Process Parameters and Product Characteristics

I recently received a note from Pankaj Sharma asking me to explain the difference between the function of the process and the function of the process item—or workpiece, if you will. As with most questions about quality management, the answer is straightforward but not necessarily simple. And Failure Modes & Effects Analysis methods can be used to explain this.

The FMEA Framework

On a PFMEA worksheet of the type found in the AIAG-VDA FMEA guideline, the second group of columns, those that come after the Structure Analysis section, are called the “Function Analysis” section. That section includes three specific columns:

For each process step, there will be a “process function” which can be derived using a Characteristics Matrix. This is an active verb-measurable noun statement of the purpose of the process step. In addition, there will be one or more key process parameters for that step.

A process parameter is a measurable element of inputs or transformational activities in a process that affects process outputs. There are many types of process parameters, depending on the type of process being studied. Some examples include pressure, temperature, speeds, feeds, voltage, time, chemical concentrations, and torque (among dozens of others).

Because the process step is transforming one or more inputs into an output, there are product characteristics that are important in that output. Product characteristics are the measurable?features of a system, subsystem, part, or component. Product characteristics are derived from product functions in DFMEA using hierarchy diagrams, partial Parameter diagrams and Interface Matrices.

Effective Process Control

Both process parameters and product characteristics play central roles in the dynamics of a process, so both can be used as part of a comprehensive approach to process control. Most process control plans are aimed at the product characteristics of the output—you measure the part or assembly in some way after the process step is completed—and the process parameters aren’t measured.

That’s the traditional way to control a process and to control quality. It’s the way humanity has been doing it since the first stone tools were chipped from flint and obsidian.

However, it’s often (even usually) more powerful to use process parameters for control. Why is that? The simple reason is that process parameters determine what the product characteristics might be after the process step is completed.

Let’s consider an example—the through-hardening heat treatment of a medium carbon steel part. To start, we have to assume that the steel has the correct composition. After all, the heat treat line has no capability to fix that kind of error. So, what is the process?

The steel is heated to a temperature of ?800-845oC (1475-1550oF). It also has to be held at that temperature so that the temperature is even in the entire part. The atmosphere in the furnace also has to be controlled so that the carbon in the steel doesn’t react with oxygen—which would decarburize the part and make it soft at the surface.

Once the part is at the correct temperature in the correct atmosphere, the part can then be quenched in room temperature oil. The part should then have a minimum hardness of 55 Rockwell “C.” Of course, the part will also have to be tempered to have the right combination of hardness and ductility, but that’s another process step.

To control this first process, you could just measure the hardness of the part after quenching. That would, of course, be a detection control. However, if the hardness wasn’t correct, the part would have to be reworked. Or, if it were decarburized, it would have to be scrapped.

As an alternative, if the temperature, cycle time, atmosphere in the furnace and quenching conditions are correct, it is a virtual certainty that the hardness would reach the minimum temperature. Measuring and regulating those factors would constitute a set of prevention controls. The metallurgy of this situation is so well understood that the proper outcome is almost as certain as the fact that a ball tossed in the air will fall to earth; the basic laws of nature and the principles of metallurgy will guarantee both of these outcomes.

You might then ask why would you ever check the hardness of a part if you are so sure of the process control? There are two reasons why you would at least like to do this on a sampling basis:

1. Something may have gone wrong with the measurement tools used to ensure that the process parameters have remained within expected limits so the hardness measurement will then be the control that matters—and will also guide the corrective action to repair the process parameter controls
2. There may be some contractual or regulatory reason that the hardness needs to be recorded

There’s also a third reason that would not be true in this example but is possible in other processes. If the relationship between process parameters and product characteristics is not well understood, then reliance on detection control, or measurement of the process output’s product characteristics may be necessary.

Nevertheless, you should not lose sight of the simple fact that prevention controls, controls that are focused on process parameters, have great economic value and are often more effective than detection controls.

Of course, deciding what controls to use—and which to omit—is another question. To learn more about that, you can read about that here.