The initial objectives for reactor D are:
- To document the energy input and power for a complete pyrolysis, and intermediate stages (dehydration, torrefaction).
- To document the background properties of the reactor - heat capacity, leakage, etc.
- To compare the progress of pyrolysis with a simpler model (evaporating a known mass of water)
I will begin with operation of the empty reactor to assess the control circuit performance, reactor durability, and thermal leakage.
I will conduct a duplicate operation replacing the empty specimen can with a can containing 100mL of water (heat of vaporization is about 62 Watt-hours for this mass of water at 100C).
In these experiments, I will make a regular recording of the power input profile at constant control settings (PID parameters and temperature set-point), record the temperature, and note conditions observed at the reactor vent. I will monitor the exterior temperature of the reactor with an IR thermometer.
The temperature rise is illustrated with the upper curve and the cumulative energy input is illustrated with the lower (straight line) curve. The initial rate of temperature rise changed as the reactor began to steam - giving up remaining free water in the plaster-of-paris. At higher temperatures, the rate of temperature rise did not return to the initial high rate because of heat loss - the reactor exterior was rising in tandem with the interior but lagging on the order of 100C. I will confirm with a second test, but this plaster and sand shell does not appear to be satisfactory insulation.
This second graph shows the calculated energy input required per degree of temperature increase. The peak (approximately the area above the interpolated straight line), reflects the water evaporation effect, which should not be present in a re-test. But the rising energy requirement for raising temperature illustrates that heat loss is occurring. If energy were only raising the reactor interior temperature, the graph would be a horizontal line because specific heats are relatively constant.
Interpolating the rising heat consumption, the temperature rise rate will decline to only about 4 degrees per hour at the threshold of pyrolysis (300C) and less than 3 degrees per hour at the ultimate target of 500C. Totally impractical rates.
Test 2 Results
The second test confirms suspicions of the first results: The temperature rise profile in this graph is logarithmic, and declines logarithmically after heat addition is stopped at 85 minutes. Although the heat demand for evaporating water from the plaster is gone and better reactor temperature was achieved, heat loss remains excessive. In an adequately insulated reactor, the temperature would have risen linearly, and then remained constant after heat addition was stopped. I will have to replace the plaster shell with another material.
Test 3 Results (figure is mislabeled as Test 2)
This test is for the reactor with fiberglass insulation (R4.3 air conditioning duct board). The temperature rise is linear on each interval: setpoint was initially set at 250C, then raised to 400C, then 450C. Each time the reactor temperature rose linearly to the setpoint and became nearly constant. Heat input was controlled by the Arduino circuit (though there are some control logic errors remaining - integral response is not working properly). This test produced much smoke as fiberglass binding agents were volatilized from the insulation and masking tape that secured inner layering of the fiberglass was presumably pyrolyzed. I will repeat this test with some water in the reactor to observe reactor response with a known heating load present, and to assess any changes in reactor properties that may have occurred after release of the insulation volatiles.
Tests 4 and 5 were reactor troubleshooting tests, there were some Arduino program errors which were corrected (and the posted program in this forum was also corrected). The reactor has some leakage, however, so further modification is needed. However, Tests 6 and 8 are compared here to evaluate the prospect of measuring useful pyrolysis data. In test 6, the reactor setpoint was 400C and 100 grams of water was put in the sample can. In test 8, the reactor setpoint was 415C and NO water was put in the sample can. Both tests’ data are shown here:
The upper curve corresponds to test 8 with no water. In test 6, temperature rise was delayed despite the same power input. The setpoint was reached about 18 minutes later. In both tests, the steady state temperature was maintained by steady application of about 200 watts - which represents the reactors heat loss at around 400C. So the difference between test 6 and 8 is 18 minutes of net addition of 300 watts, or 90 watt-hours of energy. This agrees well with the calculated energy requirement to heat and vaporize 100 grams of water: 0.1kg x 1.16 Wh/kg C x (100 - 30) + 0.1kg x 626 Wh/kg = 71 Wh
The reactor has an overall heat capacity, independent of samples that may be put in it. At the end of test 8, the temperature drop was observed after power disconnect. The steady-state power had been 182 watts, and the temperature drop was 9.4C/minute. Accordingly, the overall heat capacity C = (182 watts x 60 sec/min)/(9.4C / min) = 1162 J/C = 0.32 watt-hour/C (a more limited set of data from Test 3 give a result of 0.36 watt-hour/C, a good confirmation check, though probably less accurate). Compare to the reactor heating time from 30C to 400C: (400 - 30) x 0.32 watt-hour/C / 500 watt heating rate = 0.237 hour = 14 minutes. This agrees well with observation. The reactor is well characterized now. It is time to do some pyrolysis!
Test 11 is pyrolysis of 131 g of wood (fir, salvaged from a box-spring frame - old). The pyrolysis gas is directed through capture stages. The reactor setpoint is 462, for “fast” pyrolysis to maximize oil.
The biochar yield was 34g (26%), 31g of condensate was collected in ice bath, and an additional 7g of mass was collected in a filter canister (could be some additional liquid, because condensation was observed in the tubing after the condenser at peak smoke output. In total, 55% of the initial biomass was retained as char, condensate and filtrate. There was also leakage from the reactor and tubing system that released smoke, and smoke passed the filter too.
By comparing the heat input for Test 11 with Test 8, it may be determined that Test 11 consumed an additional 42 watt-hours of energy. This can be accounted from attributing heat of vaporization to the mass of collected condensate (which appears to be mostly water) 31g x 626 Wh/kg = 19.4Wh and the greater reactor temperature in Test 11 than in Test 8 0.32 Wh/C reactor heat capacity x 60C greater temperature = 19.2 Wh. These two heat consumptions total 92% of the graphically estimated heat input difference.
It would be interesting to compare the pyrolysis energy requirements of various biomass feedstocks, their various biochar yields, and their various condensible and filtrable fractions. However, there must be some reactor improvements before these tests could be performed:
- A reactor with sealed electrical connections to the heating element.
- A reactor that is reliably air-tight
- A more efficient condenser with greater flow capacity
- A more effective filter
- A pyrolysis-gas-handling chain that is better sealed
Without these improvements, I think the results of further tests will not yield reliable data.
I am pleased with the progress that has been made, and I hope these experiments can inspire others to take up further work.
Experimenters should note that freshly made biochar at pyrolysis temperature will spontaneously combust if exposed to air. When I open the reactor to make the final char mass measurement, the char appears black on all surfaces, but will soon develop embers and white ash will start to form. Soon after measurement, it is necessary to quench the biochar and cool it with water.