Understanding and Controlling Process Variation

Let’s discuss sources of variation and how to measure, monitor, and control processes to minimize the differences from one part to the next.

Statistical process control (SPC) is a set of tools that provide insights into the changing nature of processes.

Product designs include the design engineer’s desired dimensions along with an allowance for variation. Engineers use tolerance analysis to determine a range of sizes that will enable assembly into the final product.

If every component complied with the design’s nominal values without variation, every assembled product would work as intended. Unfortunately, components, materials, and parts vary. They vary within or beyond the design specification because the variation occurs as a direct result of the processes that create the part and not the design intent.

Manufacturing processes create parts that are different from one another. Even a well-controlled and stable process has variation.

When the design calls for restricting the variation with a tolerance smaller than the natural range of variation, parts will still have dimensions outside the specifications. Manufacturers may include an inspection to sort out faulty parts, thereby incurring higher yield losses and higher costs.

Some ‘out-of-spec’ parts might only be found after assembly into a final product, compounding the cost of the error. Where is the root cause of the out-of-spec parts? What element of the process led to the faulty product?

When the design does not account for the normal variation of the manufacturing processes, it is inevitably the designer’s fault. Variation happens.

In this webinar, we will discuss the various sources of variation and how to measure, monitor, and control processes to minimize the differences from one part to the next.

Statistical process control (SPC) is a set of tools that provide insights into the changing nature of processes. Making variation visible allows the entire team to minimize excessive variation.

Even if we accept that no two parts are the same, it is not safe to assume that all parts are within the expected or defined range of values.

A process that is not monitored or controlled will generally create results with excessive variation.

This Accendo Reliability webinar originally broadcast on 14 June 2016.