Volume 3, Issue 2

Reliability Edge Home

Case Study Report Introduction:
Integrating Weibull Analysis into Bodyshop Reliability Engineering

Guest Submission

Dave Whetton
Engineering Quality Manager, Comau Estil UK

Comau Estil UK designs and builds vehicle assembly systems for the automotive sector. Recent advances in lean manufacturing have greatly increased the importance of sound reliability engineering and valid reliability predictions, especially in highly automated manufacturing facilities. This introduction and the case study below demonstrate some of Comau Estil UK’s efforts in response to this demand.

Background: Historically, our reliability models were based upon classical calculations of MTBF and MTTR. We realized that this method carried with it the assumption of constant failure rate. However, the method was easy to use in the absence of statistically acceptable analysis tools. In an effort to improve upon existing methods, Ford Motor Company investigated several options for developing standardized reliability reporting based on calculating actual time-to-failure and time-to-repair distributions. As a result of this study, Ford adopted the use of ReliaSoft’s Weibull++ and BlockSim software for reliability improvement and performance modeling. This has been complemented by Ford’s new corporate Reliability and Maintainability (R&M) specification, which requires standard reports incorporating analysis outputs from the ReliaSoft software. Consequently, the American Institute of Quality (AIQ) has been commissioned to roll out strategic training to promote awareness among the supplier base about the advanced statistical techniques. The Ford expectation has been formalized in the five points of FRED (Failure Reporting, Evaluation and Display). Evolving supplier selection criteria is placing increasing weight on standardized reliability reporting using output from ReliaSoft software for predictive models.

Comau Estil UK Initiatives: Comau Estil UK has seized the initiative to apply advanced statistical techniques by retaining consultancy support from Jambor & Associates and applying comprehensive statistical methodologies for vehicle program launch support, many of which are consistent with the five points of FRED. Our application of ReliaSoft software has been piloted in the Jaguar Halewood and Jaguar Castle Bromwich Body Assembly Plants. Comau Estil UK’s specific case studies have targeted the “X” Type Jaguar bodyside and framing systems and the “S” Type Jaguar front structure system. Even before its acquisition by Ford, Jaguar placed high emphasis on reliability techniques such as Machinery FMEAs, reliability allocation models and control plans. Jaguar has welcomed Comau Estil UK’s approach to analyzing reliability using the ReliaSoft software, not only as the basis for reliability improvement but also for the benchmarking of existing facility to incorporate the results into our throughput simulation model as a predictor of future performance. The FRED report presented below, which represents Comau Estil UK’s standardized reporting format, was executed to evaluate the impact of integrating aluminium joining technology into a facility where spot welding has been historically dominant.

Conclusion: Our current capability includes four workstations with the Weibull++ and BlockSim software. We expect our use of the software to increase and result in more licensed workstations and further training from AIQ. Based on case studies that Comau Estil UK has performed to date, we have identified recommendations for enhanced capabilities that would make the software more useful for bodyshop-specific applications. The areas of innovation would be:

  • Automated scaling of the eta parameter to reflect machine busy time when utilization is below 100%.
  • The ability to start a simulation at an arbitrary point in the life of the tool instead of assuming a starting point of zero.
  • More flexible capability to define shift patterns.
  • Automated capability to combine similar reliability entities into “typical” models (e.g., geometry setting stations, etc.).

Back to Intro

Case Study Report:
Integrating Weibull Analysis into Bodyshop Reliability Engineering

 

Station 003 Front Structure
Jaguar Castle Bromwich
PROJECT – X202
Bodyshop
30/8/02

Dave Whetton
Engineering Quality Manager, Comau Estil UK

Objective(s)
The primary objective of this FRED report is to identify an opportunity to improve and raise the baseline of acceptable reliability levels of the production equipment. This reliability growth can only be accomplished through root cause analysis and then by engineering a sound, verifiable fix.

Introduction
The following report represents an Availability study of the X202 Front Structure Station 003 in the Jaguar Castle Bromwich Plant. ReliaSoft’s Weibull++ MT and BlockSim MT (version 1) software applications were used to produce the FRED charts, as prescribed by Point 1 of the Ford Corporate R&M specification.

Methodology
The raw data set used was from the equipment fault detail report of POSMON (the facility’s event tracking system) beginning 07:30 on 28th May 2002 and ending 12:00 on 19th July 2002. The study included all components in Station 003, as shown in the BlockSim block diagram in Figure 2. During this period, 882 events were recorded in the specified shift pattern. Events were analyzed at three levels: Station, Component (e.g., Rivet Gun, Proximity) and Fault Code.

To make the POSMON downtime log readable by Weibull++ MT, some pre-processing was required. First, the data set was filtered to remove any events that fell outside the specified shift pattern and then events were removed that had a continuous duration over planned downtime. The data set was then entered into the Weibull++ data entry form. For the Weibull analysis, the Rank Regression estimation method was used because of the completeness of the data. Time-to-failure and time-to-repair (or failure duration) distributions with their parameters were then used to build a reliability block diagram in BlockSim. Within BlockSim, a simulation was run for 10,000 minutes (approximately 2 weeks of production) with results calculated for instantaneous availability 10 times. The simulation resolution was set to run 100 inner loops and 10 outer loops. This means that 100 simulation points were generated for each reliability entity and results were returned for each of the 100 runs; then the simulation returned results 10 times at system level, each time with a new stream of random numbers for the simulation points. This yielded 10 different system reliability values and 100 reliability entity values. The system reliability at the specified time returned by the simulation was the average of these 10 reliability values. The simulation results were used to estimate MTBF, MTTR and Availability at the System level.

Similar steps were taken to work down the availability hierarchy from Component level to Fault Code level, as shown on the FRED chart in Figure 1.

Results
Figures 1 through 4 demonstrate the results of the analysis.

Figure 1: Front Structure Station 003 FRED Tree

Figure 1: Front Structure Station 003 FRED Tree

 

Figure 2: Front Structure Station 003 Block Diagram

Figure 2: Front Structure Station 003 Block Diagram

Figure 3: Front Structure Station 003 Pareto

Figure 3: Front Structure Station 003 Pareto

 

Figure 4: Front Structure Station 003 Rivet Gun FRED Tree

Figure 4: Front Structure Station 003 Rivet Gun FRED Tree

Conclusions
The immediate opportunity for reliability improvement follows a critical path of Component = Proximity (98.53% availability) to Fault Code = 102062 waiting part present (98.5% availability).

The study has highlighted the importance of having data in electronic format that can be operated on for pre-processing before entry into the analysis software. The study also demonstrates the value of the ReliaSoft software for performing the analysis and creating graphical representations of the output. Finally, the study illustrates the value of automated analysis to point out the dominant areas and sources of failure as an aid to prioritize reliability improvement efforts.

For Comau Estil UK’s throughput simulation model, the most easily understood input would be the MTBF from the FRED tree generated by BlockSim. It would be preferable to use MCBF, which represents the failure frequency based on station/component busy time.

Back to Intro

End Article

 

ReliaSoft.com Footer

Copyright © 1992 - ReliaSoft Corporation. All Rights Reserved.
Privacy Statement | Terms of Use | Site Map | Contact | About Us

Like ReliaSoft on Facebook  Follow ReliaSoft on Twitter  Connect with ReliaSoft on LinkedIn  Follow ReliaSoft on Google+  Watch ReliaSoft videos on YouTube