15. Narrative on the lecture Professor Graham B. Wallis invited me to give in 1965, and the errors and lack of attribution in his publications on thermal stability.
My correspondence with Professor Wallis began with my letter to him dated October 28, 1963. I had read one of his articles on two phase flow and boiling, and I was so impressed by the article that I felt moved to tell him that it was “by far the most lucid and straightforward article I have encountered . . .” He replied in a letter dated November 21, 1963, after which we corresponded several times.
In 1965, in response to my request, and on the basis of correspondence in which I presented unpublished, original work on thermal stability, Professor Wallis’ letter of June 30, 1965 invited me to lecture at his summer course on two phase heat and fluid flow. The lecture took place at Dartmouth on July 9, 1965.
Several weeks after the lecture, Professor Wallis sent a letter to Nucleonics in which he used (without attribution) original, unpublished material from my letters and lecture, and expressed his agreement with my Nucleonics article. His letter was published in the October 1965 issue of Nucleonics under the title “A Vote for Adiutori”.
Professor Wallis’ letter of July 8, 1966 enclosed the thermal stability section of his notes for his 1966 summer course on two phase heat and fluid flow. In the notes, he derived a thermal stability criterion for a specific boiler type starting from the general criterion in my Nucleonics article. An error in his derivation resulted in an erroneous criterion that failed to account for the effect of the boiler plate on thermal stability.
In the Hale and Wallis article published in 1972, the above erroneous derivation is repeated, the error in the resultant criterion is recognized and described, and the error is erroneously attributed to a lack of rigor in the general criterion. The article then presents an alternative derivation that results in Criterion (24), the correct criterion for the subject boiler type. Criterion (24):
· Is identical to Criterion (6) in my letter to Professor Wallis dated June 14, 1965.
· Was presented in my lecture at Professor Wallis’ summer course on July 9, 1965, and its application to Berenson’s boiler and results was discussed.
To my surprise and dismay, the Hale and Wallis article contained no attribution to my letter or my lecture.
The article also states
The limitations imposed on Berenson’s data by the high wall resistance were first discussed by Wallis in his (“A Vote for Adiutori”) comments on Audiutori’s (sic) original paper.
The above statement is in error. It should state that the limitations imposed on Berenson’s data by the high wall resistance were first publicly discussed by me in ”Author’s reply” published in the December, 1964 issue of Nucleonics, and privately in my letter to Professor Wallis dated June 14, 1965, and in my lecture given on July 9, 1965 at Professor Wallis’ summer course on two phase flow and heat transfer.
In my letter to the editor of I & EC Fundamentals dated August 18, 1972, I requested that he publish my “letter to the editor” that corrected errors and described the lack of attribution in the article by Hale and Wallis. Predictably, the editor saw no need to publish such a letter.
Correspondence that resulted in an invitation from Professor Wallis to lecture at his summer course on two phase flow and heat transfer held at Dartmouth College.
In May of 1965, I received a flyer from Professor Wallis announcing a two week summer course on two phase heat and fluid flow to be given at Dartmouth by Professor Wallis, Novak Zuber, and John Collier. My letter to him dated May 21, 1965 indicated that I did not wish to enroll in his course, and made the following offer:
Some time near the end of the two weeks, I would be pleased to present one or two one hour lectures dealing with the results of my conceptual research on two-phase heat and mass transfer. . . . (I) offer to present the lectures without fee and at my own expense.
The reason I offered to lecture “without fee and at my own expense” is because I had tried to present and publish the application of the theory presented in my Nucleonics article, but had been prevented from doing so by reviewers and editors who (I feel) rejected my work because they correctly sensed that it was revolutionary, and that I was not a member of the club. As a result, I was extremely anxious to present the results of my work to an audience competent in two phase heat and fluid flow, such as persons who had completed Professor Wallis’ course.
There is a possibility that we might fit you in as a “special topic” on the last day of the course. Before deciding one way or another, I will need to consult with Collier and Zuber about the amount of time they may be able to give up to make room for you. Also, as I am not familiar with your work, I find it hard to find a basis for a rational decision. Perhaps you can supply me with more evidence which will convince me. I will try to be quite objective and not be influenced either by your own opinion of your work or the opinions of those who have resisted its acceptance.
My letter to Professor Wallis dated June 14, 1965 was my response to “supply me with more evidence which will convince me”. The letter presents my view on the relationship between heat flux and interface temperature difference in nucleate boiling, and applies the thermal stability theory presented in my Nucleonics article to the boiler used in “Experiments on Pool-Boiling Heat Transfer” by P. J. Berenson, Int J Heat Mass Transfer, v 5, pp 985-999, 1962. The letter gives the general criterion for thermal stability, and then states:
For a boiler such as Berenson used for his thesis, this general form may be reworked into the particular form
-(1/hs + tw/kw)-1 <s dq/dTw (6)
where the right hand side of equation (6) of course refers to the slope of the pool boiling curve. From this criterion, it may easily be seen that:
1. The statement that such a boiler is necessarily stable at all points of the pool boiling curve is simply not true.
2. The stability of such a boiler can be improved markedly by simply increasing the steam side coefficient, decreasing the thickness of the wall, or increasing the thermal conductivity of the wall. . . . (Berenson) intentionally made the boiler plate very thick in spite of the fact that he wanted the boiler to be as stable as possible since his thesis subject was transition boiling.
It is important to note that although Criterion (6) is quite simple, it was original. It was the result of the first application of my theory of thermal stability—the first thermal stability criterion derived for a specific type boiler. Nothing like it had ever appeared in the literature, although I had tried unsuccessfully to arrange its publication.
Thank you for your letter of June 14th.
I would be glad to have you give a lecture on July 9th in the afternoon on a subject of your choice. . . .
. . . I agree with your analysis and comments on Berenson’s work which seems quite obvious once you have pointed it out.
Berenson’s article was derived from the Sc. D. thesis he submitted in 1960 at MIT. The thesis adviser was Professor Peter Griffith. The first sentence of the thesis abstract states:
An experiment, utilizing a condensing fluid as the heat source, was performed to determine the heat flux vs. temperature difference curve for transition pool boiling from a horizontal surface.
(Compare Professor Wallis’ “seems quite obvious once you have pointed it out” with the Argonne comment “The most consistent fundamental error throughout the paper treats temperature as an independent variable”. In other words, the Argonne critics considered it irrational to take derivatives with respect to temperature because temperature is not an independent variable. Therefore, in their view, the “obvious” Eq. (6) was irrational and of no use.)
The lecture I gave on July 9, 1965 at Dartmouth.
My lecture at Dartmouth on July 9th is described in my lecture notes. The first part of the notes is in the form of a manuscript. The second part of the notes is in the form of an outline.
The lecture began with a review of the conventional view of thermal stability, and a description of what is wrong with the conventional view. The general criterion for thermal stability is given without derivation in the form
dQin/dTw <s dQout/dTw
in which Q is heat flow rate (not heat flux). (I usually express the general criterion in terms of heat flux rather than heat flow rate. In this case, I used heat flow rate because I feel that the criterion is easier to comprehend if it is first presented in terms of heat flow rate rather than heat flux.)
The notes state:
The criterion results from idealizing a generalized system and uncoupling it at the boiling interface. The uncoupled pieces are then analyzed to determine whether they would fit together in a stable manner. The derivative on the left hand side . . . answers the question “How is the heat flow into the boiling interface affected by the temperature of the interface?” The right hand side . . . answers the question “How is the heat flow out of the boiling interface affected by the temperature of the interface?”
A boiler of the type used by Berenson is described, and its stability analysis is described analytically and graphically. It is shown that, for such a boiler,
dQin/dTw = -A/(1/hs + t/k)
where A is the area of the boiler plate. Thus the stability criterion for this boiler type is given by
-A/(1/hs + t/k) <s dQout/dTw
Berenson’s transition region article is discussed, and the lack of transition region data noted in the following:
If you will plot up some of (Berenson’s) data on linear graph paper, you will find that, in 17 of the 20 experiments, virtually no data was obtained in the transition region, in agreement with the new theory of thermal stability.
With regard to the design of a boiler such as the one used by Berenson, the notes state:
. . . the designer can exert a very strong influence on the thermal stability of the boiler in any of the following simple ways:
1. Minimize the thickness of the boilerplate
2. Maximize the thermal conductivity of the boiler plate.
3. Maximize the heat transfer coefficient of the fluid condensing on the heat source side of the boiler plate.
The phenomenon of burnout (ie a large increase in temperature resulting from an incremental increase in heat flux) in so-called constant heat flux systems such as nuclear reactors and electrically heated boilers was discussed, and it was pointed out that burnout does not necessarily result in such systems. For example, if the temperature coefficient of reactivity of a nuclear reactor were sufficiently negative, the stability criterion would be satisfied at all values of the boiling interface temperature, and there would be no burnout.
The manner in which thermal instability can result in undamped, oscillatory behavior was described.
The so-called dry wall phenomenon was discussed, and it was suggested that it is thermally induced rather than hydraulically induced.
Hydraulic stability and its resemblance to thermal stability were discussed.
I thoroughly enjoyed giving the lecture, and sensed that it was well received. I was particularly honored by John Collier’s request for my autograph.
Professor Wallis’ letter to Nucleonics published in the October, 1965 issue under the title “A Vote for Adiutori”.
Several weeks after my lecture, I received a letter from Professor Wallis dated July 27, 1965 in which was enclosed a copy of a “letter to the editor” he had sent to Nucleonics. The “letter to the editor” had obviously been prepared using (without attribution) original, unpublished material on thermal stability from my letters to Professor Wallis and my lecture of July 9, 1965. The “letter to the editor” discussed Berenson’s boiler and results, and expressed agreement with my Nucleonics article. It was published in the October 1965 issue of Nucleonics under the title “A Vote for Adiutori”.
I was dismayed that Professor Wallis’ “letter to the editor” did not state that it was based on the original, unpublished stability criterion in my letter to Professor Wallis dated June 14, 1965, and that I had discussed its application to Berenson’s boiler in the lecture I gave at his summer course on July 9, 1965. However, I was so elated that Professor Wallis had had the courage to publicly endorse my work that I made no effort to have the lack of attribution corrected.
I certainly did not anticipate that Professor Wallis would one day present my Criterion (6) as his own original work, or that he would credit himself with being the first to consider the effect of boiler plate thickness on thermal stability, both of which he did in Thermal Stability of Surfaces Heated by Convection and Cooled by Boiling by Hale and Wallis, I & EC Fundamentals, Vol. 11, No. 1, 1972.
The fundamental error in Professor Wallis’ notes for his 1966 summer course on two phase flow and heat transfer.
Professor Wallis’ letter of July 8, 1966 enclosed the section dealing with thermal stability in his lecture notes for his 1966 summer course on two phase heat and fluid transfer. At that time, I briefly perused the material, then filed it.
In preparing this narrative, I examined the subject lecture notes more carefully, and was amazed to find the following in Section 14.6.5 entitled “Pool Boiling From a SurfaceWhich is Heated by Convection (Perhaps Condensation) from a Fluid at Constant Temperature T1”, the boiler type used by Berenson.
The stability criterion now becomes
-h1 – mb < 0 (14.108)
Thus a negative value of mb can be counteracted by having a suitably large heat transfer coefficient h1. This was used experimentally by Berenson15 and Owens29.
(mb is the slope of the boiling curve—ie is the slope of a curve of heat flux versus the temperature difference at the boiling interface.)
Criterion (14.108) states that, for the subject boiler type, the boiler plate has no effect on thermal stability. This violently disagrees with my “Author’s Reply” in the December, 1964 issue of Nucleonics, my letter to Professor Wallis dated June 14, 1965, and my lecture at Professor Wallis’ summer course. Criterion (14.108) also disagrees with the criterion that underlies Professor Wallis’ “A Vote for Adiutori” published in the October, 1965 issue of Nucleonics!
The correct expression of Wallis’ incorrect Criterion (14.108) is Criterion (6) in my letter to Professor Wallis dated June 14, 1965.
The errors and lack of attribution in the article on thermal stability by Hale and Wallis published in I & EC Fundamentals in 1972.
The article by Hale and Wallis claims to demonstrate that
dqin/dTw < dqout/dTw (1)
. . . eq 1 (the general criterion for thermal stability presented in my Nucleonics article) . . . is valid only if the thermal resistance of the wall is neglected.
This statement is simply not true. What the article by Hale and Wallis actually demonstrates is that if the derivatives in Criterion (1) are not evaluated correctly, the result will be a criterion that is incorrect.
The correct evaluation of the derivatives in Criterion (1) is quite simple. It requires only that both derivatives be evaluated at the boiling interface. (Note that I covered this on page 8 of my lecture notes for Wallis’ 1965 summer course.)
Hale and Wallis evaluated one of the derivatives in Criterion (1) at the boiling interface, and evaluated the other derivative at the opposite interface. Of course the result they obtained was incorrect. They erroneously concluded that Criterion (1) is incorrect, but they should have concluded that their analysis was incorrect.
The article by Hale and Wallis states that, for a convectively heated wall, Eq. 1 results in the stability criterion
hin > - dqout/dTw (5)
The article also states
Wallis and Collier (1967) applied eq 1 to several different design situations . . .
In other words, in the notes for their 1967 summer course at Stanford, Wallis and Collier appraised the stability of convectively heated boilers using Criterion (5), a criterion that erroneously indicates the boiler plate has no effect on thermal stability. This same erroneous criterion had previously been presented in Professor Wallis’ notes for his 1966 summer course.
The article indicates that, sometime between 1967 and 1972, Hale and Wallis deduced that Eq. (1) was incorrect. They then deduced alternative methodology that resulted in their Criterion (24) which states that the boiler plate has a strong effect on stability.
Criterion (24) in Hale and Wallis is identical to Criterion (6) in my letter to Professor Wallis dated June 14, 1965. Criterion (24) is presented as original work by Hale and Wallis, and there is no attribution to my letter to Professor Wallis, or to my lecture at Professor Wallis’ summer course in 1965.
Amazingly, the article also states
Criterion (24) was used by Wallis (1965) in analyzing Berenson’s data.
This statement is amazing because it raises the questions:
· How could Professor Wallis deduce Criterion (24) in 1965, when he could not do so in 1966 or 1967?
· If Criterion (24) cannot be deduced from Criterion (1) as Hale and Wallis claim, how did Professor Wallis determine it in 1965?
The Hale and Wallis article also states:
The limitations imposed on Berenson’s data by the high wall resistance were first discussed by Wallis in his comments on Audiutori’s (sic) original paper. ( Wallis’ comments were published under the title “A Vote for Adiutori” in the October, 1965 issue of Nucleonics).
The above statement is not true. The truth is that the limitations imposed on Berenson’s data by the high wall resistance were first discussed publically by me in “Author’s Reply” in the December, 1964 issue of Nucleonics, and privately in my letter to Professor Wallis dated June 14, 1965, and in my lecture given on July 9, 1965 at Professor Wallis’ summer course on two phase flow and heat transfer.
(Limitations on Berenson’s data were also discussed in my manuscript entitled “Transition Boiling—the Relationship Between Heat Flux and Thermal Driving Force” submitted for publication to the AIChE Journal on March 19, 1964, and rejected on May 5, 1964.)
The first public discussion of “the limitations imposed on Berenson’s data by the high wall resistance”.
The first public discussion of “the limitations imposed on Berenson’s data by the high wall resistance” is the following from my “Author’s Reply” published in the letters section of the December, 1964 issue of Nucleonics:
. . . Rohsenow states on p. 138 of (Modern Developments in Heat Transfer, W. Ibele, ed. (1963): “With condensing vapor as the heat source on one side of a wall, any point on the entire (pool boiling) curve can be reached under stable conditions.” My concept of thermal stability demonstrates that this conclusion is wrong and the experiments by Berensen (sic) (“Transition boiling heat transfer from a horizontal surface, Sc.D. thesis, MIT, ; also Int. J. Heat Mass Transfer, 5, 985-999 ) proved it, although apparently neither Berensen nor Rohsenow realized it.
In his experiments, Berensen built just such a boiler as described above, and he reported the results of 20 experiments, each of which purports to cover the entire pool boiling curve. If the reader will plot Berensen’s results on LINEAR graph paper, he will find that there is essentially NO data in the so-called transition region for 17 of the 20 experiments. The reason for this predominant lack of data was that his boiler was usually NOT thermally stable in the transition region. The fact that this was not recognized is doubly important for the following reasons:
1. Since nothing was known about thermal stability at that time, Berensen understandably did not look for thermal instability, and as a result he did not report it, even though it was there.
2. Owing to a lack of understanding of thermal stability, Berensen correlated the transition region “results” of the 17 experiments which contained no such data! The three experiments which contained the desired data did not “agree” with the majority and were therefore treated as oddballs!
3. Since nothing was known about thermal stability at that time, Berensen understandably did not design his boiler with thermal stability in mind. As a result, he INTENTIONALLY built in a very thick boiler plate, virtually guaranteeing that he would not get the desired data.
Berenson made the boiler plate very thick (2.25 inches) in order to improve the accuracy of the heat flux results which were based on data from thermocouples imbedded in the boiler plate. Berenson’s results were presented on log log paper, and this obscured the fact that very little data had been obtained in the transition region. If the results had been plotted on linear paper, the lack of transition region data would have been readily apparent.
My unsuccessful attempt to correct the errors and cite the proper attribution in the article by Hale and Wallis.
In a letter dated August 18, 1972 to the editor of I & EC Fundamentals, I requested that he publish my “Letter to the Editor” that corrected errors in the Hale and Wallis article, and cited the lack of proper attribution.
Predictably, the editor (Professor Pigford) saw no need to publish such a letter.
When I read the article by Hale and Wallis, I was dismayed and resentful because:
· The stability criterion that was the main thrust of the article was my original work.
· I had been unable to arrange for the publication of the stability criterion.
· It was my guess that Hale and Wallis had no difficulty arranging for the publication of my criterion.
My response to the criticism by the Argonne Seven and other critics, published in the December, 1964 issue of Nucleonics, closed with the following prophetic statement:
That my concept of thermal stability has not emanated from the universities or the national laboratories is perhaps regrettable. However, it is no less true that it represents new knowledge to the science of heat transfer. It is now incumbent on the universities and those who control the scientific literature to see to it that this new knowledge is not wasted—and to make it possible for me to publish my work in a less brief and more satisfactory manner. Whether they shall prove equal to the task is a matter for their conscience—not mine.
Persons who control the American scientific literature have done nothing to make it possible for me to publish my work in a less brief and more satisfactory manner. In fact, they have made it so difficult that, in spite of my many attempts to arrange for the publication of my work in American Journals, none of my work has ever been published in an American peer reviewed Journal.
In 1964, one of my articles was accepted for publication in an American peer reviewed Journal (the AIChE Journal), but it was never published there because a member of the club (to whom I had naively mentioned the article) pressured the editor to “unaccept” it prior to publication. (The editor of the Journal, Professor Harding Bliss of Yale University, told me it was the only article ever accepted for publication in the AIChE Journal, and then not published.) The article was finally published in a Japanese Journal thirty years after it was accepted for publication in the AIChE Journal! And it was just as timely and important in 1994 as it had been in 1964!
The article was A Critical Examination of the View that Nucleate Boiling Heat Transfer Data Exhibit Power Law Behavior, Japanese Society of Mechanical Engineers International Journal, Series B, Vol. 37, No. 2, 1994, pp 394-402.
The background on my accepted-but-not-published article is related in Narrative on my 1964 paper accepted for publication in the AIChE Journal, but never published there.
The fact that researchers at MIT, a world class engineering school, did not know how to optimize the design of an experimental boiler to operate stably in the transition region was eloquent testimony to the originality of my Nucleonics article. The fact that Berenson’s boiler was unable to operate stably throughout most of the transition regions tested is eloquent testimony to the validity of my view of thermal stability in general, and the validity of my general criterion for thermal stability in particular. Unfortunately, the originality and validity of my work has served only to unite peer reviewers and journal editors against its publication.