Date: 2009/08/24

 

Purpose

 

This is a continuing series of tests to explore the relationships between tar production and various critical measurements in the reactor.  The goal is to be able to use temp and pressure readings for gasifier diagnostics and establish a formal set of "conditions needed" for clean gas production.  Our hypothesis is that we can most accurately correlate tar production with temps maintained at the reduction restriction, which approximately measures how well x temp has spread and filled the hearth area for tar cracking.  The graphs at the end of the report plot tar production against this top of reduction temp, as well as other variables that might be contenders for relevant control or indicating parameters.

 

The main change in this test run vs the 2009/08/03 run is a new method to separate soluable tar from insoluble soot on our filter paper.  Simple "greyscale" of tar filter test has not been helpful in distinguishing between tars and soot in the prefilter gas stream.  Thus we are trying to establish a solubility separation method to better track tars separate of the tar/soot mix.

 

The secondary change in this run is a different gas flow rate cycling scenario, explained in more detail below.

 

Other tests in this series are linked to from the GEK User Pages and Run Reports page. 

 

Methods

  

 

Start Up

The reactor reduction cone was cleaned and filled with 1"-1/8" mesh sieved crushed char. The reactor was closed and filled with walnut shells via auger. A setpoint of 4 inH2O reactor pull rate was set, and the reactor was started through the ignition port/tube assembly with a propane torch. The port was left open for a few minutes and then closed. The reactor was generally held at 4 inH2O for 40 minutes to allow the reactor to approach thermal equilibrium (as judged by stabilization of the nozzle air temperature [T_air_in]). The electromechanial buzzer circuit on the dark room clock used for tar sampling caused GCU resets (gaps in data during start up). Isolating the clock resolved the issue.

 

Experimental Run

We desided to move to a dual "pyramidal" experimental design for the pressure setpoints. This experimental design was intended to avoid the effects of a quick change in flow rate, which was part of the experimental design on the 2009/08/03 test. While this "pulsed" experimental design may be useful to determine response under varied loads, it does not allow the reactor to reach more stable thermal equilibrium conditions. By repeating the same structure twice during the same run we can see how repeatable results are within the nominally same run conditions of a single run.

 

After start up, the GCU was programmed to ramp down from 8 to 1/4 ramp up from reactor pressure (vacuum relative to atmosphere) of 1/4, 1, 2, 4, 8 inH2O. The reactor pressure was held constant with the PID loop for 5 minute intervals. The reactor pressure was then ramped down at the same pressure points over 10 minute intervals. The choice of a slower ramp down speed was made to allow for the apparent slower cooling of the reactor thermal mass.

Structure of the described run (2009/08/24, upper) and previous run (2009/08/03, lower)

 

900°C

After the experimental run was over, we attemped to achieve T_tred temperatures above the ~850°C peak seen over the experimental run to see if we achieve lower tar levels. The run period was realitively short (starting after 220 minutes).

 

 

Tar Sampling

Spot samples were drawn at the reactor output, BEFORE the filter train, as in the previous run.  However, we increased the spot size from 0.28" to 0.375" dia. and switched from a 60 mL syringe to a small sampling pump (KPV-20A) run for 15 seconds (@ 5V) to pull 450 mL of of gas. The gas sample was drawn from the bottom filter port, through 16" of 1/8" neoprene tubing and through the filter media (ceramic wool insulation strip) sandwiched between two metal plates clamped down with wingnuts (shown above). We expect the woodgas to be much cleaner after the filter, though we are not characterizing the filter media during this run.

 

Filter strips ("exit" side up). In sets of ten, increasing right to left, top down.

 

Filter strips after spot removal with Xacto knife, grayscale chart for manual interpretation, 3 mL syringe.

Set of ten vials

 

Cropped images of vials (click for larger image)

 

Samples were placed in vials containing 3 mL 99%+ isopropyl alcohol and sample (sample IDs from 10-43, left to right). Samples were in solution for ~4 hours and shaken occasionally. Lacking access to a spectrophotometer, vials were photographed in front of an ambiently backlit piece of translucent plastic in sets of 10. The images were Gaussian blurred (5 pixels) to remove JPG 

artifacts, and a pixel color sampled from the horizontal center of the vial between the miniscus and bottom of the vial. The HSV 

values of the pixel were recorded. The saturation value showed the highest signal and was used here (all three values generally covary).

 

This value does not as of yet provide us a quantitative number regarding tar content [mg/m3], but does indicate what conditions lead to reductions in tar.

 

Results

Critical Temperatures

Sensor signal problem seen at ~180 minutes.

Combustion TC is now sheathed in a mild steel tube (SS sheath was oxidizing)

Reactor Flows

Reactor Vacuum        

Contour Plot of Vertical Temperature Distribution with Time

Reactor Conditions and Tar Samples

Despite auger issues, pyrolysis temps rarely if ever spiked (from low fuel) like the previous run. This may mean we did manage to keep fuel levels high throughout the run.

 

Measured Tar Correlations with Reactor Conditions

The regression fits assume a linear fit. Although hard to confirm within the noise of the data, there may be a non-linear dropoff in tar content, here seen to occur at approximately 800°C. There have been anecdotal reports of a steep dropoff in tar content related to top of reduction temperature. The response may be non-linear, seeing this trend in future runs will help to confirm this.

 

From Wikipedia:

"An interior value such as R² = 0.7 may be interpreted as follows: 'Approximately seventy percent of the variation in the response variable can be explained by the explanatory variable. The remaining thirty percent can be explained by unknown, lurking variables or inherent variability.'"

 

Data points between 10< sample #s <40 (the main experiment) where included in the above plots. Sample 39 was excluded due to thermocouple error (seen at ~170 mins above).

 

 

Correlation between Tar Solution Saturation and Spot Exitside Greyscale

Sample number (x-axis).

There is a moderate correlation (r^2=0.6) between the amount of tar in solution (as measured by digital photo solution color saturation, y-axis) and the visually indexed greyscale color of the sample spot (on the exit side of the filter media, x-axis). This is somewhat hard to interpret, but one explanation would be that increased tar content leads to better binding of soot onto the filter, disallowing it to pass and deposit on the opposite side of the filter.

 

Reactor Temperature Ranges Through Experimental Run

Temperature ranges seen through the experimental run (~85-215 min). Note that the ranges may be biased slightly by the erroneous reading around 170-180 minutes. Mean value indicated by black line, outliers by circles. Lower and upper end of box are the first and third quartiles, respectively (representing the middle 50% of the data).

 

Run Description
Run Name:	Instrumented Walnut Shell Run
Run Location:	Shipyard, Berkeley, CA
Operators:	Bear, Jay
Date:	08/24/09
Fuel
Type:	Walnut Shells
Moisture Content:	9.4% wet basis, 10.3% dry basis via microwave equilibrium method (08/19/09), when hopper was filled and sealed
Angle of Repose:	ND
Ash Content:	ND
GEK
Version:	v3
Reactor Type:	Imbert
Air Nozzle Size:	3/8" street 90°C with 3/8" cap drilled on center with 5.5 mm dia. hole, 2.75" above reduction top
Reduction Cone Height:	6
Top Reduction Diameter:	3
Bottom Reduction Diameter:	6
Tar Fence:	yes
Tar Fence Height:
Details:
Filter:	v3
Filter Media:	Char, run before, remixed
Fill Height:
Details:
Gas Motive Force
Ejector	yes
Jet Nozzle Type (eg barb, plug):	drilled plug
Jet Exit Position:
Fan	no
Power Source/Voltage:
Engine	no
Make:
Model:

 

GCU Setup Form
Run Name:	Instrumented Walnut Shell Run
Date:	08/24/09
GCU Version:	1.0
Firmware:	in development (PID control of reactor with servo controlled ejector air, timed grate shaking (1 min intervals, ~40° fwd/back rotation)
	Use
Thermocouples:
TC0	T_bred - bottom of reduction - 1" in from cone bottom, through manometer port
TC1	NA
TC2	T_tred - top of reduction - inside 1/4" mild steel pipe quarter round, welded closed and positioned on the top edge of the reduction cone
TC3	T_air_in - air in - inside air riser, through hole drilled co-axially with the riser tubethrough street 90°. 1 1/2" below top of riser.
TC4	T_comb - combustion - first TC of profile assembly. centered by ring rod assembly, 1" in front of air nozzle hole, at same elevation. Sheathed in 3" length of 1/4" mild steel tube.
TC5	T_1in - 1" above combustion. 2nd TC of assembly...
TC6	T_2in - 2" above combustion.
TC7	T_3in - 3" above combustion.
TC8	T_4in - 4" above combustion.
TC9	T_6in - 6" above combustion.
TC10	T_8in - 8" above combustion.
TC11	T_gas_out - installed in cowling gas exit port
TC12	T_flare - installed in 1 1/2" tangential entrance tube
TC13	T_gas_flowmeter - installed in 1 1/2" to 1/2" reducing T just after filter, before union based gas flowmeter
TC14	NA
TC15	NA
Drivers:
FET BANK 1:
VOLTAGE	12 V
FET0
FET1
FET2	Grate Fwd Relay (30A automotive) (FET # not confirmed)
FET3	Grate Rev Relay (30A automotive)
FET BANK 2:
VOLTAGE
FET4
FET5
FET6
FET7
Servos:
SERVO1	Ejector Air Control
SERVO2
SERVO3
Pressure:	Part (7002,5004,4006,7007,5010,7025,5050)
P0	7007 - P_comb - pressure at combustion, via profile assembly
P1	7007 - P_reactor - taken from reactor manometer port
P2	7002 - P_gas_out - gas out flowmeter differential pressure
P3	7002 - P_air_in - air in flowmeter differential pressure (flowmeter installed on V3 air cowling inlet)
I/O:
RS-232
CAN-BUS
ANALOG0
ANALOG1
ANALOG2	Auger plunger on sense (12V motor line, through voltage divider)
ANALOG3
Frequency Counter

 

 

Run Data and R Code:

run_data_20090824.zip