More companies are leveraging high speed vision systems to inspect multiple quality characteristics on their products.
This testing directly on the line is a major advancement in inspection and measurement methods. Automation allows data to be collected quickly and efficiently and 100% inspection becomes feasible. A common question that arises is, “Does my 100% automatic inspection system replace the need for Statistical Process Control?”.
Background
More companies are leveraging high speed vision systems to inspect multiple quality characteristics on their products.
For example, in a high volume baking operation, a vision system can test for bun height, bun length, slice thickness, topping distribution, surface color, and more. This happens automatically on the line at high speeds. In bottling or other plastic manufacturing, a vision system may inspect multiple dimensions and surface properties.
This type of multi-characteristic testing occurring directly on the line is a major advancement in inspection and measurement methods as it replaces the need for an operator to do manual testing with hand held gages. Automation allows data to be collected quickly and efficiently and 100% inspection becomes feasible. Furthermore, the line does not need to be stopped.
A common question that arises is, “Does my 100% automatic inspection system replace the need for Statistical Process Control?”. The answer is an emphatic NO! This article will discuss how SPC complements and enhances the information gained from efficient inspection systems.
SPC vs. Inspection
Historically, quality was viewed as “conformance to specifications”. This “Product Control” approach to quality leads solely to a reliance on inspection systems for ensuring only “quality” products are delivered to customers. Controlling the product means that the product (incoming or outgoing) is inspected to ensure that all product conforms to the stated specifications. Although automated high speed inspection systems make the inspection process much more efficient, these systems complement SPC rather than replace the need for it.
A critical weakness of inspection systems is that they do not PREDICT that a problem may be on the horizon (as SPC does). Furthermore, an inspection system does not indicate when a process change has occurred, but such information should be highly valuable to manufacturing personnel, because a process change often leads to changes in performance of the product–even if the product still meets specifications.
Additionally, SPC can indicate when variation is getting worse (or better) in a process, and this information is extremely important in predicting product performance and customer satisfaction. Inspection systems do not evaluate variability.
The table below contrasts Inspection Systems with Statistical Process Control.
|
Inspection Systems |
Statistical Process Control |
|
Inspection implies that we are alerted when product characteristic is out-of-specification. By this time, the product is non-conforming and must be scrapped, reworked, or downgraded. Product control is reactive.
|
Control charts alert us when significant changes to key product or process characteristics occur. Note that we typically will react long before we produce non-conforming product and crises may be averted. |
|
Inspection does not distinguish between parts that are very close to target vs. parts that are nearly out of specification — all are treated as “good”. However, almost always our customers would prefer to receive parts close to target.
|
SPC and Process Capability assessments drive us to achieve stable (predictable) processes that have minimal variation. Results are generally consistent with customer desires (closeness to target) |
|
Inspection does not inherently monitor the evolution of the process over time. We are only alerted to the issue when non-conforming product is produced. As such, root causes of process changes leading to non-conforming product may be difficult to determine
|
Because we are alerted as soon as significant process changes occur, we can usually determine the root cause(s) of the process change, and act to prevent future occurrences. |
|
Inspection systems focus on key product (output) characteristics but do not prevent issues by controlling key input characteristics such as raw material and processing characteristics. |
SPC provides optimal benefits when we move upstream in the process. When the significant process variables that affect a key process output are being controlled, then the process output is predictable.
|
Inspection systems may provide assurance that customers receive conforming products. However, without tracking key process characteristics, issues will not be avoided using proactive techniques.
Making Use of SPC and Inspection Data
For key product characteristics that are being inspected, control charts may be constructed using the inspection data. The control charts will provide important insights regarding whether the process average or variation is changing significantly (e.g. trends, process shifts). Even if 100% of the product meets specification, the variability may cause unnecessary expense, assembly issues, or inconsistent product performance.
The selection of appropriate subgroup sizes is still an important consideration when designing a control chart, even when data on every part produced is available. For example, using large subgroup sizes (because the data is available) may result in overly sensitive control charts which detect statistically significant process changes which are not important from a practical perspective.
We still need to capture the expected within subgroup and between subgroup variation to set up proper control limits that allow us to distinguish common cause and special cause sources of variation.
Moving Process Control Upstream
As mentioned earlier, the real benefits of SPC are realized when key process characteristics that affect product quality are monitored and controlled. These characteristics are not the same features that are typically being inspected. Rather, they are predictive of the features that are being inspected and that our customers require. For example, these key characteristics may be production conditions or material properties–or features of supplied components.
By controlling the key characteristics that predict product quality and product performance, we may avoid costly issues that result from undetected process changes and that are only uncovered at the earliest via product inspection.
Utilizing Inspection Data to Complement SPC
The availability of inspection data may be used to confirm that our upstream process controls on key characteristics are appropriate. For example, if a glass temperature is being controlled to prevent distortion (feature inspected) during a molding process, we can determine whether significant changes in glass temperature do in fact lead to changes in glass distortion. Simple graphical tools like scatter plots or quantitative methods like regression may be used to confirm these relationships. The availability of process “output” data allows us to quickly determine whether we are controlling the right characteristics and if so, whether our charts are sensitive enough to detect issues before the final output is adversely affected.
Summary
Advances in automated inspection processes and equipment have provided significant benefits to manufacturers but the need for SPC has not been eliminated. SPC is a vital tool allowing progressive manufacturers the ability to detect process changes, reduce variation, and prevent costly issues. In addition to providing quality assurance, automatic inspection systems provide data that can improve the effectiveness of the SPC.
Steven Wachs, Principal Statistician
Integral Concepts, Inc.
Integral Concepts provides consulting services and training in the application of quantitative methods to understand, predict, and optimize product designs, manufacturing operations, and product reliability. www.integral-concepts.com