How to shoot your credibility in the foot

Simple. Follow Alan Fletcher’s example. He hasn’t done the testing yet, he claims, but when I saw Alan’s announcement of the pool, my immediate reaction was “Ew! WTF?”

Missing in action: self-critique. What could be a problem with this? How could this pool create doubt regarding the work Alan has undertaken?

This obvious lack of self-critique is a prominent feature of the models of Rossi behavior that I use. Rossi seems totally naive about how his work would appear to others. If he is criticized, they are snakes and clowns. A simple desire to verify is full-on grounds for exclusion. Jed Rothwell wanted to bring his own measuring equipment to a demonstration (such as thermometers.) No, visit not allowed. Rothwell at that time was a strong supporter of Rossi. He didn’t take it personally, remained supportive, because he had friends who privately told him they had witnessed impressive tests, and he trusted them.

Fletcher seems to think that nobody could question his honesty. It is not that I’m questioning it, and the most likely source of mistrust would be from Planet Rossi. And I will explain below what I suspect may happen.

So, first, a critique of his explanatory post:

On LENR Forum, Alan Fletcher wrote:

WARNING: all numbers here are from memory. All the data has been posted in the forum … somewhere …

I don’t see any major problem with the numbers, but I haven’t verified them, either. If I have time, I’ll support this post with references to sources. There is one that is important, so I’ll be sure to find it. When we are talking about what is in manufacturer documentation, exact quotes can matter; and what is actually there has been misrepresented by some. The difference between what the documentation actually says and what they “quote” shows interpretive bias. They misread, or at least select from among possible meanings, the meanings that support what they believe, which they then present as what the documentation “says.”

The specification for the Gamma L 0232 refers (variously) to the MINIMUM flow of 32 l/hr at 2 bars back-pressure (presumably when the pump is set to maximum stroke length and rate). The label just says “Dosing Rate”.

MINIMUM is put in capital letters. The fact is that there are multiple mentions of 32 l/hr in Prominent documentation.  (I suspect that I have written 12 l/hr somewhere; if so, nobody has pointed it out. It’s hard to find good help.) There is one document where Prominent calls 32l/hr a “maximum flow rate.”

The [German label] doesn’t say “dosing rate.” It says “dosierleistung.” Smith gives two translations, one in his graphic image label, “dosing capacity or maximum flow,” the other is “dosing rate” or “maximum flow.” The manual generally talks about capacity, and Prominent documents mostly call 32 l/hr a “minimum capacity,” but one refers to it as “maximum flow,” as I recall.

[correction added because Alan shows the English label for his pump. That is translated from the German, because “dosing rate” is not the ordinary English that would be used for maximum flow setting. But it is close enough that someone might just stick with the cognate word. Missed is that the specification is a capacity, not a rate, per se. The pump works across a wide range of rates, according to setting, and the term for nominal maximum rate setting would be capacity — even if actual flow rate can sometimes exceed this to some degree.]

I looked up dosierleisung.  Most translations use “capacity,” but “rate” is used sometimes. This would depend on context, I’m sure. What is the engineer reading the specifications looking for? They want to pump up to X amount of fluid per hour. They want to choose a pump guaranteed to do that. They want to know “capacity,” the most they can depend on being able to pump. The actual rate will depend on settings (and there is also calibration, part of the pump set-up).

If they are using a metering pump, they probably want to know precision as well, and they would not use a pump in a way that drastically reduces precision. If all they want to do is pump a lot of water, they would not use a metering pump, the expense would be unwarranted.

Long ago, I thought of the early E-cats as needing a float valve that would regulate level, otherwise, with fixed input flow (as Rossi was setting), there would either be too little or too much flow. In the former case, the reactor would run dry and perhaps overheat; in the latter, it would flood and overflow, and if this were not carefully monitored, this could create apparent XP. Is that a feature or a bug?

Rossi’s “engineering” is literally half-assed.

(He has a high “get ‘er done” intensity, which has impressed many, but the execution may be incomplete. Petrol Dragon, anyone?)

As mentioned, “capacity” implies maximum, so for Smith to treat this as a maximum was not stupid, though it was not fully precise. The value he used was still — almost certainly, but we will find out — a rough maximum, a design floor that would not be greatly exceeded at maximum pump setting. Alan is consistently using the interpretations and translations and erroneous quotations of Planet Rossi, betraying a strong bias. Some of these are subtle, some are not.

The specification indicates that an individual pump may deliver between -5% and +10% of nominal, and that the “repeatability” between runs is 2%. (So the range for ONE run is -7% to +12%)

However, it delivers more at lower back pressure. There is a warning (hidden deep in the documentation where even expert witnesses can’t find it) that at VERY (undefined) low back pressures the pump can deliver “several times” more than the specification.

It is not in the basic pump documentation, it is in a brochure that is about dosing pumps in general. It does not state what is claimed. The difference may be crucial, and if Alan doesn’t know this, he has not carefully followed discussions on ECW or this blog. I covered this pump issue at Pumped up or stupid mistake. I could not find the original Prominent document, but there is an image on E-Catworld.

“When metering at atmospheric pressure the pump can achieve several times the stated feed rate.”

Consistently with what he may believe, Alan has substituted “the specification” for “the stated feed rate.” Almost certainly, the document means “the reading displayed on the pump,” whereas “specification” would refer to some value given in the manual (and not on that page, making this statement vaguer if it means what Planet Rossi thinks.)

This is an error mode, the pump is not designed to be accurate at atmospheric pressure (not carefully defined here, but it would probably mean zero back pressure), not to mention forward pressure. The error mode is possible because fluid may leak past the valves in the pump, increasing flow over what is created by the action of the solenoid. It is unlikely, however, that at atmospheric pressure this leakage is more than a relatively small fraction of the volume being pumped, at maximum stroke length and stroke rate. The warning in that document would be referring to smaller settings, almost certainly.

This whole pile of meshegas by Rossi and friends is based on not looking at the word “capacity.” Capacity implies a maximum. Not an absolute maximum, and when the manual says minimum capacity, it is saying that the capacity, the maximum flow rate, is at least the stated figure, so one may rely on the pump to deliver that volume at that pressure.

Smith’s mention of this in his supplemental expert report is then not entirely and fully accurate, because most pumps would be able to deliver a little more than the specification, but the substance likely remains. It is not the face-palm stupidity that Planet Rossi claims it to be.

Rossi/Anonymous posters claim that the flow is much higher :

They do. Rossi gives a specific number, as Alan correctly reports:

Mats Lewan interview

Mats Lewan : By the way, since you know these pumps, what capacity do they have at the actual pressure in the plant—about 0.2 bars?

Rossi : “About 75 l/h.”

Blog: 60l/hr

I covered the first blog post in an update on OMG! Good news! There is another, but it merely says to Rossi “you were right.” So the correct figure for “blog” is 68 l/hr.

Rossi has also indicated that there was a recirculating pump into a pressurized return tank ( (I think – On JONP?) .

That’s a huge can of worms. Rossi introduced, quite late, alleged features of the system, possibly too late to ask him much in depositions about them. A recirculating pump, like any pump in the “customer area,” could create at least two possible fraud modes (or even accidental failure, though that’s more difficult). A possible generic error, affecting many Rossi “demonstrations,” would be over-pumping, causing overflow water. Error in pressure measurement could then make his reported steam temperature represent saturated steam over flowing water (or merely very hot water). If the inoperative reactors did not have disconnected and clearly shut-off flow, but merely had pump settings reduced to zero, they would flood water into the steam system. Not having a steam trap — Rossi removed one — would allow this to escape unnoticed. As well, a return pump could introduce air into the system, which could then confuse the return flow meter, as has been discussed ad nauseum.

It’s a total mess, and, again, with cleverness or naivete, Rossi removed all the evidence that could be used to study possibilities, as if nobody could possibly question the report of an effing nuclear engineer! Who do they think they are? (That’s a standard ele argument.)

The initial setup has:

Stroke and rate set to maximum

Suction from a bucket, between 42cm and 62cm below the pump center-line.

The suction tube is about 5 feet long, 1/2 inch inner diameter.

The discharge tube is about 7 feet long, 1/2 inch inner diameter, but with a more restrictive “Tee” to the manometer.
Discharge level with the center-line — effectively 0 bar.

Learn to write “barg,” Alan. Good for the soul. But we know what you mean. The conditions look decent. This does not:

So, I’m taking bets (in yankee dollars, or in quatloos) what the FIRST timed run will show.

Each bettor will make a prediction in l/hr L, and bet in $ or quatloos. Those betting in dollars will automatically be entered for quatloos too.

Each bettor will PAY $1 or Q1 for each Liter between his/her/its prediction and the actual result : this will go into a “pot”.

The WINNER (closest to the actual result) will receive HALF the total pot. The project will receive the other half.

In the case of tie[s] the HALF share of the pot will be split equally between the co-winners.

The Dollar results/winner and Quatloo results/winner will calculated separately.

Okay, so imagine that the result comes out something like 40 l/hr. A winner is declared. Planet Rossi doesn’t accept the results and suspects that Alan gave the winner the data before announcing the results, so he controlled who would win and split the proceeds with the winner, getting, then, 75% of them. For a scientist to run a pool on the results of his work would obviously be considered unethical. It could create interpretive bias.

Alan will have some choices: he said he will announce his “first results,” but first results often incorporate errors that are then found. That is why scientists don’t generally publish immediately, until they have time to go over everything. In open public science, that’s bypassed, but, still, there should be time for critique and further study before definitive results are released. “We have this result, but we have not yet checked everything thoroughly.”

I consider this scenario (dirty pool) extremely unlikely, don’t mistake this for an allegation of impropriety on Alan’s part. Only naivete. By announcing a pool, he is damaging his credibility as a neutral observer, in favor of encouraging more bar-room posturing.

By the way, pumps could vary greatly. Over-pumping with metering pumps is essentially a failure mode and the exact failure behavior would not be guaranteed to be the same for all pumps. Further, wear in the pump could cause overflow to be much higher for some pumps. An interesting possibility, mentioned above, is flow at zero pump setting, and forward pressure. At relatively low forward pressure, this would be easy to test.

What if all those inactive reactors were “turned off” by shutting off input heating and setting the pumps for zero flow, but the return system were pressurized, as Rossi might now be claiming? Those reactors could overflow. Yes, Rossi could claim that valves were shut to prevent this. Did those valves exist? Rossi can, in fact, claim almost anything, but all this points out that actual test conditions were not clear and not necessarily visible, and that Rossi excluded the IH Engineer, Murray, in July, 2015, practically guaranteed that IH could not accept the Doral results, given that they also claim they were unable to independently confirm Rossi claims in their own testing.

All this to create a “1 MW test” that was quite a bad idea, not what an engineer would want, when the individual reactors have not been thoroughly tested and characterized and reliability determined. Cart before the horse.

Andrea Rossi.

Rossi is now claiming to target sigma five. Without a clear definition of device performance, it’s meaningless. However, if there is a clear specification for pass/fail performance, something like “generated power of X at COP Y, for period Z,” then testing uniformly manufactured items can come up with a sigma level. There is no clue that Rossi has made anywhere near enough devices to show a meaningful sigma level. Sigma five is 233 defects per million opportunities. The QX is claimed to be a 20 watt device. If 20 watts is the minimum power, a 2 KW home reactor would have 100 devices. With sigma five, it is unlikely that any of these would fail in the specified lifetime. Failure in the control systems might be more likely. Rossi’s existing systems, from his past history, may have been below sigma 1, but might be thought of as sigma 2. Sigma 2 could actually be commercially useful, if that’s a reliable estimate. Roughly two out of three devices would work as specified. So redundancy would be built into a system, allowing a relatively high level of failure with overall acceptable system performance.

Whether or not this would be practical would depend on details. Where do the failures occur? If there are insertable fuel pods and it is those which fail, and if the application allows identifying which pods are failing, sigma 2 would be good enough, only a minor nuisance, requiring engineering setup. But if the failures are delayed, not so easy. In that case, one over-engineers the application, with redundant devices, so that many can fail; as a full system approaches minimum power generation, it would then be repaired.

Rossi’s tests for IH, the Verification Test and the alleged GPT, did not look for or consider device reliability data. It was ignored entirely, yet another sign of what an abortion this was.

Additional Notes

I hope Fletcher knows how to read a graduated cylinder properly. (Many will, it’s part of a basic science education, or used to be. It’s not necessarily obvious. The water level is the bottom of the meniscus, viewed from a position normal to the surface of the cylinder.)

Fletcher has also placed his own bet.

The issue about pressure is interesting. What is the relevant pressure? Static pressure when the pump is not operating (between strokes), or transient pressure during actual flow? I can see arguments either way.