Mizuno : Publication of kW/COP2 excess heat results

Quote from kirkshanahan

Mizuno call the heat capacity Hc, and gives an equation for it as Hc = .987 + .00661 * Tout (Eqn. 1 on pg. 12).

Quote from Zeus46

if a linear fudge is really needed, Hc = 0.975 + 0.000105*Tout would be much better.

Having said that, in the limited temperature range Mizuno is working in (273-333K), a regression of Hc = .987 + .0000661*T_out works well...

Maybe a couple of zeroes have been missed off?

Quote from maryyugo

Really, this is worth arguing over?

I'd describe it as an explanation rather than an argument.

Quote from maryyugo

It would be a pity if this was all there were to it.

I'm not sure you understood said intelligent remark though... I am suggesting it's a printing error, rather than something Mizuno has been getting wrong for the last decade or two. Mainly because actually using the Hc equation as written in the report would lead to very odd spectacular results. (i.e. an erroneous COP of ~5.5, as a guess... that is, Kirk's 2.93 * 2ish)

"If you write in Engrish"

Thereyugo Maryugo

This is an arrogant monocultural monolingual slur

I ran the paper through spell/ grammar check as undoubtedly Mizuno did.

There were no errors and I doubt if Professor Nagel found many.

All the Japanese native speakers I know make sure to run their written stuff by English native speakers.

I am quite sure that Mizuno did so.

Indeed Mizuno wrote at the end of the manuscript

"Special thanks are extended to David Nagel for kindly reviewing our manuscript."

Perhaps Maryugo needs to improve her/his reading skills

Zeus "Having said that, in the limited temperature range Mizuno is working in (273-333K), a regression of Hc = .987 + .0000661*T_out works well...

Maybe a couple of zeroes have been missed off? "

I'd agree with that.. probably Mizuno originally wrote 0,0000661.

Zeroes can be lost easily during changes in text formats.

The heat capacity data I've found is rather limited

http://www.engineeringtoolbox.com/air-properties-d_156.html

yields btw 273 and 373K

Hc =0.990 + .0000514 *T_out  but the regression is poor. R2=0.77 only.

I guess Mizuno is doing isotope readings where possible... interesting to find out if the carbon is old or new.

Funds had to be re-directed to the lawyers - a crime against humanity that will be eventually undeniable.

TTM - The money lost to Rossi may end up not being a total waste as it did create awareness that caring risktakers were showing up with resource.

That opened some doors - some of which may still turn out to be useful. The criminal shame is the $5M+ that was designated to fund research had to be diverted to attorneys. Perhaps those lessons will end up having convertible value as well.

Alan - I agree with the lawyers statement. The accountants are a result of regulatory burden overreach - they are simply filling a need. I think that has a chance of getting fixed in a US to the extent where more balance in the mix might be possible.

Catching up...

@MY - Yes, that's the paragraph that I was referring to. The given equation is way off vs. the given values, given values essentially match lit. data. I got air heat capacity data here: http://www.engineeringtoolbox.com/air-properties-d_156.html

@Z46 - I see you figured out the likely equation. Now the only question is: "What did Mizuno actually do/use?" When things don't make sense in the paper, it forces us to guess, maybe we guess right, maybe we don't. There are several instances of inadequate descriptions in this paper that should have been flagged by the reviewers and corrected before publication. In the end, they make this paper LTA. maybe Mizuno found something, maybe not, but we won't know from this paper, unless a modified version is published with inadequacies fixed.

Quote from kirkshanahan

should have been flagged by the reviewers and corrected before publication.

It's a pre-print.

Quote from Zeus46

It's a pre-print.

Which normally means it has been proofed by the author and returned for publication as is. But that really doesn't matter. I hope it gets fixed, whatever it is...

There are a lot of other problems with it too BTW...

While I understand that the Maryyugo Show is a staple on any forum mentioning LENR, I wonder if giving hir constant attention is relevant to the subject clearly mentioned in the topic header, just as I wonder if it is really necessary to speak about Rossi when it seems that r, o, s, s, and i, do not spell Mizuno, and that the E-Cats and his inventor have numerous other threads where people can go on for ages about how he's a quack, a criminal against humanity and such a conman deserving no attention that some persons have to write about him daily for hours on end.

I now have some spreadsheets in readable format. I will upload them so people can see for themselves exactly how Mizuno did the calorimetry. It may take me a little while to prepare them.

Comments on “Observation of excess heat…” by T. Mizuno, Preprint J. Cond. Matter Nucl. Sci. 25 (2017)

Executive summary of these comments for those who only want the overview…

This paper has numerous errors, conflicting information, and unanswered questions in it. Therefore until it is corrected, it cannot be used as evidence of excess heat’s existence.

----------------------------------------------------------------------------------------------------------------------------------------------

In discussions on lenr-forum, it has already been pointed out that Mizuno used mass flow calorimetry to measure heat output of a putative LENR reactor. That method requires mass flow rate, heat capacity of the calorimeter fluid (air in this case), and a temperature differential. It was pointed out that the given equation for the heat capacity (Eqn. 3) was in error, and it was noted that adding in two 0’s to the multiplicative coefficient would apparently correct the discrepancy. However, until that correction is made by the author, the paper’s presented results can not be used. The discrepancy introduces a significant additional multiplier to the power output calculation. Readers need to know exactly how Mizuno calculates his output power. Guessing at what Mizuno did is not good scientific practice. Putting that issue aside for the moment, several additional comments can be made.

What is apparent in this paper is that only one branch of the decision tree for resolving anomalous readings is being investigated, namely the ‘LENR’ branch. There is no indication of any attempt to define and test other possible, more ‘mundane’ explanations for the anomalous results. Therefore, these comments will focus on that process by identifying specific anomalous features and positing mundane explanations for them, whereupon their impact and potential resolution will be discussed. Additionally, some additional comments of the ‘peer review’ type will be presented.

The extent of replication has some relevance here (replication here focuses strictly on replication by Mizuno of the reported results). It is unclear from the paper how many separate reactor-load sets were tested. Was it one calibration reactor run under different conditions or 2 or 3 reactors run separately at those different conditions? Likewise, how many ‘excess heat producing’ reactors were actually tested. The data figures give multiple results at different input power levels and treatments, but is this one reactor run many times, or several reactors each run just a few times? This needs to be clarified. Of course, the fewer reactors used, the lower the overall reliability of conclusions.

To seek ‘mundane’ explanations, the technique is to follow the general approach of a Propagation of Errors calculation and examine the error potential of each term in the relevant equations. Those basic equations are:

Pexcess = Poutput – Pinput

Pinput(heaters) = I * V

Poutput = mass flow * heat capacity * temperature difference(outlet – inlet)

In general, no serious problems with Pinput has been noted and little more will be said about that except to note Mizuno defines this as equation 2 on page 12. Other commentators on other experiments have noted however that sometimes measured input power is actually dissipated elsewhere that where it is assumed to go. In the end, that possibility may need to be eliminated.

Mizuno speaks of calculating the “caloric value of air”. This means computing the Poutput (Pout, or Hout in his paper) as noted above. He gives equations for calculating the heat capacity of air and the air density (Eqns. 1 and 4), and states more explicitly the Poutput equation as Equation 3. He also gives a ‘semiempirical’ (meaning experimentally determined) equation for air velocity as a function of power to the blower (eqn. 5).

Taking the Poutput terms in order, mass flow is given by multiplying together the air velocity, V, the exhaust opening cross sectional area, S (essentially fixed and thus of little concern), and the air density, r. The air density is dependent upon the chemical composition of the air. The biggest unknown in that involves the water content.

Mizuno states “In calculating the caloric value of air, the caloric value varies depending on atmospheric pressure and humidity in addition to temperature. The atmospheric pressure was [measured] with a [calibrated] commercially available barometer made by Sunoh company and humidity by a measuring instrument made by Empex company. There was no significant change during the 80 ks measurement period.” [Corrections indicated with brackets.]

Thus we are given no information on the air composition in any of these tests, and we are expected to ‘just trust’ Mizuno that this information is unimportant (“…no significant change…”). However, prior experience in this field has repeatedly shown CF researchers underestimate the impact of variation on their results. Cases in point are the impact of calibration constant shifts and the accuracy/precision of water loss measurements in F&P-type electrolysis cell experiments. Therefore this blanket statement by Mizuno is not adequate. Data is required. Was it raining one day? 50% RH? 10%RH? Summer or winter?

It seems eminently reasonable to assume as the air density changes (also due to T changes) that the attained velocity for a given blower power input will change. This needs to be explicitly examined. In other words, the natural variation (sometimes called the ‘error’) in equation 5 must be evaluated. The equation appears to be a linear regression fit of an exponentially varying “X” value (exp(-Wb/w)), with the usual “A” (slope) and “B” (intercept), but the “w” term is an additional constant not normally seen in simple linear regression. The values of A, B, and w are given, and from that we can see that at 0 input power, ‘somehow’ the blower produces 0.3 m/s air movement. Of course, this is just a consequence of the statistics, but it clearly indicates there are regions of validity that must be known. Specific errors associated with each constant need to be specified.

Mizuno shows the calibration data in Figure 17. Of note is that it spans (approx.) ~0.2 to 4.8 Watts blower power. Yet he states (p.12, last full paragraph) “Blower input power was approximately 4 W to 6 W.” Thus he extrapolates the calibration curve for values above 4.8W, which is not always a ‘safe’ thing to do. Good scientific practice is to calibrate over the entire experimental range, and not to do just part of it. He also says “In a usual test, the input power of the blower is 5 W, so the wind speed is 4 m/s.”, which is a less extreme extrapolation and given the flattening out of the curve observed it may not be significant. But again, what is needed is error bars on the constants he uses.

The air density itself is calculated apparently (eqn. 4). Again, no composition information is incorporated, so we can’t estimate the changes due to changing humidity. And, as usual, there are no error bars on the regression constants (the values 3.391, 201.26, and 0.41529). So one again this is LTA (less than adequate). This factor is a direct multiplier in the Pout equation, plus the real value probably affects velocity as noted above.

With regards to the heat capacity term, one problem has already been mentioned (the missing “00”). However, heat capacity also will be impacted by chemical composition, so the lack of that information is a possible issue again. As well, no error bars on the constants of eqn. 1 are given. LTA.

Finally, in what is probably another ‘typo’ type of error, the given equation (eqn. 3) is obviously missing a dT multiplier in the middle expression. One can easily detect this by doing unit analysis, which is where you make sure the units on the computed value correctly fall out of the equation. Hout is in Watts or Joules. Leaving out the dT gives either W/degK or J/degK. The dT term is present in the rightmost expression.

Moving to the last Pout term, the temperature differential, one always likes to assume temperature is so easy to measure that there ‘certainly’ is no error in the measurements, but is that really the case? Are there any indications of a problem?

Examining Figure 20 for the calibration run, the radical change in ‘room temperature’, i.e. Tin, seems suspicious. Mizuno explains this by stating “because there is no air conditioner in the laboratory” (p.14). But typically room temperature doesn’t change so quickly (seems to jump 2-3 degrees in <1 hr., did someone open a window?), at least in our experience. Some expanded explanation is needed because the alternative explanation is that there is some electrical feedback going on. Ed Storms had this problem in his first set of data generated in January 2000. He redid his grounding scheme and mostly eliminated the problem. Perhaps Mizuno has a similar problem in his lab with no air conditioning? That fact suggests an older building with perhaps sub-standard wiring. Figure 26 shows a similar plot for a different run, this time for one that supposedly showed excess heat. There, the increase is not as dramatic, but is still present. And it shows a similar shape, which adds to the suspicion level. To summarize, the temperature traces might indicate electrical crosstalk or ground loops. Otherwise the similarity in profiles seems unusual and suggests perhaps that the experiments were run at the same time on different days that had substantially the same temp profile. It would be nice to see results from a run that had a different Tin profile, just to be sure we are not looking at feedback from heater circuits or something similar.

So, summarizing the issues in computing the excess heat the following comments were made:

• No error bars on equations to allow error propagation calculation
• No factual basis given for not reporting humidity and pressure data
• Errors in equations found that may or may not just be typos

These factors lead to an inability to trust the supposed excess heat measurements reported in this paper.

Mizuno also presents several figures relating to scanning electron microscopy (SEM). We note here that several are elemental composition maps, which are captioned as if they were SEM images. This is a technicality, but they are not SEM images, they are EDX maps (Energy Dispersive X-ray (analysis)). Figures 33, 34, 37, 38, and 39 are all EDX maps. None show any useable features, thus the elements being mapped are detected as perfectly uniformly distributed in these images. This means showing the images as figures is pointless. Simply stating the above facts would be all that is needed.

However, there is one interesting point to made from noting that Figures 37, 38, and 39 are O, C, and N maps of the nickel wire surface from a reactor that purportedly showed excess heat. But Mizuno stated “Figure 34 shows the distribution of oxygen after treatment of Ni reacting metal. There is no oxygen observed. When such oxide film, nitride film or carbon are present in large amounts, it is difficult to generate excess heat. Removing impurities completely from reactant metal is one of the important conditions for successful excess heat generation.” Yet in relation to Figures 37, 38, and 39, Mizuno says “Figures 37, 38 and 39 show the distribution of oxygen, carbon, and nitrogen, respectively. These elements are widely present on the nickel wire surface. In particular, it can be seen that a large amount of carbon is present. It is possible these elements were originally present inside the reactor. However, it is clear that these elements increase on the electrode surface as the excess heat increases.” So which is it, the impurities help, or they hurt? Where are they coming from? Did ‘LENRs’ create new elements?

The other images presented offer little useful information as well. First is the problem of scales. Figures 31 and 32 are of areas just a few micrometers across, Figure 34 is almost a millimeter across, and Figure 36 is perhaps 300 micrometers across. The difference in scales makes it nearly impossible to correlate these images, unless they were to be physically related as expanded views of one area, but no such comment identifies that this is what was done. So the images are disconnected and impossible to relate except in the most general terms. What is shown differs little from pictures of any wire, so the question is: “Why are these pictures even included?” The whole section 3.2 is of minimal value and probably would be recommended for deletion in a properly reviewed paper. Otherwise a much better job of presenting information to support whatever point is trying to be made is required.

In Figure 40, Mizuno attempts to relate the amount of apparent excess heat to the reactor temperature. Of note is the flyer point at low excess heat. The tendency many people show is to ignore flyers. That may be adequate if one is trying to characterize the bulk of the data, but the fact always remains that the experimental system produced the flyer point. Something caused that. How often might it reoccur? How wide might the variation actually be? In this case there is 1 point out of 20, and the spread is almost as large as the whole span of the rest of the data. What would one see after running 100 experiments? Would the implied relationship of Figure 40 persist? At this point, there are no answers to these questions, but they are very important. If the supposed dependency actually vanishes with more work, conclusions drawn today are not worth much. Why the flyer appeared at all must be adequately investigated, such that Mizuno could produce such a point on demand. This is the full control of the experiment required to claim full replicability (internal to this study). As it stands, the conclusions reached should only be deemed provisional at best.

In the end, this paper illustrates the generic problem with papers in the CF field. There is much hidden information that might be important, but is never discussed. Alternative explanations are never considered. Errors are never quantified. In fact, this is difficult to do for a reader, leaving him or her unable to decide what the value of a given paper is. What is always needed is the raw data, which is why the Storms data set was such a contribution and why the McKubre refusal to disclose methods was so poor. Perhaps the spreadsheets mentioned by Jed Rothwell will shed additional light on these questions.

In the end, this^ comment illustrates a generic problem with online criticism in the CF field.... Namely, it looks like another case of the Unbounded Error Gambit! ...Point out a possible source of measurement error, and assume/imply it must be extremely significant.

In this case, we can check how big these effects could be:

Quote from kirkshanahan

Taking the Poutput terms in order, mass flow is given by multiplying together the air velocity, V, the exhaust opening cross sectional area, S (essentially fixed and thus of little concern), and the air density, r. The air density is dependent upon the chemical composition of the air. The biggest unknown in that involves the water content.

A unrealistically huge swing in (indoor) Relative Humdity from 25% to 70% (at 20C) gives a maximum change in water mass of 8.64g/m3

4.184 1.864 J/g.C * 8.64g/m3 * 60-20C * 0.03m3/s = 19W total possible error due to the effects of RH.

...Or in other words, a possible phantom 0.96W gained for every 1% drop in relative humidity since the control run.

Quote from kirkshanahan

It seems eminently reasonable to assume as the air density changes (also due to T changes) that the attained velocity for a given blower power input will change. This needs to be explicitly examined.

The maximum possible change in air density between 100% and 0% RH is 23g/m3. Dividing that by 1060g/m3 dry air density at 60C gives a maximum possible swing of 2.1%. The effect on the the efficiency of a centrifugal pump would be absolutely negligible (To be thorough: 2.1% / 2.31 = <1%, ...or... 11.1% / 2.31 = <5%, if Mizuno's barometer is also broken and the weather is fluctuating between extremes).

Quote from kirkshanahan

The air density itself is calculated apparently (eqn. 4). Again, no composition information is incorporated, so we can’t estimate the changes due to changing humidity

You can't eat the same piece of cake twice, Kirk.

Quote from kirkshanahan

With regards to the heat capacity term, one problem has already been mentioned (the missing “00”). However, heat capacity also will be impacted by chemical composition [i.e. relative humdity/water content - Z46], so the lack of that information is a possible issue again. As well, no error bars on the constants of eqn. 1 are given. LTA.

...Or three times.

Quote from kirkshanahan

Errors are never quantified.

Oh, the irony.

Quote from kirkshanahan

The other images presented offer little useful information as well. First is the problem of scales. Figures 31 and 32 are of areas just a few micrometers across, Figure 34 is almost a millimeter across, and Figure 36 is perhaps 300 micrometers across. The difference in scales makes it nearly impossible to correlate these images, unless they were to be physically related as expanded views of one area, but no such comment identifies that this is what was done.

Everyone likes a pretty SEM photo, and... Are you being serious? It says right there "Figure 36 is an enlarged photograph of the sample in Fig. 35", and all the scale bars are fairly obvious, it would seem.

So the executive, executive summary is... There's *very likely* a couple of typoes (one is a bit of a howler), and you only somewhat have to trust Mizuno's claim of an insignificant change in relative humidity (For a more realistic - assuming the aircon is broken - maximum swing in RH from 60% to 30%, the largest possible error is 19W).

Edit: Also, an extreme swing in barometric pressure from 1055hPa to 960hPa, could produce an additional worst-case error of 9%.

Edit 2: Alan Smith points out the need to use specific humidity rather than absolute humidity. I just say this means than my 19W error margin, has it's own error margin of about 10%...

Quote from bocijn

Lets say the error is 10 percent? 152+/-15 MJ compared to 83+/-8 MJ gives a minimum COP of 1.5, maximum of 2.2

If the error is 20 percent..................................................................................gives a minimum COP of 1.2, maximum of 2.7

If you are working in MJ, it's not really the "COP" that you are calculating.

Zeus46 said "it's not really the "COP""

True ,

it isn't really COP,

Mizuno is translated as "the output/input ratio " =1.841 .

Perhaps we should use OIR or maybe 入力出力比 (maybe Jed can clarify) Nyūryoku shutsuryoku-hi ??