Bob Higgins of Motorola Solutions did an analysis which he sent to Mats Lewans, who copied it to me and a couple of others on Vortex.

Hello Mr. Lewan,

I am enclosing my spreadsheet analysis of the data you published for Ing. Rossi’s October 6 test at U. of Bologna of his E-cat. In this analysis, I tried to incorporate reported items such as water leakage, heat loss through the insulation, difference in source water temperature and the water temperature of Tin, and the final energy stored in the E-cat when the experiment was terminated. If you have the occasion to look through this analysis, I would appreciate hearing of anything you find that may not be consistent with your observation as a first hand witness to the test. You are welcome to share this with others that may also be able to evaluate the analysis for missing features or wrong calculations.

We considered the possibility of contamination of the Tout by the hot water/steam of the heat exchanger primary input. However, because the secondary water was flowing up out of the secondary outlet and out of the brass header, and the contamination primary heat would have to pass this water to reach the thermocouple, and because the flow rate was high in the secondary, heat from the primary inlet would quickly be diverted into the secondary outlet water. Thus, very little of this contamination heat would make it to the thermocouple and cause temperature error - we regarded it as a possible minor second order error. If you draw a cross-section picture of this pipe and the flowing water, you can see how possible contamination heat from the primary inlet would likely terminate in the secondary outlet water long before reaching the thermocouple.

I would agree with everyone else that there was much that could have been improved the experiment, but the real point now it to understand the data we have and determine what information that can be derived from it with confidence.

It is interesting now how the skeptical criticism on the net seems to be switching from “doesn’t work at all” to “doesn’t work with acceptable commercial COP”. Are the skeptics now convinced there was large scale excess energy? In and of itself, this is a physics shattering breakthrough. It is clear from the data that the COP would have been much higher if the test had been run for a longer period. I am personally excited by the results and data from the experiment.

Thanks to Ing. Rossi for hosting the experiment – he was under obligation to no one to do the experiment - and to you for reporting the data.

Bob Higgins
Motorola Solutions

I suggested he subscribe to vortex, but meanwhile ...

He sent a very large spreadsheet, with a couple of interesting diagrams / plots.

I've put some of them (with permission) in my initial draft report

I extracted and annotated a couple of pictures from his spreadsheet :

a) a very nice schematic diagram of the fat-cat.
(He shows a pressure regulator at the outlet ... I don't know if this is a guess or new information!)

b) Another data plot and comments

Initially, heat is stored in the E-cat as it is filled with water and the water is being heated. This energy is accounted a lumped addition at the end. When E-cat fills at about 173 min into the experiment, steam is not yet formed and liquid spills into heat exchanger and there is measurable heat exchanger output. Some of the stored energy in E-cat is being quickly lost into the heat exchanger as the liquid water carries out heat
quickly.  This causes the spike seen just after 173 min.
At 185 minutes, steam begins to form (crosses 100C) and by 200 minutes the steam is 110C which corresponds to about .6 bar pressure (AJF : over atmospheric -- 1.6 total). At 220 minutes, the steam reaches about 120C which is about 1 bar of pressure (over ambient) and the output is probably mostly steam. At 350 minutes, the steam is down to about 116C which may make it oscillate in the pressure valve: valve closes, pressure builds up, valve opens and outputs a burst of steam, valve closes. The data is too coarse to show this possible temperature/pressure oscillation. The oscillation probably gets averaged out in the heat exchanger heat measurement. Probably needs and output measurement that
also integrates all of the heat output in one big lump as a check.
If oscillation occurs due to a pressurization valve, it could cause the temperature measurement in the reactor (green curve above) to be wrong. There could be intermittent measurements of steam and water and the steam temperature could be oscillating.

c) Overall energy balance
Test of Rossi 27kW Reactor No 2  (1 of 3 reactor cells activated)
Date: 6-Oct-11
Location: University of Bologna, Italy
Analysis: Bob Higgins
Note: See supplemental analysis at bottom of data for additional energy balance items.
Heat exchanger efficiency = 0.98 [Rossi estimate]
E-cat Insulation Value (R) = 6.0 W/(K*m^2) [Higgins, estimate]
Est. E-cat surface area = 1.4 m^2           [Higgins, estimate]
Reactor flow rate, Fp = 13.0 liters / h [Lewan, est.; changes with time, var not used]
Reactor water leakage, FL = 2.0 liters / h at 2 bar internal     [Lewan, est.]
Exchanger flow rate, FE = 640.8 liters / h     [Lewan, measured]
E-cat total heat output = 33.76 kWH  = 121.5 MJ
Total electrical input energy = 8.97 kWH  = 32.3 MJ

(That didn't copy very well ...)

d) Residual Energy in Tank

Final Temp of water in E-cat = 97.7 C
Average Temp of input water to E-cat = 25.5 C
Temp difference of remaining water = 71.7 C
Est. volume of water in E-Cat at end = 20.0 liters remaining (est. liquid volume remaining, some internal volume may be steam.)
Heat stored in E-Cat water = 1.667 kWH heat remaining in water at end of data
Weight of E-Cat =  98 kg
Average specific heat of E-Cat (est.) = 0.07 (presume an average of steel and lead) Est. Heat stored in E-Cat structure = 0.572 kWh heat remaining in E-Cat structure at end of data

Reply via email to