A Stable Process May Still Be Far From Optimal

Six Sigma has transformed the way companies approach process improvement.

It provides a disciplined structure for defining problems, measuring performance, identifying causes, implementing improvements and maintaining the results. Statistical Process Control, measurement-system analysis, control charts and process-capability studies help organizations distinguish evidence from opinion.

Yet one important question often remains after a successful Six Sigma project:

Who continues improving the process after the project is closed?

A process can be stable and capable while still operating below its economic potential.

It may produce within specification but consume too much energy. It may maintain consistent quality while leaving yield, throughput or raw-material efficiency on the table. Its target may be controlled very precisely without being the best available target.

This is where Evolutionary Operation can extend Six Sigma.

The Apparent Contradiction

At first glance, EVOP and Six Sigma may seem incompatible.

Six Sigma aims to reduce variation.

EVOP deliberately introduces variation.

But these are two fundamentally different types of variation.

Uncontrolled variation is noise. It makes production unpredictable and hides the true relationship between operating conditions and performance.

EVOP introduces small, structured and intentional changes. These variations are not disturbances. They are sources of information.

Random variation should be reduced. Designed variation should be used to learn.

Six Sigma creates the conditions under which EVOP can work reliably. EVOP then uses those conditions to move the process toward a better operating point.

Six Sigma Provides The Improvement Architecture

The most widely used Six Sigma improvement structure is DMAIC:

  • Define
  • Measure
  • Analyze
  • Improve
  • Control

DMAIC provides a disciplined sequence for improving an existing process.

EVOP is not an alternative to this structure. It can become a powerful method within it—particularly in the Improve and Control phases.

The connection becomes clear when we examine the principal Six Sigma tools individually.

Define: Select The Right Response

An EVOP program must begin with a clear objective.

Is the goal to increase yield? Reduce steam use? Improve product strength? Increase throughput? Lower raw-material cost? Reduce a by-product?

In reality, industrial processes rarely have only one response.

A higher production rate may increase energy use. Lower raw-material cost may reduce product quality. A change that improves conversion may accelerate catalyst deactivation.

The Define phase helps translate a broad ambition such as “optimize the process” into a measurable economic objective with explicit constraints.

A suitable EVOP objective might be:

Reduce energy consumption per tonne while maintaining product quality, throughput and equipment limits.

This definition prevents the program from moving toward a mathematically attractive but operationally unacceptable result.

Measure: Can We Trust The Response?

EVOP searches for small effects.

That makes measurement quality critical.

Suppose a deliberate process change improves yield by 1%, but the analytical method itself varies by 2%. The improvement may be real, yet the measurement system cannot detect it reliably.

A Gage Repeatability and Reproducibility study helps determine how much of the observed variation comes from the measurement system rather than from the process.

Repeatability concerns the variation obtained when the same measurement system measures the same item under the same conditions. Reproducibility concerns differences across operators, instruments, laboratories or other measurement conditions.

Gage R&R is therefore more than a quality-department exercise. It determines whether an EVOP program can see the signal it is trying to create.

Unreliable analytical measurements may require more repetitions, a more sensitive method or an improved measurement system before valid conclusions can be drawn.

The sequence is simple:

First qualify the measurement. Then optimize the process.

Analyze: Is The Process Stable Enough To Learn From?

Before comparing EVOP operating points, engineers must understand the existing process variation.

Statistical Process Control helps distinguish common-cause variation from special-cause events.

Common-cause variation is the normal background noise of the process. Special causes include events such as:

  • an unusual raw-material batch
  • a sensor malfunction
  • an operator intervention
  • a utility failure
  • maintenance work
  • an upstream disturbance

If one EVOP condition happens to coincide with a special-cause event, the result may appear much better or worse than it truly is.

Control charts provide a structured way to detect such changes in process behaviour. The choice of chart depends on the type of data, the sampling method and the process under study.

EVOP requires that the distribution of uncontrolled error be reasonably comparable each time an experimental condition is repeated. Otherwise, engineers may attribute the effect of humidity, feed quality or equipment behaviour to the factors deliberately being changed.

This is why SPC and EVOP are natural partners.

SPC asks:

Is the process behaving consistently?

EVOP asks:

Within that behaviour, which operating condition performs better?

Control Charts Do Not Need To Reject Intentional Changes

A common concern is that EVOP changes could trigger control-chart alarms.

This is a matter of design.

A traditional control chart assumes that a process target should remain fixed. EVOP intentionally moves selected factors between planned operating points.

These planned changes should not be treated as unexplained special causes. They should be recorded as part of the experimental structure.

Control charts can instead be applied to:

  • responses within each EVOP condition
  • residuals after accounting for the planned operating changes
  • variables that are supposed to remain constant
  • safety, quality and equipment indicators
  • the process after a new operating point has been adopted

The control chart remains valuable. Its purpose shifts from preventing all movement to identifying movement that was not part of the plan.

Process Capability Defines The Room For Safe Learning

Statistical control and process capability are related but different.

A process can be stable while consistently producing outside specification. It can also be capable but poorly centred.

Capability indices such as Cp and Cpk summarize how the process distribution relates to specification limits. Cp reflects potential capability based on spread, while Cpk also reflects how well the process is centred.

For EVOP, process capability serves another important role: it helps determine the available experimental space.

Consider two processes:

  • Process A operates close to a quality limit with high variability.
  • Process B operates well inside the specifications with low variability.

The same size EVOP change may be unacceptable for Process A and entirely safe for Process B.

Capability analysis therefore helps engineers define:

  • how far factors may be changed
  • which responses require hard guardrails
  • when an experimental step should be stopped
  • how much safety margin must be retained
  • whether variation should first be reduced before optimization begins

EVOP does not ignore specifications. It learns inside them.

Improve: From A Project Experiment To Production Learning

The Improve phase of DMAIC often uses Design of Experiments to identify important factors and interactions.

DOE is extremely powerful in laboratory and pilot-scale work. It can explore wide factor ranges efficiently and build informative response-surface models.

At full scale, however, wide changes may produce unacceptable quantities of off-specification product. Repetitions can become expensive, and management may be unwilling to expose the plant to extreme experimental conditions.

EVOP adapts the principles of designed experimentation to these constraints.

Instead of a short campaign with large factor changes, EVOP uses:

  • narrow factor intervals
  • repeated cycles
  • normal production
  • gradual movement
  • close cooperation with operations

DOE is suited to breakthrough experimentation, while EVOP is evolutionary by design. DOE often operates offline or in dedicated trials. EVOP operates during production and must avoid deliberate off-specification operation.

Within DMAIC, the two approaches can therefore complement each other.

DOE may identify the important region or major factors. EVOP can fine-tune and maintain the optimum under real operating conditions.

Control: Do Not Freeze The Improvement

The Control phase is intended to ensure that the gains from a Six Sigma project are sustained.

Typically, this includes:

  • standard operating procedures
  • control plans
  • control charts
  • training
  • ownership
  • response plans

These measures are necessary. But they can produce an unintended consequence.

Once the improved settings are standardized, the process may again become static. The new target is treated as permanent.

In reality, the best operating point moves because raw materials, equipment and economic conditions change.

EVOP can turn the Control phase from passive preservation into active maintenance.

Instead of asking only:

Has the process moved away from the target?

The organization can also ask:

Is the current target still the best one?

This is the most important contribution EVOP can make to Six Sigma.

Dmaic With Evop

The integrated workflow can be summarized as follows.

Define

Select the economic response, operating factors, stakeholders, constraints and stopping rules.

Measure

Confirm that the measurement system can detect the expected improvement. Use Gage R&R and measurement-system analysis where appropriate.

Analyze

Use SPC, control charts, process knowledge and historical data to understand noise, disturbances and possible influential factors.

Improve

Apply DOE when wide experimentation is feasible. Apply EVOP when the process must continue producing and only small, safe changes are acceptable.

Control

Monitor the improved process—but continue small-scale learning as conditions evolve.

EVOP does not replace DMAIC.

It gives DMAIC a continuous engine.

A Practical Example

Consider a reactor producing an on-spec product with stable conversion.

A Six Sigma project may discover that raw-material variation and temperature control are the main sources of inconsistency. Measurement-system analysis confirms that conversion and impurity measurements are reliable. SPC improvements reduce variation. The capability index improves, and the process operates with a comfortable margin from the specification limit.

At this stage, the project could be closed.

But the new stability also creates an opportunity.

Because the process now has lower background noise and greater capability margin, EVOP can test small combinations of temperature and feed ratio. Repeated cycles may show that the same quality can be achieved with lower energy use—or that conversion can be increased without approaching the impurity limit.

Six Sigma did not merely improve the process.

It made the process easier to optimize.

Why Previous Evop Programs Sometimes Failed

Traditional EVOP requires long-term discipline.

Measurements must be taken reliably. Operating conditions must be followed correctly. Unplanned changes must be documented. Operators, laboratories, engineers and supervisors must cooperate.

Programs can fail when:

  • measurements are unreliable
  • special causes are ignored
  • experimental conditions are not maintained
  • results are interpreted too early
  • production teams are not involved
  • management stops the program before enough cycles are completed

These are not arguments against EVOP.

They are arguments for integrating EVOP into an established operational-excellence system.

Six Sigma provides governance, roles, statistical tools and management discipline. EVOP provides the ongoing mechanism for controlled learning.

The Role Of Pia

PIA—the Process Improvement Agent—can digitize this integration.

A PIA workflow can use:

  • Gage R&R results to assess measurement reliability
  • control charts to detect abnormal process conditions
  • process-capability limits to define the safe exploration range
  • EVOP designs to propose small operating changes
  • repeated observations to separate effects from noise
  • digital records to document every decision
  • automated guardrails to protect quality and equipment

This allows Six Sigma principles to remain active after the original project team has moved on.

PIA does not introduce uncontrolled variation. It introduces only the variation required to learn—and only within the limits defined by the process organization.

Stability Is The Foundation, Not The Destination

Six Sigma has taught industry to reduce variation, improve measurement and maintain capable processes.

EVOP adds one further principle:

A process should not only remain under control. It should continue learning.

Together, the methods answer two different questions.

Six Sigma asks:

Can we operate this process consistently?

EVOP asks:

Can we operate it better?

The combination is powerful because optimization without control is unreliable, while control without continued learning can preserve an outdated operating point.

Six Sigma stabilizes the process. EVOP helps it evolve.