I finally have a simple simulation model of the Mizuno reactor which has excellent correlation to the data supplied in the excel files that Jed published in his report. I used the program to monitor all of the days he supplied data for to verify the simulation accuracy. I also constructed a separate test simulation model in order to uncover the interactions of various sections of the model. With that tool I was able to prove that an ideal system matched my expectations.
At first I was over enthusiastic about the power being generated and as time passed I found myself lowering the expectation as my model became more representative of the real life behavior of the system. I used information derived by Gigi to obtain effective values for the pump leakage and the time constant of the outer system. Gigi has not responded to my requests for additional proof to his contentions so those assumptions remain open to question. I did review his derivation and it appears excellent in my opinion. A second, inner time constant was required in order to make the simulated response match the real measurements to a good degree. We were given a big hint as to the value of this factor by the posted graphs that Jed included within his report. That constant is what establishes the sharp transients that occur with each pulse. The inner time constant is approximately 1200 seconds while the outer one is 21000 seconds. My model was carefully adjusted in an attempt to determine the amount of power being produced by the LENR reaction assumed to be taking place due to heating by the 20 watt drive pulses. I played around with the power level and pump leakage tradeoff process as I attempted to achieve the best time domain fit to the original data. Each of these factors leaves a different footprint upon the response. Power pulses are mostly seen at the beginning of the period and dissipate after the inner time constant filter works upon them. These power pulses end up depositing a fixed amount of joules into the system thermal capacitance. The leakage due to pump imperfection, etc. continues throughout the entire period of the simulation and is visible even when the pulse nature of the power pulses has smoothed out. By far the most influential detriment to the time domain behavior of the model is due to ambient initial conditions as well as its variation during the simulation. fortunately, it is possible to take those changes into consideration with a good model. I feel that I am capable of locating the true tiny power pulse hidden within these very large noise generators. I took Jed's data from October 20, 2014 to carefully match the time domain response of the water temperature he measured to that calculated by my simulation. This was just one of three main days that he published, but I gave it my most exacting effort to seek the power pulse accurately. On that day just one pulse was applied which made the fit a little easier than when working with two or three pulses although the others make a good match as well. The results suggests the application of an 8 watt excess power pulse that is time synchronized with the 20 watt heating power pulse. I could easily determine the difference between this level of power and an assumed power of 10 watts. On the lower side it is more difficult since the thermal joules added to the thermal capacitance of the system becomes quite small. I feel that the excess power must be at least 5 watts, but it is difficult to be confident at the low power levels. This excess power may be less than everyone hoped for since it represents an increase of only 40% above the 20 watt input. The accuracy of the simulation is quite remarkable. The simulated error reaches a peak of +.07 to -.085 degrees C over the entire time period. The average is tighter, having an average of .04 degrees C. Measurement noise is the dominate excursion seen, but there remains hidden within that noise a signal component. My assumption is that this is mainly due to the excess power pulse being slightly delayed from the drive power pulse as the heater comes to temperature. Also, there likely remains one or two additional time constants that impact the system response. These would be difficult to chase down and I feel that the model works well enough now. Unless someone is very interested in obtaining a copy of my simulation file I plan to put it aside after this difficult exercise. It currently is not in a form that I would want to have distributed widely since it lacks good formatting. Besides, the need for this tool vanishes if Mizuno's system does not continue to be tested. The main reason I pursued this goal was to prove to myself that his system produces excess power and for the challenge. Dave
