MFMP: Automated experiment with Ni-LiAlH

  • The past 2 days I have been manually evaluating the resonant frequency of the coil vs. temperature and what temperature the coil settles into for a given core temperature. Currently I am regulating at 1200°C and the field coil is settling in at about 178°C. The epoxy on the coil is only rated to 150°C, but I am not seeing any smoking or softening of the epoxy (of significance). I have relocated the audio PA about 2' farther away from the coil with a longer lead to the field coil. No oscillations have been seen since this change. The resonant frequency of the L-C appears to go lower with temperature, being about 3930 Hz at room temperature and about 3840 Hz with the core at 1200°C. It is more likely the coil is changing frequency with temperature based on the coil temperature going from room temp to 178°C.


    Because I have the opportunity to play a different .wav file for stimulation every time it is applied, I am going to create a number of .wav files at different frequencies and choose which to use based on the temperature. Also, the resistance of the coil is changing. At room temperature, a full gain setting produces about 6.3A peak sine wave current while at 1200°C core (178°C coil), the same drive only produces 5A sine wave peaks. So, I will also change the stimulation level based on the temperature.


    The coil current sensor has proven to be really useful. I can change the frequency of the input at each manually set temperature and adjust to maximize the coil current to find the resonance. It also lets me see when something is changing or if there is an oscillation.

  • BobHiggins

    If you're interested there's a command line utility which can generate and play sound like for example a sinewave. If this could be somehow integrated in your reactor script there would not be need anymore to have many different wave files.


    http://sox.sourceforge.net/

    http://montessorimuddle.org/20…nd-saving-tones-with-sox/


    EDIT: it's not immediately clear, but the program archive (I suggest downloading the zip version, not the exe installer) does not come bundled with a play.exe file to use it as suggested in the second link. You have to copy sox.exe into a new file and rename it into play.exe (or rename sox.exe). The program changes behavior depending on its filename.

  • Since yesterday I completed testing of the system to a core temperature of 1200°C and the coil didn't melt or smoke, I am starting a comprehensive calibration run tonight. This will reach 1200°C tomorrow (while I am awake and can still monitor things) and will begin cool-down through tomorrow night. There should be no oscillations in this run. I am using 3 different .wav files for the excitation, each having different frequencies to compensate for drift in the resonant frequency with temperature. I am also taking a stab a equalizing the amplitude of the stimulation with temperature. Additionally, I have chosen a slightly different waveform than just plain sine: 3s of sine followed by 2s of nothing - repeated 6 times. This will be sort of a hammering:




    I hate to say it but I have change the link to ALL of the experiments. I put all experiment folders under a master folder called "Experiments" that is shared. Each of the experiment links I had before are still on Google Drive under this master folder, including this one. For this latest calibration folder, look in the folder for "Experiment_Wav_20170521" for "PreCalibration_20170525". Here is the link to the top level folder:


    https://drive.google.com/drive…FkMVNaSEtaNWM?usp=sharing

  • I have chosen a slightly different waveform than just plain sine: 3s of sine followed by 2s of nothing - repeated 6 times. This will be sort of a hammering:


    Actually Bob, this is somewhat similar to the effect you get to using a PID thermostatic controller with an SCR coile driver. PID's run on a 2 second loop, and one of my colleagues at 'www.lookingforheat' is of the opinion that this is a trigger mechanism- 'waves of heat and magnetic fields'.

  • The full automated calibration completed normally in the night. This run was timed to have the hottest temperatures while I was still awake to monitor. No oscillation problems were seen with the audio PA. The data is in the Google drive in the experiments folder:


    https://drive.google.com/open?…Pc25a4cOM2bkFkMVNaSEtaNWM


    This morning I will analyze the calibration data, disassemble the reactor, and prepare for fuel loading.

  • This is the usual graph related to this run. Since the steps were not uniform in duration I had to do some modifications to make the usual script work. Hopefully future calibrations will be done with a constant step duration (also taking into account the length of the waveform).



    The coefficients for the 3rd degree polyfit are:

    Code
    1. 1.04566291e-08, 3.82707327e-05, 9.86894597e-02, -2.42502509e+00



    BobHiggins

    The waveform you posted above could be generated in real-time with the utility I previously suggested using the following command:


    Code
    1. play -n synth 3.0 sin 4000 pad 0.0 2.0 repeat 6 gain -6


    An obvious advantage is that you could control the parameters (i.e. the command line arguments) dynamically depending on reactor conditions.

  • can

    Thanks once again. I have written a program that automatically searches and dynamically finds the boundaries where the voltage steps, and then takes samples backward from each of those boundaries as settled temperature points for the calibration. Instead of averaging each of these data points to get a settled temperature, I take all of the settled data points into the fit. I also create an ambient set of sample points with P=0 to add to the data to be fit. The program spits out the 3rd order polynomial fit as:


    P(watts) = 9.09461527238551e-009*T^3 + 4.10041913857542e-005*T^2 + 0.097225895042357*T - 2.296695979152


    I have not plotted it against your curve, but I am sure it is very close. It may have differed by the number of points taken before the step change. In this case, I ignored the 3 closest points to the boundary and used the previous 77 points [Boundary-80 to Boundary-3].


    The reason I don't want to use the command line is that it doesn't have much advantage because it would take significant re-coding of the script driven control program to add the capability to compute the values for the frequency for example, and compose a command line to drive it. It is a whole lot easier change to the control program to use the .wav file which I can create in a nice sound editor.

  • I have loaded the reactor tube with a fuel comprised of the same HCL processed Hunter Chemical AH50 Ni powder + LiAlH4 + metallic Li. The metallic Li was easy enough to handle in the dry glove box with tweezers. The humidity in my dry glove box was about 5.4% and it had air, not argon. Handling in air with low humidity was no big deal. I had to use a pair of wire cutters to divide the 1/4" OD x 3/8" long chunks into smaller pieces to drop into the tube. The tube is filled with fuel. I will now weigh it and begin re-assembling the reactor system. Hopefully, I will begin the next experiment in about 3-4 hours.


    This experiment will not only have the change in fuel, but will also employ the magnetic field stimulation.

  • What magnetic field value do you expect to achieve?

    What dead volume will you have ?

    Each magnetic field stimulation will comprise 6 x [3 seconds on, 2 seconds off] of 4 kHz sine wave current with about 4.5-5 amperes of current in the coil. I expect this to produce a sine wave magnetic field having peaks of +/-140 gauss. The dead volume is only about 1cc. Initially I will soak in the 100°C-140°C range under vacuum to dry out the system. Then I will heat with the back pressure regulator set at 90 PSIA (6 barA) pretty much all the way to 1000°C. I have extended the time between 400-600°C where some H2 absorption was taking place in the last run as Arnaud suggested. There are several times where I turn off the heat and turn it back on as a stimulation. Also, at the end, when the reactor is at 1200°C, I will pump down to 2 PSIA to see what happens as the Li may begin to boil.

  • Hi Bob, do you have plans for assymetric waveforms? Suhas use sharp up and slow down, like a saw.

    That sounds like a sawtooth. In Suhas' case that is a direct DC stimulation of the dusty plasma. The current flow will be even more sharp, because as the voltage rises, there will be no conduction and then suddenly the plasma will form, the resistance will drop, the current will go up, and the voltage will fall sharply. It is a different drive system. At the moment we don't know if he is adequately measuring either his input power or his output heat. If MFMP does measure a COP big enough to not be a mistake, there will be a lot of people building reactors of this type.


    In my case, I have a limited bandwidth to drive the coil due to its inductance which I must resonate to get more current to flow (without resonance, the current would be about 17x lower and the magnetic field lower by a similar factor). In this experiment the sine magnetic field will be applied in sort of a hammering set of bursts. Each stimulation will provide 6 bursts of high current sine wave 3 seconds long.


    I have already started collecting components to build both microwave stimulated dusty plasma and DC stimulated dusty plasma reactors.

  • I started the experiment. Of course, the temperature climb will be slow and boring. The first of the DAQ files should be in the Google drive in about 40 minutes. The magnetic field stimulations will begin at 500°C.


    The fuel load is pretty light at only 0.85 grams because of all of the Li volume. I only fill the 2" end of the closed alumina tube - so, I am filling by volume. This 0.85 grams contains about 0.72 g of the Hunter Chemical AH50 Ni powder that has been (as the previous experiment) etched in ultrasonic 15% HCl at 50°C for 10 minutes. The fuel also has 0.08g of LiAlH4 and 0.05g of Li metal.


    Based on my experiments with handling the Li metal, I think that people running experiments with an increased Li concentration should consider adding the metal rather than finding an exotic way to handle the Li into their reactor (like the nano-encapsulated Li). It is just going to melt and become part of the solution anyway at low temperature.

  • The experiment is progressing perfectly under automation. The system is at 600°C. Interestingly, without venting, absorption in the system has taken the pressure from its previous limit at 90 PSIA to 1 PSIA (vacuum) all on its own. I am going to try to get a plot to post soon.

  • Here is the plot for the first 18 segments:



    The COP calculation is based on back calculation from the current temperature, current power in, and the PvT curve calculated in the calibration run. The COP is only valid where the temperature is settled, and relies upon the fact that the input power is nearly constant.

  • Mats002

    There is a scheduled excitation burst at 500, 600, 700, 800, 900°C, and then at 1000°C there are multiple bursts per temperature plateau. I am going to add an excitation curve and a radiations curve to the suite shortly and you will be able to see where the excitation bursts occur.

  • This 0.85 grams contains about 0.72 g of the Hunter Chemical AH50 Ni powder that has been (as the previous experiment) etched in ultrasonic 15% HCl at 50°C for 10 minutes.


    Maybe you can test asetic acid and longer time at some run? HCl is quite hard acid.

    In preparing Urushibara nickel HCl etching reduce greatly catalytic activity (pores) and properly etching is done with asetic acid (or base).


    In some reports Raney nickel is found to produce XH. In some time ago I tested Urushibara iron, not find XH but it outgas lot of hydrogen when heated. Maybe with lithium it works better? I did it form PCB ferrochloride and Al foil..