Accurate Measure of Field Performance

optomist1

A Sea of Statistics
Trusted Information Resource
#1
Good Day,

I have a bit of a dilemma regarding how to accurately measure the number of field defects or failures. Each fielded unit consists of an electrical harness, and on average each harness consists of 8 connectors, and within each connector are approximately 30 mechanically crimped wire to contacts or connections.

The harnesses are used to conduct 24/7 performance validation/verification tests; such that if one or fifty wire/contact crimped connection(s) fail(s) the test is terminated, the harness replaced and the test must be re-started from "0" time, very costly and more embarrassing.

Once shipped to the customer, and a contact fails, the harness is labeled defective and returned for repair of replacement.

Given that when one crimped contact fails the entire harness is classified as defective, leads me to believe that, I should merely evaluate the population of all harnesses as good or bad based on this.

Yet given the number of crimped connections in each harness or unit, it would appear to be more realistic to calculate the yield based on the total number of contacts in the larger population of all harnesses in given period of time. I only have attribute data for the period in time these units were fabricated.

Any suggestions from the Cove will be appreciated. Thank you.

Regards,
Marty
 
Elsmar Forum Sponsor
N

NumberCruncher

#2
Hi Optomist1

I will have to make an assumption that you are referring to the proportion of failures rather that simply the number. 23 failures is not too meaningful by itself. 23 failures in 23 items is disastrous. 23 failures in 10 000 items is rather better.

If you want to measure the proportion of field failures, then it's best to measure the proportion of field failures.

Calculating the raw proportion of failed contacts will give you a very low estimate of the failure rate.

Let's say you have sold 1000 units and 10 have failed in the field. That is 1% fail rate.

Alternatively, you have about 240 contacts in each harness so you have 240 000 contacts in the field. If you check the failed units, you will probably find only 1 faulty contact in each harness. That makes about 10 defects in 240 000 which is 0.004% fail rate.

But you actually have 1% fail rate for the units.

So which do you think is the more realistic figure to report to your management, 1% or 0.004%?


However, counting the number of faulty contacts per unit may be useful for a different reason. If you find only 1 (and very rarely 2) faults per harness your manufacturing process is probably in control and the faults are most likely caused by a general deficiency in the manufacturing process.

If you find that the faulty harnesses each have several defective contacts, it suggests that you have a "special cause".

Do you have different shifts? Do the faulty harnesses relate to a particular shift?

Do you buy your contacts in batches? Do the faulty harnesses all relate to a particular batch or supplier?

And so on.

Hope this helps.

NC

[Note to statisticians. Yes, I know the above calculation for faulty connections is not strictly correct. This is actually a binomial calculation so I ought to calculate the probability of getting 1 defect + probability of getting 2 defects+...+ probability of getting 30 defects. However it doesn't make any significant difference to the point that I am making. Assuming totally independent errors and a small failure rate, the most likely number of defects is 1 per unit.]
 

optomist1

A Sea of Statistics
Trusted Information Resource
#3
Hi NC,

Thanks for the response. This situation is part of my Black Belt Case Study and posses a dilemma at least for me.

I have considered using a p chart in Minitab, yet when assessing Field Performance in the context of this problem, 1 failed crimped connection is as good as 50, whether 1 or 50 failed crimped connections, the harness is classified as defective.

Given that all the related data is attribute/binary, I am left with few choices

Of course from an internal process measurement standpoint, assessing the crimp/contact operation in terms of the total number of contacts makes sense. This is one of the suggested improvements.

I guess it comes down to the question(s) or one's perspective is:

If assessing strictly field failures then failed units is the appropriate measure.

If assessing the internal crimping operation, then measuring the number of defective to accepted crimped connections makes sense.

Thus far I have stratified the field failure data, performed the fish bone C&E and the results are pretty clear.

Regarding special causes...this is in fact the case. The one operator who assembled the particular contacts/harnesses was new to this product, secured the wrong crimp tool. So there are several areas for improvement, labeling, calibration and control of crimp tools, instituting wire/contact pull test, work instructions revisions etc.

Once the suggested improvements are implemented several verification runs will be made and analyzed

Your input is clear and helpful...thank you, especially the binomial note; again from the end users perspective, whether one, two three or fifty failures, it is still a failed harness and a costly failure in terms of test time, etc.

Funny thing about statistics, sometimes the hardest part is the selection of how one should present or structure the data....the actual analysis, although important is rather straightforward.

Regards,
Marty
 
Thread starter Similar threads Forum Replies Date
I Can we use pin gauges to measure an accurate Cpk ? Capability, Accuracy and Stability - Processes, Machines, etc. 9
Ed Panek Ideas to keep quotes for consulting service accurate on not inflated Consultants and Consulting 11
N Accurate calculations of First Time Quality / First Time Yield Misc. Quality Assurance and Business Systems Related Topics 3
Q How can you justify using a more accurate Pin Gage class? General Measurement Device and Calibration Topics 16
Moumen H Controls to ensure that Quality Plans are accurate APQP and PPAP 2
R Accurate Quarterly Inventory Turn Numbers Quality Manager and Management Related Issues 3
C Accurate Measurement Method for Flash and Mismatch @0.002" Inspection, Prints (Drawings), Testing, Sampling and Related Topics 8
marmotte Definition Monitoring - Accurate definition of Monitoring - IEC 60601 Definitions, Acronyms, Abbreviations and Interpretations Listed Alphabetically 1
T Accurate Method of Internal Radius Check Inspection, Prints (Drawings), Testing, Sampling and Related Topics 18
J Printers and Plotters that do ACCURATE Mylar or Plastic Film Overlays Design and Development of Products and Processes 6
SteelMaiden What does 99.9% success look like if processes were 99.9% accurate? Training - Internal, External, Online and Distance Learning 19
Marc Keep Your Resume Accurate - Lying on Resume cost CEO his job and $25.1 million Career and Occupation Discussions 3
Marc Definition Accuracy (Accurate) - Definition Definitions, Acronyms, Abbreviations and Interpretations Listed Alphabetically 4
R 100% Inspection - 80% Accurate - Quote from Juran (?) Inspection, Prints (Drawings), Testing, Sampling and Related Topics 94
D Measuring ID's: Which is the most accurate? CMM / Sunnen Gage / Plug Gage? General Measurement Device and Calibration Topics 3
C First Article Inspections - 'how to measure' General Measurement Device and Calibration Topics 11
lanley liao Question regarding the calibration of monitoring and measure equipment. Oil and Gas Industry Standards and Regulations 0
J Should a Class 1 medical device with an option to measure body weight be considered Class 1m? EU Medical Device Regulations 0
M Gage R&R and right way to measure Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 16
P How to measure DPPM in Chemical batch manufacturing Quality Tools, Improvement and Analysis 1
silentmonkey How to measure severity if my product is designed for emergency use and failure would result in death? ISO 14971 - Medical Device Risk Management 9
A Looking for a third party to calibrate our measure equipment - South Carolina General Measurement Device and Calibration Topics 6
J Validity / outcomes measure for custom made medical device ISO 13485:2016 - Medical Device Quality Management Systems 2
D Performance specification as a Risk Control Measure, EN 14971 ISO 14971 - Medical Device Risk Management 7
J KPIs or Metrics to Measure a New Complaint Handling Process 21 CFR Part 820 - US FDA Quality System Regulations (QSR) 9
D Torque fractional unit of measure; cN-m or N-cm? General Measurement Device and Calibration Topics 2
qualprod ISO 9001 5.1.1 - How to measure System effectiveness ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 4
J How to measure a Critical Characteristic that is a Basic Dimension Misc. Quality Assurance and Business Systems Related Topics 6
qualprod Best criteria to measure Corrective Action effectiveness - Poor Maintenance ISO 9000, ISO 9001, and ISO 9004 Quality Management Systems Standards 10
1 How to measure straightness on a HEX Long Bar Manufacturing and Related Processes 7
Ron Rompen Measurement Method Challenge - Measure feature #91 Inspection, Prints (Drawings), Testing, Sampling and Related Topics 9
G Measure 3 times, for calibration report? Customer Owned Reference Standard General Measurement Device and Calibration Topics 4
M Informational The ANSM decides, as a precautionary measure, to withdraw from the market macrotextured breast implants and breast implants with polyurethane-coated s Medical Device and FDA Regulations and Standards News 0
M Informational Health Canada advises Allergan of its intent to suspend its licences for Biocell breast implants as a precautionary measure Medical Device and FDA Regulations and Standards News 0
U SI 13 - Not Every Process Requires an Efficiency Measure IATF 16949 - Automotive Quality Systems Standard 1
P NPR Metric - What if Customer has relaxed measure? TL 9000 Telecommunications Standard and QuEST 1
T Tape Measure, the right device for 2 or 3 decimals accuracy on prints General Measurement Device and Calibration Topics 8
M How do we document and measure a service we do not provide? IATF 16949 - Automotive Quality Systems Standard 1
M For a Type 1 Gage R&R I want to use 1 unit and measure it 30 times Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 5
M How to Measure Paper considering Variation and Humidity General Measurement Device and Calibration Topics 10
rob73 Advice/recommendations for equipment to measure a soft pvc tube - Optical Measuring? General Measurement Device and Calibration Topics 4
R Height gauge to measure short hard to access heights General Measurement Device and Calibration Topics 3
D Proper Tool to Measure Wall Thickness (ISO 3611 - Micrometers) Manufacturing and Related Processes 7
S Quality Measure for Engineering Services Misc. Quality Assurance and Business Systems Related Topics 1
A What are methods and tools to measure Workload analysis ? Human Factors and Ergonomics in Engineering 1
M How do I measure variation in my process ? Capability, Accuracy and Stability - Processes, Machines, etc. 3
J How to measure process quality? Is DPPM the correct metric? Capability, Accuracy and Stability - Processes, Machines, etc. 5
M Micrometer Anvil Placement to measure the diameter of Steel Wires Measurement Uncertainty (MU) 6
S Where to Get SPC Training to Measure our Processes? Training - Internal, External, Online and Distance Learning 12
M A question on Measurement System Analysis - Stability to measure got Gage R&R (GR&R) and MSA (Measurement Systems Analysis) 2

Similar threads

Top Bottom