Locked lesson.

Upgrade

About this lesson

The process is first assessed to determine if the failure modes identified in the Process FMEA have been prevented based upon the process design or management. The process control plan is then evaluated to determine if it is able to detect each of the failures listed in the Process FMEA. This prevention and detection capability is evaluated and scored.

Exercise files

Download this lesson’s related exercise files.

PFMEA Detection Rating.docx
60.8 KB PFMEA Detection Rating - Solution.docx
239.3 KB

Quick reference

PFMEA Detection Rating

Step 5 of the Process FMEA methodology is to conduct scoring of the failure modes.  The third score that is determined is the score for the detection of the failure mode or effect.

When to use

Once the severity and occurrence scores for the failure mode and effect have been determined, the methods for preventing the failure mode or detecting the failure mode or effect are identified and scored.

Instructions

Based upon the failure modes and effects, the FMEA analysis team determines the mechanisms used to prevent the failure or to detect either the failure mode or its effect as soon as possible.  Generally, if a failure mode is prevented, it does not also need a detection mechanism.  The Process FMEA has a column for both prevention and detection, which is a not-very-subtle hint that you should strive for prevention rather than detection.  Whichever method provides the lower score, that is the number that should be recorded in the adjacent detection score column.

As a general rule, the greater the level of automation, the lower score and the more reliance on in-process checks (checks done by operator while performing the step) versus post-processing checks (checks done by an inspector after the work is completed) the better.   

This is the point where Poka Yoke (mistake proofing) design features can be a big asset.  They are a form of prevention and can quickly drive the detection level to 1 or 2.  The primary sources of data for this score are the process control plan or quality plan and the product and process design documentation.  I have also found the process map to be very helpful in this area by reviewing the map for rework and repair points in the process. 

Ratings

The tables below show the scoring classification to be used for detection scores in a Process FMEA.

Example

Building on the example used in earlier lessons, a Process FMEA for a portion of a ball point pen assembly is shown below.  The colors in the detection scoring column are provided for illustration and training purposes.  Colors do not need to be shown in an actual FMEA.  In this example, the RPN calculation have been removed in order to focus on the Detection scores.  The example in the next lesson will include the RPN calculation.

Hints & tips

  • Generally, prevention scores will be lower than detection scores.
  • People make mistakes due to fatigue and distraction.  Machines don’t make those types of mistakes.  For that reason, automation provides a lower score than manual checks. 
  • In-process inspection and checks are preferred over post-processing because waiting until post-processing increases the likelihood that something might slip through.
  • If an item is checked both in-process by the operator and post-process by an inspector, use the lower score for the rating.
Login to download
  • 00:04 Hi, I'm Ray Sheen.
  • 00:06 Now it's time to discuss the third type of scoring on a Process FMEA,
  • 00:10 that is the detection scoring.
  • 00:12 This is still part of step five in the FMEA process.
  • 00:17 Detection ratings will be a bit different from the others.
  • 00:19 There will be two columns of actions or entries and we will use the score for
  • 00:23 the lower of the two.
  • 00:25 Detection is scoring the ability of the product or process, based upon the design
  • 00:29 and setup of each to detect a failure and take appropriate action to correct that
  • 00:33 failure, avoid that failure, or segregate the items for further action.
  • 00:38 Further, it needs to do this before the customer or
  • 00:40 operator is negatively impacted.
  • 00:43 Quite frankly, a better name would be undetectable,
  • 00:46 because a high score will mean that the failure is undetectable and
  • 00:50 a low score indicates that is it very detectable, or preventable.
  • 00:55 I mentioned that they were two columns for entries, one is for prevention and one for
  • 00:59 detection.
  • 01:00 The prevention control is an aspect to the process that prevents the failure.
  • 01:04 The detection control is an aspect of the process that
  • 01:07 finds a failure after it has occur, but before it goes on to the customer.
  • 01:12 You want to capture the detection method that is used for the failure mode.
  • 01:16 Keep in mind, the detection method may be occurring at another step.
  • 01:20 Such as the inspection step later in the process.
  • 01:23 One reason for capturing this is to be able to trace down your detection system,
  • 01:27 if failures start to occur in production.
  • 01:30 Well, let's consider our sources of data for the scoring.
  • 01:34 Look to the control plan for the process to identify all the inspection and
  • 01:38 test steps and what you are inspecting and testing.
  • 01:41 If you don't have a control plan, you can use the process control data.
  • 01:45 Things like yields and inspection records for an existing product.
  • 01:49 Another excellent tool is the process map that was developed at step three.
  • 01:53 Any loops, rework, repair, inspection and
  • 01:56 test steps will also indicate the information.
  • 01:59 Where failure detection data may be available.
  • 02:02 You can also use the design documentation and Design FMEA to identify Poka Yoke or
  • 02:07 mistake proofing decisions and to know what attributes are critical.
  • 02:12 Okay, now it's time to start scoring.
  • 02:14 I pulled from the standards, a description of the likelihood of detection,
  • 02:19 the detection opportunity and the criteria to use for process control.
  • 02:24 The level 10 is for failures that are absolutely impossible to detect,
  • 02:28 until the customer or user has already experienced the negative effect.
  • 02:32 There is no control in place.
  • 02:34 If a failure has never been analyzed before,
  • 02:36 it gets this rating, until it can be demonstrated that there is a control.
  • 02:40 Level 9 is very remote.
  • 02:42 The chance of detecting the failure is unlikely.
  • 02:45 It normally only comes about because of some random audit, checking the item and
  • 02:49 then finding the problem.
  • 02:51 The classic example of this is a problem that is only found in a destructive test.
  • 02:56 You find the problem, but you destroy the product while doing it.
  • 03:00 Level 8 is a remote likelihood of detection.
  • 03:03 It usually is relying on a manual inspection or
  • 03:06 test, based upon some visual, audible, or tactile signal.
  • 03:10 The problem with these is that the individual is distracted for a moment or
  • 03:13 their tension has wandered, a failure will slip right past them.
  • 03:17 And in particular, this inspection or test step is being done later in the process,
  • 03:21 not at the point where the cause of the failure occurred.
  • 03:25 Level 7 is a low likelihood of detection.
  • 03:28 The detection opportunity is again based upon a manual, visual, audible, or
  • 03:32 tactile test.
  • 03:33 But now it is occurring right at the work station, where the cause occurs.
  • 03:37 This gives the process the chance to immediately find and
  • 03:40 fix the problem, instead of the condition we had at level eight, where the process
  • 03:44 could have been making many failed items before it was discovered.
  • 03:48 Level 6 is a low likelihood of detection and
  • 03:50 it now relies on instruments and gauges, fixtures, or some other type of equipment
  • 03:55 being used manually, to conduct the inspection or test.
  • 03:59 The test is done post processing.
  • 04:01 Meaning it's not done at the workstation where the cause exists.
  • 04:04 The score is better than 8,
  • 04:06 because the equipment that is used should provide consistent data.
  • 04:10 Level 5 is moderate likelihood.
  • 04:12 The opportunity is to use the same manual test equipment or fixtures, but
  • 04:16 now using them inline at the workstation,
  • 04:19 instead of at some later point at an inspection station.
  • 04:23 Level 4 is a moderately high detection and
  • 04:26 we're now using automation equipment for our control mechanism.
  • 04:30 Automation doesn't get tired, or the equipment be misused.
  • 04:34 However, this is a level 4 because the automation is post-processing the item.
  • 04:39 To get to a level 3, the automation must be in line and
  • 04:43 processing the item in real time.
  • 04:46 The likelihood of detection now is high.
  • 04:49 And when the automation finds a problem with an item being processed,
  • 04:52 it locks that item out of further production, until the issue is resolved.
  • 04:57 Level two is very high, now we are finding things before they become a failure.
  • 05:02 The cause has started, meaning the error occurred, but
  • 05:05 it is found and fixed before it becomes a defective unit.
  • 05:09 And this detection is through an automated process.
  • 05:12 Level 1 is for
  • 05:13 the Poka Yoke mistake-proofed system, that prevents errors even from occurring.
  • 05:19 Any problem is almost certain to be detected and
  • 05:21 in fact, often an effective part can't even be made.
  • 05:25 So let's look at the FMEA form.
  • 05:27 In some cases, there's a prevention control, but for this example,
  • 05:31 in most cases, it is a detection control.
  • 05:34 There is little automation at this time in the process, so
  • 05:37 most things are done manually.
  • 05:39 In fact, the overwhelming method listed in detection methods,
  • 05:43 are visual inspection by the operator or the inspector.
  • 05:46 And since those are only a visual inspection,
  • 05:48 there's no equipment involved, they get a score of 7 or 8.
  • 05:52 Depending upon whether they are done in line or post processing.
  • 05:56 However, notice in line 14,
  • 05:58 that we have a visual inspection by the inspector, but a score of a 3.
  • 06:03 This is because that failure model of a defective decal
  • 06:06 is also addressed in the stock room, with the automated system in use there.
  • 06:11 So we can use the lower of the two scores.
  • 06:14 The other interesting point is that the two failure lines where the cause was no
  • 06:19 standard, were found to be Poka Yoke-ed for the occurrence.
  • 06:22 And therefore, they can't exist, so they get a score of 1.
  • 06:26 Well that wraps up the scoring for Process FMEA.
  • 06:30 As you can see, detection is very much related to the amount of automation that
  • 06:34 you have in your process.

Lesson notes are only available for subscribers.

PMI, PMP, CAPM and PMBOK are registered marks of the Project Management Institute, Inc.

Gift this course