Rossi Lugano/early demo's revisited. (technical)

LDM,

Do you suppose it is the hollow area in the center of the Lugano device that speeds up the settling time (less mass in the hardest place to heat)?

I can't see the ribs helping, but instead hindering the settling time a bit.

Also, how hard is it to change the emissivity settings for your simulation?

LDM,

Why does your simulation show a cooler inside than the middle of the outside?

LDM,

I may have been confused as to where edge of the cylinder was in the slice image. So the lines are the coil? The core temperature still seems low compared to the real thing, but no need to split hairs at the moment.

The emissivity question is about something else, which is best tackled after the simulation behaves as good as is reasonable.

Quote from LDM

What is your measured temperature data ?

What is your difference between inside and outside ?

And where is your internal thermocouple located ? In the center ?

At 736 C outside, the core temperature is about 812 C, so about 80 C difference at that temperature.

The internal thermocouple is right in the center of the cylinder, both of the diameter and length.

The external thermocouple is cast into the outside surface, with the tip just visible in the casting.

Quote from LDM

Also, is your cast resting on the floor (tiles) or is there room between the tiles and the cast so that no heat is leaking away by contact ?

In any case it is preferred to have the cast at at least such a height above the floor that a normal convective heat flow can develop and that almost no radiated heat is reflected from the floor upwards again.

The casting is supported by the Kanthal A1 coil leads, which are doubled over and twisted so they are both strong enough to support the Cylinder and not get as hot as the coil itself, protecting the electrical connections. So the Cylinder is held about 6 cm above the fire bricks, supported only by thin wire (24 AWG x2) connected to the 14 AWG solid copper AC wiring by push connectors. I have considered removing one brick below the Cylinder so that reflected heat is not a problem.

Quote from LDM

Did you keep track of the room temperature and if it changed in that time frame compensated for it ?

Yes, I always have an ambient temperature thermocouple going. It rarely changes more than a degree from start to finish, unless experiments run for many hours.

There is an image of the Cylinder at 1107 C external T not too many posts back that shows the ambient sensor and has a description of it, It has radiant heat blocked by a fire brick.

Almost all experiments have a complete data log of all three thermocouples, and the thermocouple timestamps are used for all notes. (Since the cool down to ambient takes so long, I have been stopping the data logs around 150-200 C, since that is where the inside and outside temperatures tend to become the same.)

The real emissivity at 735 C is between 0.94 and 0.95 . It is probably 0.945, but the pyrometer only has two decimal places. I have calibrated the pyrometer from about 50 C all the way to 1107 C against the external thermocouple that is cast directly into the surface of the Cylinder.

.

.

.

.

.

.

Extra details of construction:

.

.

LDM ,

The Cylinder is 6.6 cm long. The coil ends 2.5 mm from the ends of the cylinder.

One end runs a bit cooler than the other; the Durapot seems to have separated from the coil slightly for a few wraps on one side (1/3 of the diameter) of the cooler end, and it is hotter on the opposite side in the same place. I have been measuring the cooler part. The excess on one side seems to balance out the cooler part, and the coil hasn’t burned out. This means I should be a little short when calculating radiant and convective power overall. So far it has been working out fairly close to COP 1. The cooler end I have also made 1.1 cm, instead of 1 cm for calculations.

The segments I have been using are (in order) : 1.1, 1.0, 2.5, 1.0, and 1.0 cm. The second last segment is almost always the same or slightly (1-5 C) higher than the 2.5 cm middle segment. The one end has been cooler right from day one.

The cylinder end faces (circle area) are complicated due to a lot of temperature variation from the center to the outside rim. For simplicity I use a flat plate model with the same area as the circle for calculating power, and use the respective end segment temperature and emissivity as the end face values. Seems to be a reasonable compromise, and actually quite close to IR pyrometer values. It was a complete nuisance to back the pyrometer over 1 m away to get a good average temperature over the appropriate focal area and by aiming precisely at the cylinder ends, but I did it enough times to convince myself that the above cheat method was close enough.

There are plenty of ways to improve the equipment and methods further, but that is beyond the original scope of the project. I welcome anyone else to make a replication in kind, and improve on whatever they want.

I am actually surprised at how much abuse this Cylinder has taken so far. It is still hooked up, so if there are any particular temperatures (within reason) or something that you would like to know or be tested, I can test them out for you.

The poured Durapot is certainly not as dense as manufactured extruded alumina (like the Rods in Lugano).

LDM ,

Try this and see how it compares: (I haven't added all the 0.39 E data to the calculation sheet, just the final one). I used The Thermal Wizard for radiation/convection calculations rather than build a giant spreadsheet to handle all those details. I just dragged and dropped the results into the respective cells after inputting the necessary details. (It takes a little practice to grab the values without lopping off a digit, since a cut-paste doesn't work with the web page.)

The coil is 1.9 cm OD.

Bumping this over here...

.

Quote from THHuxleynew

Paradigmnoia:

As always, thanks for your terrier-like pursuit of technical data. I was not aware of that image: you are right, it is revealing.

In the interests of accuracy I'm going to disagree with part of what you say, though not the overall conclusion.

If that picture is genuinely one of the December test then I think we can definitively answer Jed's point about not knowing anything wrong with that test, and even say what the issue was.

Those, like Alan, who believe Rossi's stuff sometimes work might perhaps consider that Rossi's innacuracy in doing experiments is so high that he could never possibly know himself the difference between working and not working. There is juts no evidence out there his stuff is more than parlour tricks of a specifically scientific variety.

The Ferrara tests where document as two significant test runs: December and March. They are different (in lots of ways). Like any magician, Rossi has a habit of ringing the changes.

In the December test run the report says:

So, from the report: the entire calculation comes from the (integrated power) WH measurement from the PCE-830. This is then divided by 96 (the test time in hours) to get the inferred power consumption.

They know power was output at the same rate throughout the experiment because the output was shown there the whole time from the 1s per frame time lapse photography.

This test has TCs for temperature so I'm not inclined to think there are big errors on the output side.

You might think that because (to get a constant output) a constant input is needed, and anyway if not constant the total energy in vs total energy out is all that matters, this instrumentation is pretty good. That, I'm sure, is what Jed thought.

The problem is that those checking this experiment were not experts in AC power measurement - or if they were trusted Rossi to have a vaguely sane setup.

The photo shows Rossi's setup to be not vaguely sane. The negative PFSigma and PWSigma show that at the time this photo was taken the power was drawn from two phases with reversed clamps (it would not be reasonable for the control box in reality to source power back into the mains supply).

The positive WH indication shows that for some of the test power must go through a phase with a correct polarity clamp.

The report notes that the control box switches power in some proprietary (= don't ask questions about it) way, but that the total power remains the same. That may be true, but because of the partly reversed clamps that photo is absolute proof that the WH indication (the only one used by the experiment checkers) will read lower than expected.

So we have positively identified a Rossi-esque false positive for this (December Ferrara) experiment.

Some details (a little speculative, but I think the logic is solid):

No-one needs to change the clamps. In the photo phase 31 has a maximum real power of < 38W (however measured + or -). Phases 12 & 23 have a negative real power (both with reversed clamps) because the total negative power can only be obtained at the stated total negative PF if both phases contribute. Note that the average power-weighted PF over all phases is -0.48, but the max (absolute value) PF for any phase is -0.52, which means both of the high actual power PFs must be close to -0.5, and therefore also the 31 phase PF must be positive (clamp the correct way round).

The control box can deliver a constant large power with a measured much lower WH energy, by switching between power through 12 & 23 (as here) and power through 31. All that is needed to get the stated 360W average is for switching to have the correct duty cycle to make the positive 31 contribution slightly more than the negative 12 & 32. Note that the voltages are weird: this is not mains line voltage: I guess measurements were done on the black box output.

Speculative notes on the March Test

The March test was different in quite a number of ways, and (typically for Rossi) exhibited a much lower COP. My guess (very speculative) is that the absent testers (only Levi was present) noticed some anomalies in the power measurements, or at least raised questions of a possible loophole. Therefore it was decided to redo the test closing those loopholes. (Wisely, given we have positive evidence of such an error). From the March test part of the Ferrara report (linked above):

The clamp ammeters were connected upstream from the control box to ensure the trustworthiness of the measurements performed, and to produce a nonfalsifiable document (the video recording) of the measurements themselves.

Which opens the possibility that the voltage measurement and current measurements in the December test were from different sides of the control box? They clearly did not trust the readings.

What is notable is the COPs obtained:

December:

Pout = 2034, Pin = 360, COP=5.6

Note that a Pin of 2034 would lead to COP=1, and this is plausible given the QSigma of 1.96kVAR (we cannot trust power calculation from phases and therefore also cannot trust the stated PF).

March:

Weird switched system, with duty cycle of 35%.

COP=2.9

Interesting that if the system were not switched the COP would be 1. Not enough data on this system however to be clear what is the false positive mechanism. Notice however that for these tests, where there was at least some level of external supervision, Rossi completely changed his setup - moving to a more complex system - more difficult to analyse - but with lower performance, for his repeat test. That is very suspicious!

THH

Display More

Here is the source of the photo. Taken by the author of the story during a visit on Dec 14 2012. (small PDF)

http://www.borderlands.de/net_pdf/NET0113S13-15.pdf

I think I now have an idea of what is going on, intuitively. It will be complicated to demonstrate, but it should deal with the funky early ecat power.

1) There are four wires hooked up to the pre-flange type ecat.

2) Italy wiring code for 3 phase is:

Ground.........Yellow-green stripes

Neutral.........Blue

Phase1..........Brown

Phase2..........Black

Phase3..........Grey

.

.

.

.

.

The resistors are wired in series (apparently) but hooked up with 3 phases and a neutral. The meter is hooked up as though it is 3 phase 3 wire. Now think about that arrangement with probably the blue neutral hooked to a series connection (double wires into a connection block from the cylinder end). Imagine short phase angle control pulses for 360 W and 810 W longer pulses of 3 phase AC being fed into that arrangement. The current flow patterns in those cases I think could be quite complex. (I'll draw a little picture in a bit.)

The November test with the hot spot was hooked up the same.

.

.

Maybe like this? (It is possible the phases could be mixed up.)

I'm not sure how to attempt to model the current flow for this with triacs. Ugly no doubt.

.

What Lugano should have looked like:

.

I pulled a brick out from below the Cylinder to make convection work freely, and picked up another true RMS meter to monitor current downstream of the controller at the same time as the voltage. That should allow the controller power consumption to be measured properly and give better measurements of the Cylinder power consumption. I am looking for one of those cable splitter blocks for the outlet to double-check the Kill-A-Watt measurements. (I might have the parts to make one).

While repeating several power steps as used previously and checking current with the new Fluke meter, the Cylinder coil failed on one end while starting measurement of the 400 W step. After determining for a few seconds that the thermocouples were not shorting out, I turned the voltage to maximum to see if a Nernst glow could be achieved. The coil failure area glowed for an extended time, but it did not seem that it was being increased by the extra voltage. Resistance of the alumina was measured through the broken coil while still 1000 C and found to be several thousand ohms, rapidly increasing as it cooled, to the megaohm range before becoming too much to measure.

The Kill-A-Watt was found to be reporting only about 0.1 A higher than the Fluke meter reported at low output voltages (35-50 V AC) and the gap decreased as the voltage increased. Voltage was essentially the same. The Kill-A-Watt meter is only used on the electrical outlet to power controller connection (pure sine). So for total input power, the KAW unit seems quite accurate.