit failed (equipment not working) Could fail a good component Conducted designed Experiment to optimise a monitoring scheme
Condition Monitoring II Discovered potential problems with kit Found an optimum scheme Developed control charts Discovered that the number of tests per day was not the major influence The worse the product quality, the more likely the test kit would fail to work properly
Condition Monitoring III Other examples: 1. Ohio – monitoring of large weighing equipment (50 Tonnes) Effected by by weather – and animals 2. Monitoring of measuring equipment used for calibration – Electrolux
General Problems Lack of good data Spend time to collect this But then USE IT Must drive it on! Must see benefits quickly!
Best Strategy Involve the operators directly makes it ‘easier’ for the engineers Work as a team