where expertise comes together - since 1996

Web's Largest Portal of Nondestructive Testing (NDT)
Open Access Database (Conference Proceedings, Articles, News), Exhibition, Forum, Network

All Forum Boards
Technical Discussions >
Violations of Unit Measure
Career Discussions
Job Offers
Job Seeks
Classified Ads
About NDT.net
Articles & News

2207 views
00:36 Feb-23-1999

Terry Oldberg

Engineering, Mechanical Electrical Nuclear Software
Consultant,
USA,
Joined Oct 1999
42
Violations of Unit Measure

In a 1995 paper*, a colleague and I reported an error in NDT's foundation. In brief, all of its various flaw detection methods violate the Unit Measure axiom of probability theory in relation to theories of these methods' reliability. As all of NDT's theories of this reliability are stated in terms of probabilities, it follows that all of NDT's theories of this reliability are wrong.

A hole having been identified in their field's foundation, one might have expected the NDT community to get right to work to repair it. However, so far as I know, nothing has been done four years later. Any ideas on overcoming this inertia?

* "Erratic Measure," in NDE for the Energy Industry 1995, pp.1-6. The American Society of Mechanical Engineers, New York, NY, 1995.




 
01:27 Feb-23-1999

Rolf Diederichs

Director, Editor, Publisher, Internet, PHP MySQL
NDT.net,
Germany,
Joined Nov 1998
602
Re: Violations of Unit Measure : In a 1995 paper*, a colleague and I reported an error in NDT's foundation. In brief, all of its various flaw detection methods violate the Unit Measure axiom of probability theory in relation to theories of these methods' reliability. As all of NDT's theories of this reliability are stated in terms of probabilities, it follows that all of NDT's theories of this reliability are wrong.

: A hole having been identified in their field's foundation, one might have expected the NDT community to get right to work to repair it. However, so far as I know, nothing has been done four years later. Any ideas on overcoming this inertia?

: * "Erratic Measure," in NDE for the Energy Industry 1995, pp.1-6. The American Society of Mechanical Engineers, New York, NY, 1995.
-------------------------------------

Terry,

thanks for this very interesting topic.
You may found already NDTnet articles of this subject,
could you mention some of the papers with wrong theory?
I would like to point out some resources below:

ECNDT'98 Session: Reliability and Validation
http://www.ndt.net/abstract/ecndt98/reliabil.htm

Proceedings European-American Workshop
Determination of Reliability and Validation Methods on NDE
June 18 - 20, 1997 Berlin, Germany
http://www.ndt.net/publicat/bibliog/euam97.htm

Another big part was published in July '98 'http://www.ndt.net/v03n07.htm'
"The Reliability of NDT techniques and procedure of validation"
(in conjunction with the train accident)

Christina Nockeman et. al.
Strategies of Validation in General Testing compared to Experience in NDE
http://www.ndt.net/article/0798/nockem/nockem.htm

1999 SEP 21-24 Boulder,
2nd European-American Workshop Determination of Reliability
and Validation Methods of NDE. Contact: Christina Nockemann BAM.
I don't think that Christina Nockemann (BAM RT Dep. Berlin)
post to this forum but you can contact her at: Christina.Nockemann@bam.de.

You mentioned the paper * 'Erratic Measure,' in NDE for the Energy
Industry 1995, pp.1-6. The American Society of Mechanical Engineers,
New York, NY, 1995.

Could you organize a copyright for NDTnet? In that case we all could
read it easily and continuing a great discussion, however some experts
should join in.

Rolf Diederichs



 
01:18 Feb-23-1999

Terry Oldberg

Engineering, Mechanical Electrical Nuclear Software
Consultant,
USA,
Joined Oct 1999
42
Re: Violations of Unit Measure : : In a 1995 paper*, a colleague and I reported an error in NDT's foundation. In brief, all of its various flaw detection methods violate the Unit Measure axiom of probability theory in relation to theories of these methods' reliability. As all of NDT's theories of this reliability are stated in terms of probabilities, it follows that all of NDT's theories of this reliability are wrong.

: : A hole having been identified in their field's foundation, one might have expected the NDT community to get right to work to repair it. However, so far as I know, nothing has been done four years later. Any ideas on overcoming this inertia?

: : * "Erratic Measure," in NDE for the Energy Industry 1995, pp.1-6. The American Society of Mechanical Engineers, New York, NY, 1995.
: -------------------------------------

: Terry,

: thanks for this very interesting topic.
: You may found already NDTnet articles of this subject,
: could you mention some of the papers with wrong theory?
: I would like to point out some resources below:

: ECNDT'98 Session: Reliability and Validation
: http://www.ndt.net/abstract/ecndt98/reliabil.htm

: Proceedings European-American Workshop
: Determination of Reliability and Validation Methods on NDE
: June 18 - 20, 1997 Berlin, Germany
: http://www.ndt.net/publicat/bibliog/euam97.htm

: Another big part was published in July '98 'http://www.ndt.net/v03n07.htm'
: "The Reliability of NDT techniques and procedure of validation"
: (in conjunction with the train accident)

: Christina Nockeman et. al.
: Strategies of Validation in General Testing compared to Experience in NDE
: http://www.ndt.net/article/0798/nockem/nockem.htm

: 1999 SEP 21-24 Boulder,
: 2nd European-American Workshop Determination of Reliability
: and Validation Methods of NDE. Contact: Christina Nockemann BAM.
: I don't think that Christina Nockemann (BAM RT Dep. Berlin)
: post to this forum but you can contact her at: Christina.Nockemann@bam.de.

: You mentioned the paper * 'Erratic Measure,' in NDE for the Energy
: Industry 1995, pp.1-6. The American Society of Mechanical Engineers,
: New York, NY, 1995.

: Could you organize a copyright for NDTnet? In that case we all could
: read it easily and continuing a great discussion, however some experts
: should join in.

: Rolf Diederichs

I provide the following reply to Rolf Diederichs' comments:

"Erratic Measure" is presently available in print form only. For convenience, I shall provide a synopsis of the portion of the paper which deals with violations of Unit Measure.

What one means by a "concrete object" will be critical to the discussion. Lazar Mayants provides a definition in his book The Enigma of Probability and Physics, which I shall summarize. In brief, a concrete object has a definite value for each of its properties and differs in the value of at least one property from every other member of the set. Each concrete object is an aggregation of concrete objects with the result that a set of concrete objects is a partition of the union of its elements. For example, a recently developed non-destructive test for cervical cancer defines a spacewise partition of human cervixes. The tissue within each element of this partition is an example of a concrete object and is itself composed of concrete objects, such as cells.

In assessments of the reliability of tests, 4 types of event play a leading role. The four types are true positive, false positive, true negative and false negative. An event of one of the 4 types happens when a test assigns a concrete object to a state which is positive or negative and when a definitive test, the test's "golden rule," is available to determine whether the assignment has been true or false. For example, the previously referenced test for cervical cancer assigns each element of its partition of human cervixes to positive or negative for cancer and the assignment can be determined to be true or false via a biopsy. A written procedure for the biopsy provides the test's golden rule.

Probability is defined as a measure of an event whose value is 1 when the event being measured is certain to occur. (For an exposition of the properties of a measure see, for example, Paul Halmos's book Measure Theory.) This property of probability is termed Unit Measure.

When an event has occurred, its probability is 1. This implies that there is a one-to-one relationship between the set of true positive, false positive, true negative and false negative events which have been generated by a test and the related set of concrete objects. If this relationship is, in fact, one-to-one, Unit Measure is empirically preserved. Otherwise it is empirically violated.

The previously referenced test for cervical cancer preserves Unit Measure. As I will demonstrate, NDT's tests for the detection of flaws violate it.

NDT's tests usually fail to define the associated golden rule. In doing so, they commit the fundamental error of failing to define the quantity which they are measuring. In order to conduct a study of the reliability of NDT, the researcher must fill this gap by providing an ad hoc rule. In doing so, the researcher supplies the missing definition of what quantity the test is measuring.

In the following discussion, "true" will mean that a positive or negative event is true by an ad hoc or golden rule. "False" will have the analogous meaning.

In NDT"s tests for the detection of flaws and the vernacular of NDT, the generation of an indication is synonymous with a positive event and a true positive event relates to a concrete object which is a flaw. It follows directly that NDT's flaw detection tests violate Unit Measure, for the existence of true positive events implies the existence of false positive events but no concrete objects relate to the latter. I will reference this type of violation as a "Type A" violation.

The previously referenced test for cervical cancer assigns each element of the partition of human cervixes to a positive or a negative event. However, the identity of NDT's negative events is murky.

According to NDT's vernacular, the generation of an indication is synonymous with the event of a positive but the identity of the negative events is unclear in NDT's procedures or vernacular. However, in estimating the Probability of Detection, NDT's researchers have established the convention that exactly 1 false negative event corresponds to each flaw which is unrelated to true positive events. With the establishment of this convention, additional violations of Unit Measure emerge.

One of these follows from the implication that the event of a negative occurs when an indication is not generated. There must be true negative events but they are not related to concrete objects. I shall reference this as a Type B violation.

At least two, additional types of violation can arise from an ad hoc rule, the distribution of indications and the distribution of flaws. One occurs when the rule relates more than one true positive event to a single flaw. I shall reference this as a "Type C" violation. The rule which counts an indication as a true positive if it is within a centimeter of a flaw creates a Type C violation if more than one indication is within an centimeter of this flaw.

Another occurs when the rule relates more than one flaw to a single, true positive event. I shall reference this as "Type D" violation. The rule which counts an indication as a true positive event if it is within a centimeter of a flaw creates a Type D violation if more than one flaw is within an centimeter of this indication.

I have demonstrated that 1) violations of Unit Measure of Type A are inherent in NDT's methods of flaw detection and vernacular 2) violations of Type B are inherent in NDT's methods of flaw detection and the vernacular of NDT's researchers 3) violations of Types C and D may occur. A study exhibiting violations of Types A, B and D is reported by E.R. Bradley, et al in "Steam Generator Group Project. Task 13 Final Report. Nondestructive Examination Validation," NUREG/CR-5185, 1988. U.S. Nuclear Regulatory Commission, Washington, DC.

Among the tests violating Unit Measure are a number of those which are used to inspect the nuclear reactor pressure boundary. The author of these tests, the American Society of Mechanical Engineers has thus far refused to address the issue.

Terry Oldberg





 
02:42 Feb-23-1999

Robert A. Day

Engineering
Milky Way Jewels,
USA,
Joined Nov 1998
40
Re: Violations of Unit Measure I read your paper at the time. I was unable to understand it so I gave copies to statisticians I knew and they were also unable to understand it. You might want to make your exposition clearer?

"A hole having been identified in their field's foundation," requires more than a conference proceedings. I would recommend republishing in the ASNT research journal which will get substantially broader interest than a conference proceedings.

You might also look at Ames Center for NDE , John Hopkins Center for NDE , the Electric Power Research Institute , and the NRC for possible interest in this area as they either have programs or interest in the reliability of NDE.

Keep in mind the burden of proof is always on he who wishes to over throw the established order.

Regards,
Robert (Rocky) A. Day
Second Sound
Ultrasonic Transducers
904 Cortland Avenue
San Francisco, CA 94110 - 5633
(415) 641-4947
Fax: (415) 641-5502

: In a 1995 paper*, a colleague and I reported an error in NDT's foundation. In brief, all of its various flaw detection methods violate the Unit Measure axiom of probability theory in relation to theories of these methods' reliability. As all of NDT's theories of this reliability are stated in terms of probabilities, it follows that all of NDT's theories of this reliability are wrong.

: A hole having been identified in their field's foundation, one might have expected the NDT community to get right to work to repair it. However, so far as I know, nothing has been done four years later. Any ideas on overcoming this inertia?

: * "Erratic Measure," in NDE for the Energy Industry 1995, pp.1-6. The American Society of Mechanical Engineers, New York, NY, 1995.

:




 
03:26 Feb-23-1999

Terry Oldberg

Engineering, Mechanical Electrical Nuclear Software
Consultant,
USA,
Joined Oct 1999
42
Re: Violations of Unit Measure : I read your paper at the time. I was unable to understand it so I gave copies to statisticians I knew and they were also unable to understand it. You might want to make your exposition clearer?

: "A hole having been identified in their field's foundation," requires more than a conference proceedings. I would recommend republishing in the ASNT research journal which will get substantially broader interest than a conference proceedings.

: You might also look at Ames Center for NDE , John Hopkins Center for NDE , the Electric Power Research Institute , and the NRC for possible interest in this area as they either have programs or interest in the reliability of NDE.

: Keep in mind the burden of proof is always on he who wishes to over throw the established order.

: Regards,
: Robert (Rocky) A. Day
: Second Sound
: Ultrasonic Transducers
: 904 Cortland Avenue
: San Francisco, CA 94110 - 5633
: (415) 641-4947
: Fax: (415) 641-5502
:

: : In a 1995 paper*, a colleague and I reported an error in NDT's foundation. In brief, all of its various flaw detection methods violate the Unit Measure axiom of probability theory in relation to theories of these methods' reliability. As all of NDT's theories of this reliability are stated in terms of probabilities, it follows that all of NDT's theories of this reliability are wrong.

: : A hole having been identified in their field's foundation, one might have expected the NDT community to get right to work to repair it. However, so far as I know, nothing has been done four years later. Any ideas on overcoming this inertia?

: : * "Erratic Measure," in NDE for the Energy Industry 1995, pp.1-6. The American Society of Mechanical Engineers, New York, NY, 1995.

: :
Robert:

I'm happy to provide clarification and join in discussion. However, I believe you have misrepresented the scientific method by stating that the burden of proof is on the author of a peer reviewed article to reprove his thesis. Actually, the rule is that I am not allowed to republish the same material and that the thesis stands until it is refuted. It would not be necessary for you or anyone else to agree with the way in which I have supported my thesis in order to refute it.

The issue is whether Unit Measure is preserved by NDT. The thesis of the paper is that this axiom of probability theory is violated in a large class of NDT's tests. This thesis can be refuted by publication of a proof that Unit Measure is preserved. Such a publication has not, to my knowledge, been forthcoming.

While the statisticians you consulted did not understand the paper, my co-author is a statistician and an academic statistician read and understood it as a referee. I can't address your statisticians' difficulties without knowing what they are and neither they nor you have ever contacted me on them. Perhaps you would be willing to contribute their comments to the forum.

Terry




 
01:14 Feb-24-1999

Robert A. Day

Engineering
Milky Way Jewels,
USA,
Joined Nov 1998
40
Re: Violations of Unit Measure Terry -

I'm the wrong guy to convince and this is the wrong forum for this debate. You do have a burden to convince and I stand by my characterization that the new theory's advocates carry the burden of proof. You are correct that the same paper cannot be submitted to a peer review journal after publication in the ASME proceedings. You would have to rewrite the paper. My suggestion is that you do so to make it more understandable to people in field and to get more attention to your theory. Publication in a peer review journal is a necessary precondition to starting a discussion on violations on unit measure in any case.

If you contact people working on NDE reliability this also will give you an opportunity to convince them, or be convinced in turn. I cannot provide you with a critique or an affirmation since I'm not working in this field.

Rocky

: : I read your paper at the time. I was unable to understand it so I gave copies to statisticians I knew and they were also unable to understand it. You might want to make your exposition clearer?

: : "A hole having been identified in their field's foundation," requires more than a conference proceedings. I would recommend republishing in the ASNT research journal which will get substantially broader interest than a conference proceedings.

: : You might also look at Ames Center for NDE , John Hopkins Center for NDE , the Electric Power Research Institute , and the NRC for possible interest in this area as they either have programs or interest in the reliability of NDE.

: : Keep in mind the burden of proof is always on he who wishes to over throw the established order.

: : Regards,
: : Robert (Rocky) A. Day
: : Second Sound
: : Ultrasonic Transducers
: : 904 Cortland Avenue
: : San Francisco, CA 94110 - 5633
: : (415) 641-4947
: : Fax: (415) 641-5502
: :

: : : In a 1995 paper*, a colleague and I reported an error in NDT's foundation. In brief, all of its various flaw detection methods violate the Unit Measure axiom of probability theory in relation to theories of these methods' reliability. As all of NDT's theories of this reliability are stated in terms of probabilities, it follows that all of NDT's theories of this reliability are wrong.

: : : A hole having been identified in their field's foundation, one might have expected the NDT community to get right to work to repair it. However, so far as I know, nothing has been done four years later. Any ideas on overcoming this inertia?

: : : * "Erratic Measure," in NDE for the Energy Industry 1995, pp.1-6. The American Society of Mechanical Engineers, New York, NY, 1995.

: : :
: Robert:

: I'm happy to provide clarification and join in discussion. However, I believe you have misrepresented the scientific method by stating that the burden of proof is on the author of a peer reviewed article to reprove his thesis. Actually, the rule is that I am not allowed to republish the same material and that the thesis stands until it is refuted. It would not be necessary for you or anyone else to agree with the way in which I have supported my thesis in order to refute it.

: The issue is whether Unit Measure is preserved by NDT. The thesis of the paper is that this axiom of probability theory is violated in a large class of NDT's tests. This thesis can be refuted by publication of a proof that Unit Measure is preserved. Such a publication has not, to my knowledge, been forthcoming.

: While the statisticians you consulted did not understand the paper, my co-author is a statistician and an academic statistician read and understood it as a referee. I can't address your statisticians' difficulties without knowing what they are and neither they nor you have ever contacted me on them. Perhaps you would be willing to contribute their comments to the forum.

: Terry

:




 
07:10 Feb-25-1999

Terry Oldberg

Engineering, Mechanical Electrical Nuclear Software
Consultant,
USA,
Joined Oct 1999
42
Re: Violations of Unit Measure : Terry -

: I'm the wrong guy to convince and this is the wrong forum for this debate. You do have a burden to convince and I stand by my characterization that the new theory's advocates carry the burden of proof. You are correct that the same paper cannot be submitted to a peer review journal after publication in the ASME proceedings. You would have to rewrite the paper. My suggestion is that you do so to make it more understandable to people in field and to get more attention to your theory. Publication in a peer review journal is a necessary precondition to starting a discussion on violations on unit measure in any case.

: If you contact people working on NDE reliability this also will give you an opportunity to convince them, or be convinced in turn. I cannot provide you with a critique or an affirmation since I'm not working in this field.

: Rocky

: : : I read your paper at the time. I was unable to understand it so I gave copies to statisticians I knew and they were also unable to understand it. You might want to make your exposition clearer?

: : : "A hole having been identified in their field's foundation," requires more than a conference proceedings. I would recommend republishing in the ASNT research journal which will get substantially broader interest than a conference proceedings.

: : : You might also look at Ames Center for NDE , John Hopkins Center for NDE , the Electric Power Research Institute , and the NRC for possible interest in this area as they either have programs or interest in the reliability of NDE.

: : : Keep in mind the burden of proof is always on he who wishes to over throw the established order.

: : : Regards,
: : : Robert (Rocky) A. Day
: : : Second Sound
: : : Ultrasonic Transducers
: : : 904 Cortland Avenue
: : : San Francisco, CA 94110 - 5633
: : : (415) 641-4947
: : : Fax: (415) 641-5502
: : :

: : : : In a 1995 paper*, a colleague and I reported an error in NDT's foundation. In brief, all of its various flaw detection methods violate the Unit Measure axiom of probability theory in relation to theories of these methods' reliability. As all of NDT's theories of this reliability are stated in terms of probabilities, it follows that all of NDT's theories of this reliability are wrong.

: : : : A hole having been identified in their field's foundation, one might have expected the NDT community to get right to work to repair it. However, so far as I know, nothing has been done four years later. Any ideas on overcoming this inertia?

: : : : * "Erratic Measure," in NDE for the Energy Industry 1995, pp.1-6. The American Society of Mechanical Engineers, New York, NY, 1995.

: : : :
: : Robert:

: : I'm happy to provide clarification and join in discussion. However, I believe you have misrepresented the scientific method by stating that the burden of proofis on the author of a peer reviewed article to reprove his thesis. Actually, the rule is that I am not allowed to republish the same material and that the thesis stands until it is refuted. It would not be necessary for you or anyone else to agree with the way in which I have supported my thesis in order to refute it.

: : The issue is whether Unit Measure is preserved by NDT. The thesis of the paper is that this axiom of probability theory is violated in a large class of NDT's tests. This thesis can be refuted by publication of a proof that Unit Measure is preserved. Such a publication has not, to my knowledge, been forthcoming.

: : While the statisticians you consulted did not understand the paper, my co-author is a statistician and an academic statistician read and understood it as a referee. I can't address your statisticians' difficulties without knowing what they are and neither they nor you have ever contacted me on them. Perhaps you would be willing to contribute their comments to the forum.

: : Terry

: :Rocky

My understanding of the role of peer reviewed publication in science is that it similar in effect to a finding in a criminal or civil trial. The issue can be reopened but the burden of proof is on the side which disputes the finding.

Perhaps you are alluding to the burden which is placed on any scientist to explain his theory in sufficient detail for it to be overthrown. I believe I have done this with respect to my allegation of Unit Measure violations, for it is clear that such an allegation can be overthrown by a proof that Unit Measure is preserved.

I do offer to discuss the paper in forums such as this one. Such discussions give participants the opportunity to probe for weaknesses in my position. I am happy to have them do this and encourage them to submit any weaknesses they think they have found to the rigors of peer reviewed publication. However, the present state of affairs is the unsatisfactory one in which the process of science has found a hole in the foundation of NDT and the NDT community is doing nothing about it.

Terry




 
00:43 Mar-19-1999

Rolf Diederichs

Director, Editor, Publisher, Internet, PHP MySQL
NDT.net,
Germany,
Joined Nov 1998
602
Re: Violations of Unit Measure : In a 1995 paper*, a colleague and I reported an error in NDT's foundation. In brief, all of its various flaw detection methods violate the Unit Measure axiom of probability theory in relation to theories of these methods' reliability. As all of NDT's theories of this reliability are stated in terms of probabilities, it follows that all of NDT's theories of this reliability are wrong.

: A hole having been identified in their field's foundation, one might have expected the NDT community to get right to work to repair it. However, so far as I know, nothing has been done four years later. Any ideas on overcoming this inertia?

: * "Erratic Measure," in NDE for the Energy Industry 1995, pp.1-6. The American Society of Mechanical Engineers, New York, NY, 1995.

:
--------------------
We are pleased to receive the copyright from ASME for a "e-reprint" of the article "Erratic Measure," by Terry Oldberg and Ronald Christensen in NDE for the Energy Industry.
The article will be published in NDTnet May 1999 issue.

Abstract:
The paper exposes an inconsistency in the field of non-destructive examination (NDE) and points out the dangers of the confusion which is created by it. The physical objects which occupy a statistical population occupy a partition of the complete set of physical objects under consideration. However, many of NDE's procedures test physical objects which do not occupy such a partition. The inconsistency is set up when NDE's scientists represent physical objects which do not occupy partitions as the elements of populations in their studies of NDE's reliability. The bogus populations invalidate the definition of probability as a measure of an event whose value on a certain event is 1.

With probability invalidated as a measure of a test's reliability, NDE's scientists have proceeded by representing a different measure of an event as a "probability." The value of this measure on a certain event varies between 0 and 2 in one U. S. Nuclear Regulatory Commission study. This makes the reader who interprets the study's measure as a probability wrong on a number of counts. For example, the apparently perfect certainty which accompanies a value of 1 for the study's Probability of Detection is actually perfect ambiguity because this measure's value on the certain event of a defect is 2. Dire consequences are imaginable if a reactor engineer were to act on the USNRC's representation.

The authors recommend avoidance of the consequences of this and other instances of confusion by the translation of past reporting. The translated reports would use a language which discriminated proper from improper probabilities and pop-ulations. For the longer term, they recommend the refocusing of NDE to yield only proper probabilities and populations.




 
07:59 Dec-05-2006
Zakary
Rigoberto http://koshkin.org/


 


© NDT.net - The Web's Largest Portal of Nondestructive Testing (NDT) ISSN 1435-4934

Open Access Database, |Conference Proceedings| |Articles| |News| |Exhibition| |Forum| |Professional Network|