DPO Question
Six Sigma – iSixSigma › Forums › Old Forums › General › DPO Question
 This topic has 13 replies, 8 voices, and was last updated 17 years, 9 months ago by Anonymous.

AuthorPosts

February 5, 2004 at 8:05 pm #34472
Joseph RosenblattParticipant@JosephRosenblatt Include @JosephRosenblatt in your post and this person will
be notified via email.Dear Forum Members,
I am trying to learn Six Sigma. At present I am a complete novice or worse. If my question is unworthy of this forum then please forgive me.
My question is: How does it help to count Defects Per Opportunity instead of defects per unit? As an example, you have are sending 10 bills with ten opportunities per bill. If you have 10 defects on 1 bill or 1 defect on 10 bills the DPO is the same however the number of units without defect is not the same. Therefore the amount of effort needed to correct the defects is not the same. Am I looking at this incorrectly?
Another thing I do not understand is how to weight defects. I would thing that some defects are worse than others. Going back to my example whether the bill has the customers middle initial wrong or the amount being billed wrong the DPO is incremented by 1.
Thanks for any light you can shed.
Joseph Rosenblatt0February 5, 2004 at 8:18 pm #95122Joseph,
DPU does not take into consideration the opportunities. eg.
Defects=99 Units=7528 Opps=2 Total opps=units*opps=15056
DPU=defs/units=0.013
DPO=defs/total opps=0.006575
The DPO is more accurate if you have multiple opportunities. If Im counting missing bolts and there are 2 bolts that go into my part, I have 2 opportunities to miss a bolt.
You can then take your DPO, multiply it by 1,000,000 to get your DPMO(defective parts per million opportunities) which can the be translated into a sigm value.0February 5, 2004 at 8:33 pm #95123Hi Joseph
DPO Or DPMO(Decfects per Million Oppurtunities) is a better way of evaluating your process capability.DPU will not give you how many fields were defects in the form..it will only give you 1 single umit as Defective.
Its always a good idea to distinguish beetween Critical fields & non Critical fields in the form you are using….This will help you to understand & track those defects which are more important to your business.
Hope this clarifies..regards
0February 6, 2004 at 3:44 am #95133
Tim FolkertsMember@TimFolkerts Include @TimFolkerts in your post and this person will
be notified via email.
On the contrary, I think it is a very good question. It is easy to get caught up in six sigma – in the jargon, in the process, in the statistics. The challenge is to let the tools help you, not let the tools control you.
You have to decide what is important enought to measure and improve, and you have to decide what is an “opportunity”. If getting the total on the bill correct is important, then by all means count the DPMO for the total. After all, if all ten customers send back a bill because each one has a defect, that is a lot worse than if one customer sends back a bill because it has ten errors.One the other hand, if you are trying to see how often the clerical staff mistypes a figure, then you would be better off counting each number entered as an opportunity. That then gives a good way to estimate how many errors you might expect on a given bill. (Of course, that assumes your errors are random, which is often not the case.)
The same principle applies to severity of mistakes. Here you might consider FMEA (failure modes and effects analysis) to decide how important a problem is. Or estimate the “cost” of an error. A typo in a name on a bill doesn’t cost much. A wrong number costs the time of your acct. dept. to find the old bill, verify the error, fix it, mail a new copy, receive the payment late, send a salesman to smooth things out, … . Multiply the (defect rate) x (cost per defect) to determine the impact on the bottom line.
One size does not fit all. Not all defects were created equal. Man does not live by six sigma alone.
Tim0February 6, 2004 at 9:42 am #95141Well put Tim …
0February 12, 2004 at 1:22 pm #95360
chngagentParticipant@chngagent Include @chngagent in your post and this person will
be notified via email.i think you want to be careful when using dpu versus dpo. again you have to consider who your customer is. if you are reporting dpo do you think your customer cares how many opportunities you have for something to potentially go wrong?
dpu, dpo, dpmo all these are mathematical derivations of one another!0February 12, 2004 at 6:16 pm #95396
Mahendra ShahParticipant@MahendraShah Include @MahendraShah in your post and this person will
be notified via email.Hello Joseph,
For achieving the total Value in applications of Six Sigma TQM methodology you need to apply both DPO as well as DPU to measure performance. For optimization you visit each of the critical items [ activity or opportunity ] and review it from the following perspective of importance to Total [ Unit ]. This will provide a better accountability in terms of Form, Fit, and Functions to entire product or service and each one assigned weight in terms of total value.
Let us visit classical definitions Unit and Opportunity or function.
Unit is normally a Product or Service where there are multiple functions incorporated in each product or service offering.
Opportunity is exercise / performance of each of the subfunction of Unit.
Having created matrix of importance factors [weight matrix of Form, Fit and performance requirements for Functions] and value of same to overall performance of product or service [ Unit] will provide a model for performance optimization. For Unit a Classical tool called Pareto analysis is used to achieve TQM Six Sigma performance ].
To apply the Value Optimized Six Sigma TQM in areas of Systems / Products / Services / Operations and developments, I have found that taking a real world Product /Service and developing the matrix model and then addressing performance based solutions helps more efficiently in applying the Six Sigma TQM methodology.
In nutshell you need to address both DPO,opportunity in terms of subfunctions and DPU, Units performance of multiple functions.
am trying to learn Six Sigma. At present I am a complete novice or worse. If my question is unworthy of this forum then please forgive me.
My question is: How does it help to count Defects Per Opportunity instead of defects per unit? As an example, you have are sending 10 bills with ten opportunities per bill. If you have 10 defects on 1 bill or 1 defect on 10 bills the DPO is the same however the number of units without defect is not the same. Therefore the amount of effort needed to correct the defects is not the same. Am I looking at this incorrectly?
Another thing I do not understand is how to weight defects. I would thing that some defects are worse than others. Going back to my example whether the bill has the customers middle initial wrong or the amount being billed wrong the DPO is incremented by 1.
Thanks for any light you can shed.
Joseph Rosenblatt0February 13, 2004 at 3:57 am #95424
Jonathon L. AndellParticipant@JonathonL.Andell Include @JonathonL.Andell in your post and this person will
be notified via email.It’s a legitimate question.
If you can count “opportunities” per unit somewhat objectively and consistently, then DPO (or DPMO, whichever you prefer) compares the defects against the number of times the process is used.
If counting opportunities is challenging (as in many transactional processes), then DPU is an acceptable fallback position.
Either way, don’t get bogged down over the number, other than to compare beginning vs ending performance.0February 13, 2004 at 9:38 am #95437Joseph:
In the early 1980’s, some engineers at SGS published an article called ‘Spatial Yield Analysis.’ In this article, the engineers described a method for calculating random and systematic defect density from %defecive. When we tried this at General Instruments, we found that the random defect density portion in the facility was quite consistent, but that some similar products had depressed yields due to the presence of a systematic defect density. (The model assumes a Poisson distribution.)
Later, we applied these methods at Motorola and we were able to identify a number of product families that were similarly affected. In fact we improved the technique by plotting out ‘test bin’ categories on to our spatial maps. These tests were used so much that one night a Mac caught fire in the main office area!!
The DPO method cannot identify depressed yields within a product family, let alone within a facility, or even remotely from facility to facility. Yet nowadays this is seen as the basis of the Six Sigma zscore?
As someone mentioned previously, defects obey a Poisson distribution, and tolerance outliers from a normal distribution are nonconforming, not defective.
Good luck!
Andy0February 15, 2004 at 12:16 pm #95517Hi,
Just trying to elaborate a bit more. We really cant say that with only DPU at hand we cant calculate process Sigma(Z). Yes we can if and only if both of the following conditions hold true :
1. There are a considerable number of defect opportunities per unit.
2. Defect opportunities are same for all Units in the sample. All your units are of same category, as in your example. If you are only looking at Bills of same category having a constant number of Defect opportunities.
If both of them are true, you do not need to have DPO for Z calculation. HOW, here is how :
The defect distrubution may be approximated as a Posson distribution with a average defect probability which is Mean of the distribution which is nothing but = Lamda=DPU.
Now, P(X~ B)= exp(Lamda) * (Lamda) ** X / factorial(X).
Where, X is the defect distribution of the project Y.
and P(X) is the probability of X defects.
Hence, if we say, Yield= Probability of Zero defects=P(0)
P(0)= exp(Lamda) * (Lamda) ** 0 / Fact(0) = exp(Lamda)
=exp(DPU).
Now, from Z table, we find process sigma value.
But the application of the above case is rare, generally this happens in manufacturing process(assembly lines) like Pumps, Engines etc where Opportunities per unit is very very large.
Hope this helps to establish the importance of DPU.
Regards.
Pallab B.
MBB, TCS.
0February 15, 2004 at 12:16 pm #95518Hi,
Just trying to elaborate a bit more. We really cant say that with only DPU at hand we cant calculate process Sigma(Z). Yes we can if and only if both of the following conditions hold true :
1. There are a considerable number of defect opportunities per unit.
2. Defect opportunities are same for all Units in the sample. All your units are of same category, as in your example. If you are only looking at Bills of same category having a constant number of Defect opportunities.
If both of them are true, you do not need to have DPO for Z calculation. HOW, here is how :
The defect distrubution may be approximated as a Posson distribution with a average defect probability which is Mean of the distribution which is nothing but = Lamda=DPU.
Now, P(X~ B)= exp(Lamda) * (Lamda) ** X / factorial(X).
Where, X is the defect distribution of the project Y.
and P(X) is the probability of X defects.
Hence, if we say, Yield= Probability of Zero defects=P(0)
P(0)= exp(Lamda) * (Lamda) ** 0 / Fact(0) = exp(Lamda)
=exp(DPU).
Now, from Z table, we find process sigma value.
But the application of the above case is rare, generally this happens in manufacturing process(assembly lines) like Pumps, Engines etc where Opportunities per unit is very very large.
Hope this helps to establish the importance of DPU.
Regards.
Pallab B.
MBB, TCS.
0February 15, 2004 at 1:08 pm #95519Hi,
Have you personally confirmed this method in any simulations, or in practice? How to you handle systematic defects? How do you handle correlated opportunities?
Andy
0March 4, 2004 at 3:19 pm #96403Andy,
yes I have implemented this method already. Can you please elaborate your question a bit further ?
Regards.
Pallab.
0March 4, 2004 at 6:38 pm #96431The reason I asked the question this question is because a ‘similar’ model was used widely in the realy 1980s but had an additional term known as the ‘functional entitlement.’ These were not due to random ‘point’ defects, but due to line or area defects. One of the advantages of this method is that it was possible to identify a member of a product family with lower yield than other members of the family. (Normally one would expect all members of a product family to be subject to the same random point defect density.)
The software to perform this analysis was so successful that one night one of the Macs burst into flames! Once this model is tried and tested the notion of DPO becomes just another Dr. Harry et al howler.0 
AuthorPosts
The forum ‘General’ is closed to new topics and replies.