Kirk Shanahan's critique of LENR experiments

Quote from JedRothwell

If you are correct, your findings will void many aspects of electrolysis going back to Faraday.

This is another of your ridiculous strawman arguments. My ATER/CCS proposal is based on the fact that the CF cell designs have a particular design flaw. It would *only* impact past work where that design flaw was also present. As well it *requires* a shift in the steady state to occur, and past work is unlikely to have an issue like that. That would be difficult to find in past work since normally the possibility of a secondary chemical reaction occurring in the cell is minimal. The F&P-type cells have, as I said, a design flaw that capitalizes on another oddity of CF electrolysis, namely that the electrolysis gases are not kept separate. When electrolysis is done, normally the gases from the electrodes are routed to separate destinations, such as into glass jars placed over the electrodes as I did in my high school chemistry lab.

That does not exclude that a small number of past studies not in the CF field might have the capability to show similar effects. I wouldn't know as I'm not an electrochemical specialist. However. the fraction of prior electrochemical/calorimetric experimentation that meets these criteria is likely less than one-millionth.

Your citing of the flawed 'group of 10' paper again demonstrates you are emotionally committed to the idea that your heroes are right and anyone who disagrees is a villian, just as interested_observer noted. And no, to repeat, that paper does not address the issues I raise about F&P-type electrolysis/calorimetry.

Quote from maryyugo

An aluminum plate with ONE thermistor per plate? ABSURD!

What you described is a typical Seebeck calorimeter. McK was trying something out. His 98 EPRI report listed results from many calorimeter designs, most just minor tweaks n established designs. (The infamous M4 run that Krivit trashed (as I've been told) was from this same study. One letter = one calorimeter.) The Al plate one was attempting to use the idea that the heat transfer within the Al plates was fast in relation to losses, so the plate would show an even temperature distribution instead of hot and cold spots. As I recall he wasn't as successful as he wanted, and did very little more with it. I'll try to look it up and post some details next week.

Quote from maryyugo

any non-uniformity of temperature within the device is mitigated with circulation of the coolant.

At least that's the idea. Placing a CF cell in a Seebeck calorimeter still has the heat loss pathways concentrated in the gas phase part of the cell, and I assume leads are also nonuniformly distributed inside the calorimeter which can lead to the need I noted with Storms's flow calorimeter, namely that you need to model it with at least two zones instead of it being one homogeneous lump (represented by one equation for the whole cell/calorimeter).

McK did not detect any significant excess heat with the Al plate Seebeck calorimeter as I recall. I'll check that too,

Quote from JedRothwell

Cells with this design have been used since the 1840s. So, if there is an error, it has distorted the literature for 170 years.

Only if there was a secondary reaction occurring only once in awhile, which is the unique chemical feature of a standard F&P experiment. Otherwise there wold be nothing to give the issue away. Hint: This is why systematic errors are so important. They can often go unnoticed for a long time.

Quote from JedRothwell

You are the first to discover this fundamental error. You will be one of the most famous people in electrochemistry if you are right.

Yes, the point of my paper was that this error mode is novel and previously unrecognized for the reasons described above. Will I be famous? Unlikely.

Quote from maryyugo

Are most LENR proponents and experimenters similarly resistant?

Specifically in ref to calorimetry, no. F&P were criticized early on because they used a single point measurement to do calorimetry and that is always potentially susceptible to accidentally being located in a hot (or cold) spot. Many CFers then improved their methods by switching to spatially integrating calorimeters like Seebeck or mass flow types. As well, Storms and Miles made extensive favorable comparative studies of their isoperibolic methods (a single point method) and integrating methods. So, in general the calorimetry done is done well by usual standards. That's why I conclude in my 2010 paper that they are probably at the limits of error (a conclusion specifically ridiculed by Hagelstein in his 2015 "MIT" CF course).

But...they all categorically reject my specific systematic error. However, their reasons fall into 3 classes and all are wrong or irrelevant. First, Szpak and Fleischmann used the idea that they already knew all about at the electrode recombination, but they were not thinking about what I was describing, but about at the electrode *electrochemical* recombination caused by dissolved oxygen, which was limited to <2% in most cases. But my idea was about gas bubbles combining at the electrode and burning which is different and not limited to any particular fraction of available gases. Second, Storms' 2006 Comment on my 2002 paper listed several supposed reasons why my ATER mechanism couldn't work, but I replied to his paper and showed why he was wrong on those points. He chose to ignore my reply and claim he had addressed all my issues in the 2007 book. Further, that didn't address at all the simple mathematical fact that I could zero out a 780 mW excess heat signal with a 3% change in calibration constant. My mechanism might be wrong, but the math isn't, and it needs an explanation of some sort, but the CFers won't even acknowledge I lined out a problem with their method. Third you have the 10 author critique that Jed keeps citing, that misrepresents what I said and then proves the misrepresentation wrong, and then claims to have dealt with my objections. Another extreme form of denial on their part.

Their biggest problem is that they believe the only significant source of variation (error) is the baseline noise, which runs on the order of 25-75 mW (some people claim even better these days). My 2002 paper strongly suggests that is not true, and there is a source of variation that is 10X (maybe larger, depends on the experiment) the full baseline noise.

Quote from maryyugo

Have you (or anyone) calculated this for existing setups

No. I have piddled around attempting to do so but in all other cases I have looked at (which is many) there is insufficient information presented to be able to assess this. In principle, this information exists in their lab notebooks and computer records, but no CFer has been willing to try to evaluate the possibility in their setups. That is however potentially one of the best ways to prove my CCS error is not important to their study. Their unwillingness to do so is another sign of pathological science.

Quote from maryyugo

I only follow your argument generally-- math was never my strong suit (vector analysis did me in in college) and I've forgotten what I learned.

But all I use is algebra and arithmetic (with a little statistics thrown in). Arithmetic to recalculate the excess heat curves. Algebra to explain why the lumped parameter approach all the CFers use is LTA. Stats to calculate the new calibration constants under the assumption of zero excess heat. All pretty basic, no rocket science involved.

I will look at the paper you ref next week...

@MY, I promised a review of the paper you referenced

(http://www.lenr-canr.org/acrobat/ZhangWSconstructi.pdf)

This is a very short paper describing a homemade Seebeck calorimeter’s performance. It actually has a lot of useful information in it, so thanks for pointing this out, as it will serve me well as an example of how ignoring the CCS/AETR concept gives researchers false confidence in their methods.

To begin, let’s look at my primary focal point, their calibration equation and maximum calibration range values. They compute Pout in Watts from their SEC’s measured EMF (E) in Volts as Pout = -.0743 + 5.899 * E + .0017 * E^2

Their calibration data is shown in Figure 1. For simplicity, I will use 4V in the following calculations. For a 4V signal, the Pout is computes to be 23.54 W. In Table II they give error values for the Pex (excess power) as .02-.04 W, which are roughly 10% of the excess power values (or less). However, what if there is a CCS during the runs? Let’s assume a 1% change in the linear term coefficient, and let’s make it an increase, which would give a positive excess, which is what happens when you calibrate with inert electrodes and then get ATER per my proposals. That means we calculate a Pout for that condition of 23.83 W. That is a 0.27W (or 270 mW) excess power. So a 3% change would be 3X that, or 810 mW. Right in line with what Storms observed. And that translates to ~100% of the reported excess powers instead of the 10% error they claim. So, do you think they should be worried about a CCS, just from that fact? I do.

Let’s look a bit further. They also claim to see excess heat in an isoperibolic mode, but to get that they went to significant efforts to ‘tune’ a ‘control’ cell to the same response as the experimental cell. But is that a correct thing to do? I don’t think so. Their control cell is made up with light water and a Pt cathode. They use a different H2SO4 concentration in it, and the position the electrodes at different spacings. Then, to calculate excess power, they use a calibration constant determined from the _control_ cell applied to the experimental cell. That assumes they have made the cells indistinguishable. Unfortunately, I am not convinced this is true. Chemically speaking H is not a good ‘control’ for D. H and D always show significant isotope effects.

As an example, consider the thermal conductivity of H2O vs. D2O (see ref [1]). In ref 1 what is notable is the following: The thermal conductivity is roughly parabolic shaped, falls at different values for H2O vs. D2O, and the maximal values are at different temperature points for H2O vs D2O. That means there is a significant difference at all temps., with D2O being slightly less conductive (~7% at maximum values, which are at different temperatures (slight pressure dependence noted as well). So, does H make a good _thermal_ control for D? I don’t think so. Is the control cell _identical_ to the experimental cell? Not in a chemical sense. So did the isoperibolic calorimetry show true excess heat? Probably not.

So the upshot is that both their methods report excess powers on the order of the noise, *if* one realizes that a slight CCS (1-3%) encompasses the reported signals, and if one realizes that the physical construction of the cells is NOT different from other F&P type cells, meaning that a CCS should be quite possible if the internal heat distribution of their closed cells changes due to the onset of at the electrode (*non-electrochemical*) recombination. Note that there are still power lead and sensor lead penetrations through the thermal boundary of the calorimeter, meaning uncounted heat losses. With an SEC though, it is very difficult to calculate how much of the input heat is actually measured, unlike the mass flow calorimeter case. That’s important to know though since the greater the amount of uncounted heat lost, the bigger the CCS-type error potential.

There are a few more things I could say, but it probably isn’t worth much to do so at this time. The biggest problem is that their SEC gives excess power signals on the order of a 1% CCS, but they do nothing to consider this problem. The date on this pub is 2008, 6 years after my original paper was published, 3 after I responded (in 2005) to the 2004 Szpak, Miles, Mosier-Boss and Fleischmann comments, and 2 after I responded to Storms' 2006 comments. Does it seem like they are 'addressing my concerns' to you??

[1] Thermal Conductivity Coefficients of Water and Heavy Water in the Liquid State up to 370 ^oC

B. Le Nelndre,' P. Bury, R. Tufeu, and 8. Vodar

Journal of Chemical and Engineering Data, Vol. 21, No. 3, 1976, 265

http://pubs.acs.org/doi/pdf/10.1021/je60070a018

Quote from kirkshanahan

He surrounded a CF cell with Al plates and placed one thermocouple (maybe thermistor or such, details are fading...) in each plate (i.e. for a total of 6 T sensors)

Quote from kirkshanahan

I'll try to look it up and post some details next week.

OK, I need to correct myself on this point. In fact the Al plates are a latter part of the experiment that is used to show the Seebeck calorimeter is working as expected. They are not part of the Seebeck calorimeter. The experiments where this is all going on are called the “HH” series.

The 6 plates are in fact what McK calls ‘thermoelectric devices’ and abbreviates as TE’s or TED’s. There is no description given of what they are. He does say a few things about them and his calibration methods…

From the Calorimetry introductory material:

“The Seebeck heat flow calorimeters were based on the measurement of some or all of the output power as a voltage proportional to the heat conducted across a set of thermoelectric devices (TE’s).” (pg. 3-3)

“The temperatures of both the hot and cold faces of the TE’s were measured and used in the function that converts the TE’s voltages to output power.” (pg. 3-3) (This may be because he notes a slight temperature effect.)

From the description of the HH series experiments:

Specifically from the “Calorimeter Calibration” subsection: “The entire Seebeck calorimeter is comprised of the electrolytic cell within six thermoelectric elements, which completely enclose the cell by contiguously forming the six surfaces of an elongated cubic cell. The sum of the voltages generated by the thermoelectric devices reflects the power passing through the TEDs.” (pg 3-32)

From the subsection “The Second Experiment”: "In Figure 3-14, the TED voltages for the top and bottom elements are multiplied by a geometric factor of 3, to account for the smaller area of these surfaces." (pg 3-34)

I *assume* his TEDs are comprised of many thermocouples as one would expect, but there is no actual description to confirm that.

McK calculates excess power (Pxs) via a 3 term equation.

Pxs = Pout – Pin – Pcp where

Pout = K1 (Sum of TED voltages) (no comment on if the 3 multipliers are used or not) (K1 is the cal constant)

Pin = Icell * ( V cell – 1.54) (this assumes no recombination, the cell is open, i.e. gases escape)

Pcp = K2 * the partial derivative of the temperature difference between the cell and bath with respect to time.

“K2 accounts for the heat capacity of the cell and its contents”

K1 and K2 are not given. (Pg. 3-36)

Bottom line: I’m not exactly sure how he does his computations since the coefficients aren’t given.. He notes the baseline noise is supposedly ~5mW, and seems to take this as the error bars on his results. This ignores the basis of a CCS problem. He reports several runs with excess powers as computed above in the 25-75 mW range, i.e. within the usual standard noise levels, but of course he is claiming much better noise levels.

Note: This was work done in 1993-1994, reported on in 1998. I hadn’t published anything on the systematic error found yet. A few months after following my looking at this work is when Ed Storms posted his experimental results that led to my 2002 publication.