Thanks for the detailed answer.
Harry
On Tue, Aug 14, 2012 at 6:26 AM, Jed Rothwell <jedrothw...@gmail.com> wrote:
> Harry Veeder <hveeder...@gmail.com> wrote:
>
>> how does he determine the ouput ?
>
>
> Briefly: The abstract says that at the lab they have a precision flow
> calorimeter. Here they are using crude thermometry in an uncontrolled
> environment. That is, a room full of people crowded around the machine, with
> currents of air and so on. Not a constant temperature incubator. That is
> obviously inaccurate but you cannot transport a flow calorimeter.
>
> The cell is equipped with two wires. One for calibration which I think is
> nichrome. It is gray, anyway. The other, working wire is constantan (Isotan
> 44) treated by Celani to be a lot more porous and absorbent.
>
> By "thermometry" I mean they turn on the 48 W heater or the working wire and
> watch the temperature stabilize at 120 deg C. That is the minimum
> temperature below which this material will not load, and no effect can be
> seen. This is straight DC power coming from a high quality power supply. As
> you would expect when there is no excess the temperature is very stable. The
> temperature stabilizes for a while even with the working wire. This morning
> it was flat. No indication of excess heat. When excess heat begins it
> fluctuates considerably, climbing and falling, from one minute to the next.
>
> With this kind of gas calorimeter, the increase in temperature is
> proportional to the excess heat, although not linear. When I did similar
> calorimetry years ago with Mizuno I found the response was stable,
> repeatable and predictable, and the fact that it is not linear is
> unimportant. (With something like LabView you can just tell it to be linear
> anyway. Throw in a fudge factor, or probably nowadays tell it to figure out
> the fudge factor.)
>
> Rob Duncan told me that the major problem with this arrangement would be
> changes in heat loss because of changes in convection. Convection dominates.
> If anything, he expects convection would increase as the gas moves faster,
> and this would lower the temperature.
>
> There is one thing that might raise the temperature slightly. The cell has a
> leak. It is initially pressurized to 20 atm. It loses 1 atm over 8 hours.
> That could not explain the anomalous temperature increase for two reasons:
>
> 1. The temperature rise happens too soon.
>
> 2. A leak is probably fairly steady, causing a steady, linear increase in
> temperature. It would never decrease. It would not fluctuate rapidly.
>
> When they brought the cell to Texas it had a variety of different instrument
> types attached, with LabView software written by various physicists and
> other non-experts. The people at NI looked at it -- actually, Truchard, the
> president and CEO himself looked at it, I gather -- and said "let's get rid
> of everything but the cell." They replaced all instruments, computers, the
> interface box etc.; they put in the latest version of LabView and rewrote
> the code. So now it is as good as any instrument I have ever seen. It looks
> like a product brochure illustration. Except the method is still crude. At
> one point Truchard said, "what this needs is an IR sensor for the surface
> temperature." He jumped in his car, drove to an electronics store and came
> back with a handheld IR sensor. He said: "This was on sale. I got a great
> deal on it!" The IR sensor is sitting on the table. That's the way the NI
> engineer told me the story, anyway. They say it is typical of Truchard.
>
> Input power is steady at 48 W both in Texas and here. Anomalous output was 5
> W and climbing when I last saw it. In Texas it peaked at 21 W. I think
> Celani said that is a typical result. In other words, 48 + 21 = 69 W. I
> think that even crude thermometry should be adequate to measure a difference
> as large as this.
>
> I would call this a trade-show demonstration. That is, not something
> perfectly convincing in itself, but something that gives you feel for what
> the product is like. I doubt that the ENEA labs are incapable of measuring
> the difference between 48 and 69 W.
>
> - Jed
>
