Author: TP Fanning, BSc, senior process excellence engineer, KCI Manufacturing Ireland, Athlone Is overall equipment effectiveness the best fit for high-volume manufacturing? Are there viable alternatives? At KCI, an Acelity company, we asked these questions and found an approach that best suited our needs. Could this approach help your company, too? The challenge to improve quality and reduce costs is a constant driver in medical-device companies. As a result, the Athlone manufacturing site targeted three key objectives in 2012:

  • Leverage overhead by achieving significant volume increase and adopting an ‘in-house make-versus-buy’ model;
  • Product and process changes to drive direct material cost-reduction with no negative impact on availability and service levels;
  • No increment to the installed base, fixed or variable overhead.

Measuring stability and stability thinking

[caption id="attachment_18567" align="aligncenter" width="724"]Stable operations Fig 1: Production-line hourly output stability graph (September 2012)[/caption] The measure defined for stability is output per hour. In practice, when plotted, histograms of hourly output tend to be negative-skewed distributions where conventional descriptive statistics are not adequate to describe them, i.e. they are non-parametric. For ‘non-parametric’ distributions, more often alternative parameters apply:
  • 1st Quartile: 25th percentile value of hourly output,
  • 2nd Quartile: 50th percentile value of hourly output, mid-point in the distribution, e.g. sensor misalignment or grippers stuck;
  • 3rd Quartile: 75th percentile value of hourly output, >Q3 are hours of high output and ‘special causal’ drivers, e.g. false-alarm reset or preventive maintenance overrun.
Between the 1st Quartile and 3rd Quartile are 50% of all hours and also the ‘common causal’ frequent events. Using the parameters, we can express output stability (stability factor) as a ratio: SF = Q1 divided by Q3 [caption id="attachment_18568" align="aligncenter" width="890"]Stable operations Fig 2: Stable operations lifecycle[/caption] Stable operations require a paradigm shift. The emphasis is on understanding and eliminating variation drivers around Q1 daily output performance. Only when output stability has been demonstrated are we entitled to shift the mean. In essence, we get to move the mean by attacking the variance.

Using stability factor to establish line targets

[caption id="attachment_18569" align="aligncenter" width="687"]Stable operations Fig 3: Production line regression plot[/caption] We set a stability factor target of 0.75, but how does this relate to output? Regression-analysis studies assessed how effective quartile performance would be as a predictor of stability factor and, in turn, output. This study illustrated the highest correlation was between Q1 and stability factor (see Fig 3). To achieve stable operations, the following are prerequisites: 1. One common metric at all levels across the entire organisation (stability factor); 2. A robust measurement system. We established a framework for unit measure that was equal for all and, under this, we specified the rules:
  • Smallest entity, the unit is granular – output per hour;
  • Hours off target (
3. Data drives all decision-making to ensure confidence in the data. Data collection was as follows:
  • By hour: data integrity – off-target defect hours leading to direct root-cause ‘drill down’;
  • By shift: rolling 12-hour shift performance, causal data integrity;
  • Daily: cross-functional challenge defective hour data, direct and root cause (‘Five Whys’);
  • Weekly: data trending to identify common cause variation (80/20 Pareto rule);
  • Monthly and quarterly: key improvement projects driven by data review.
4. Adherence to scheduling rules, which were summarised as:
  • No internal optimisation rules;
  • Agreed planned production interval.
Our data rule was ‘digitisation’: one data source meant visibility of data to all. Firstly, we started with manual data collection and, secondly, we used leverage control systems to assist data collection. This is a proven approach in production and support environments.

How we got it done

[caption id="attachment_18570" align="aligncenter" width="895"]Fig 4 Fig 4: Stability program oversight[/caption] If we have an operational roadmap and approach, then we can focus on a line of sight equal for all, based on output versus target at an hourly level. The review structure was framed as follows:
  • Daily: all decision takers to drive the hour;
  • Weekly: to improve the process;
  • Monthly: to improve business processes.
Process leaders were empowered with a brief for execution of improvement by data analysing and process focus. There was one data source, a common database and easy online availability of the information. The recurring message was ‘one single approach across the board’. [caption id="attachment_18571" align="aligncenter" width="895"]Fig 5 Fig 5: Production line stability outcome example[/caption]

Stability outcomes in summary

Business impacts:
  • Output uplift in the available hours;
  • Required less available hours to produce more output, thus eliminating overtime;
  • Cost reduction in direct material demanded five significant product and process design changes. The newly established focus on output stability aided a quick recovery within 12 weeks without ‘stock out’ or risk to supply continuity;
  • Reduction in planned maintenance downtime, meaning more available production hours instead of overtime;
  • We began the stable-operations journey with an initial measure of 0.46 SF. From here, we set a target of 0.75 with a world-class measure being >=0.9 SF. Within an 18-month period, we had hit a 0.86 SF, and from there, shifted the mean to complete the stable-operations cycle (as shown in Fig 5).
Stability program as enabler to continuous improvement:
  • Every product, every interval – from four to two weeks;
  • Reduction in finished goods inventory.
Lean methodology and kaizen events:
  • Changeover time reduction of 58%: single-minute exchange of die (SMED) work content combination charts;
  • Staffing levels: visual staffing plans, operator flexibility training and certification;
  • Leaning batch design history record (DHR): 60% reduction in manual data field entry, thus greatly improving the Good Documentation Process (GDP).
Process improvement projects focused on defect and variance elimination:
  • Ultrasonic welding characterisation; design of experiments, vital tolerance setting, in process statistical process control;
  • Eliminate vision-system false fails.
Risk-centre maintenance strategy:
  • Failure mode, effects and criticality analysis risk assessment;
  • Failure remediation strategies.
Objectives were met while growing functional competencies – a catalyst to continuous professional development. Many thanks to Jim Grant (director – process excellence, AWT Global Manufacturing Ops), along with the manufacturing staff at KCI who supported this success. ©2014 KCI Licensing, Inc. All rights reserved. DSL#15-0078.UK (2/15)