Description

This curriculum spans the technical and organisational challenges of deploying statistical process control in complex, cross-functional environments, comparable to a multi-phase operational excellence initiative involving process engineers, quality teams, and data analysts across manufacturing and regulated production settings.

Module 1: Defining Process Performance Metrics and Baselines

Selecting primary versus secondary metrics based on operational ownership and data availability across departments.
Establishing baseline performance using historical data while accounting for seasonal variation and known process changes.
Resolving conflicts between operational teams on metric definitions, such as yield versus throughput in manufacturing lines.
Implementing data validation rules to prevent inclusion of outlier events (e.g., equipment failure) in baseline calculations.
Aligning metric precision with measurement system capability, particularly when gage R&R results indicate marginal reliability.
Determining the required sample size and data collection frequency to achieve statistically stable baselines without overburdening operations.

Module 2: Measurement System Analysis and Data Integrity

Designing gage R&R studies that reflect actual production conditions, including multiple operators and shift patterns.
Addressing non-replaceable measurement devices by adapting ANOVA methods for nested or expanded studies.
Classifying measurement errors as systematic versus random based on control chart patterns and correlation with process variables.
Implementing calibration schedules that balance cost, regulatory requirements, and process sensitivity to drift.
Handling attribute data in inspection processes through Kappa studies, especially when pass/fail decisions involve subjective judgment.
Integrating MSA results into data governance policies to restrict use of unreliable data in performance dashboards.

Module 3: Process Capability and Specification Limits

Distinguishing between natural process limits and specification limits when customer requirements conflict with process stability.
Calculating Pp/Ppk versus Cp/Cpk based on whether the process is in statistical control during the assessment period.
Negotiating revised specification limits with customers using capability data and risk-based justification.
Handling unilateral tolerances in capability analysis, particularly in safety-critical or regulatory environments.
Adjusting for non-normal data using transformations or non-parametric methods without masking underlying process issues.
Documenting capability assumptions and limitations for audit purposes, especially in regulated industries like pharmaceuticals.

Module 4: Control Chart Selection and Implementation

Choosing between X-bar/R, X-bar/S, I-MR, or attribute charts based on subgroup size and data type consistency.
Setting initial control limits using phase 1 data and defining criteria for phase 2 monitoring transitions.
Responding to repeated out-of-control signals when root causes are operationally unavoidable (e.g., material batch changes).
Managing false alarm rates by adjusting rules (e.g., Western Electric) based on process criticality and detection sensitivity.
Integrating control charts into real-time SCADA or MES systems with automated alerting and escalation protocols.
Handling missing data points due to equipment downtime or sensor failure without compromising chart validity.

Module 5: Root Cause Analysis Using Statistical Tools

Selecting between ANOVA, regression, and DOE based on the number of suspected factors and feasibility of controlled experiments.
Using multi-vari studies to isolate positional, cyclical, and temporal variation sources in high-volume processes.
Interpreting interaction effects in factorial designs when process steps are interdependent or sequential.
Applying logistic regression to model binary outcomes (e.g., pass/fail) when continuous response data is unavailable.
Validating root cause hypotheses with confirmation runs that account for process drift since initial analysis.
Managing stakeholder resistance to statistical conclusions that contradict long-standing operational beliefs.

Module 6: Design of Experiments for Process Optimization

Defining factor ranges that are both statistically meaningful and operationally feasible within equipment constraints.
Choosing between full factorial, fractional factorial, or response surface designs based on resource and time limitations.
Blocking known sources of variation (e.g., shift, raw material lot) to isolate experimental effects.
Handling hard-to-change factors by using split-plot designs and adjusting error term calculations.
Optimizing multiple responses using desirability functions when trade-offs exist between quality and throughput.
Translating experimental results into standard operating procedures with clear control parameter settings.

Module 7: Sustaining Gains and Control Plan Development

Assigning ownership of control chart monitoring and response actions to specific roles within shift operations.
Integrating control plans into change management systems to assess impact of equipment, material, or personnel changes.
Defining response plans for out-of-control conditions that escalate based on severity and frequency.
Updating process capability and control limits after implemented improvements, with documented revalidation.
Using automated data collection systems to reduce manual entry errors in ongoing monitoring activities.
Conducting periodic audits of control plan adherence and effectiveness during internal quality reviews.

Module 8: Advanced Topics in Non-Normal and Multivariate Processes

Applying non-parametric control charts (e.g., run charts, CUSUM on ranks) when data transformation fails to normalize distributions.
Using multivariate control charts (e.g., T²) to detect shifts in correlated process variables without inflating false alarm rates.
Interpreting contribution plots to identify root variables driving multivariate out-of-control signals.
Modeling time-series data with autocorrelation using ARIMA-based control schemes instead of traditional Shewhart charts.
Handling processes with multiple operating modes by implementing mode-specific control limits and baselines.
Validating stability in low-volume or custom production environments using cumulative sum or Bayesian updating methods.