Update April 4th 2012
The 6hp Lister is now running sweetly at 17:1 compression ratio on woodgas and producing about 2750 electrical watts when maxxed out.
We have a lot more instrumentation, datalogging and servo control to do - but we have the basics running now - and are confident we are on the right track.
What fooled us was the amount of ignition advance we needed - the sensor is triggered by a ferrous washer placed a full 12" on the rim before TDC. On a 24" diameter flywheel that equates to 57 degrees BTDC!
We will be doing more tests on power and fuel consumption as and when we can fit them in - but at least we now have the system running smoothly.
Incidently, the original 6hp Lister Startomatic genset alternator was only rated at 2500W - so we are right in the zone.
Woodgas works well at 17:1 - so forget the rumours, the idle back chat and get converting those diesels to woodgas. We now have an open source demonstrable system - right here at APL.
It's been a long drawn out project - but thanks to the dedicated and determined help of Ron Ohler, Marcus Hardwick Andy Schofield, and all the gang who came by last weekend we have finally got our act together.
- Full report and video to follow shortly on youtube and APL GEK Wiki.
Background.
In February 2010, a group of gasifier enthusiasts met at APL with the intention of converting a 6hp Lister CS type diesel engine to spark ignition so that it could run on 100% wood gas without the need for "dual fuelling".
By converting the popular Lister CS type engine to spark ignition, allowing it to run reliably on woodgas, will open up a whole new fuel source to those, such as off-gridders, that currently use Lister type engines for home-power production.
After a long weekend of tinkering we got the Lister to run just before 7pm on the Sunday evening. The short video below was made of the first trial run of the engine after the spark conversion was completed. Further runs produced much more smoother running, once the wood gas mixture and timing adjustments had been optimised. Other videos taken at the time of the original conversion can be found on YouTube - just search for "Lister Spark Conversion".
For the purpose of this conversion, it was discovered that a long-reach spark-plug was almost a perfect fit into the injector hole in the top of the Lister cylinder head. The spark-plug could be clamped in place, down onto it's sealing washer using a short length of steel tube, and the existing injector hold down clamp. Pictures of the original spark plug and mounting method are on the GEK Forum here.
The tip of the spark-plug protrudes neatly into the spherical pre-combustion chamber in the Lister head, from where it ignites the compressed air and wood gas mixture. To ensure that the spark plug fires at the right time, it needs to be synchronised with the timing of the crankshaft, or more appropriately, the camshaft. As the camshaft on the Lister is not quite so easy to access, it was decided to take the ignition timing from the flywheel, and accept that the spark-plug would spark twice - just before the power stroke and just before the end of the exhaust stroke. The second of these sparks is redundant and has no useful purpose.
Compression Ratio.
For the initial tests, we fitted additional cylinder head gaskets, which had the effect of lowering the compression ratio from about 17:1 (for normal diesel running) to about 13:1. As the conversion was purely experimental, and we had no guarantee that it would be successful, we chose to "play safe" and use a low compression ratio which would not cause pre-detonation of the air/gas mixture.
Subsequent information has come to light, from the Indian Institute of Science (IISc), which describes how diesel engines have been successfully converted to spark ignition and run at the standard 17:1 compression ratio, without any problems or detrimental effects. Indeed the research paper went into significant depth to dispel some of the myths about wood gas at high compression ratios, and proved that it was beneficial to power output to run at a high compression ratio - thus offsetting to a certain degree some of the engine power derating on woodgas.
It has therefore been decided that subsequent experimentation with the Lister will be done at 17:1 compression ratio, which has the benefit of maximising the power output from the engine, and increasing its efficiency on wood gas.
Further to our original experimentation, it was agreed that a better location for the spark-plug would be in through the side of the hemispherical combustion chamber - making full use of the removable cast iron plug, which is fitted to some Indian Lister engines, where the compression change over valve (COV) would otherwise be fitted. This has the additional benefit that the original diesel injector can be retained, allowing the engine to be started on diesel, biodiesel or vegetable oil and then switching over to wood gas once the gasifier is generating good engine grade gas.
Marcus Hardwick has modified the COV plug to fit a long reach spark plug - images below.
This was a modification of a spare JKSon COV bung to accept an M10 long reach spark plug. The plug is a NGK CR7E (other plugs may have a different threaded shank length). A quickie layout was done with SketchUp to determine the approximate depth of bore for the plug seat (a flat, gasketed seat not a chamfered one). 2 3/8" was picked as the target depth.
The boring was done on a 1948 vintage LeBlond Regal lathe. First a through hole was made at 9mm, the tap size for the M10x1.0 spark plug threads. Then the bore was step drilled to 7/8" to accept the 16mm spark plug socket. Indexing the depth of bore was done simply by using the graduated scale on the tail stock quill. Zeroing the start of each drilling operation was by using a thin strip of sheet metal across the hole and bringing the tail stock up so the tip of the drill just kissed the sheet metal. After the 5/8" step was bored, a 5/8" piloted counter-bore was used to make the flat face for the plug/gasket. After the 7/8" diameter was reached the bung was tapped for the M10x1.0 threads. A slight chamfer was made on both ends of the threaded portion with a countersink to clean-up & deburr.
The 2 3/8" depth worked out pretty good. Counter-boring another 20-30 thousandths would likely make it just right. MH
This approach would make the Lister a versatile multi-fuel engine, with the reliability and convenience of starting on diesel and then the economy of running on wood gas. With engine power available initially on diesel, this would facilitate the start-up process of the gasifier.
Further Developments.
With this approach in mind for future experimental and development work, the intention is to complete a demonstration Lister engine at APL, characterise its performance on woodgas and make our findings available to all, in keeping with the open-source nature of this project. In due course, a kit of parts will be made available to convert Lister and other small diesel engines to spark-ignition.
Now with the October '10 Workshop approaching rapidly, there has been a spate of new activity on the Lister project, so I thought it would be a good idea to start documenting some of the work so far.
Requirements.
A modern spark ignition system consists of the following components:
1. Crank angle sensor
2. Timing circuit - usually a microcontroller
3. Power transistor
4. Ignition coil
5. Spark plug
The crank angle sensor is used to detect a timing reference point, about 20 degrees before top dead centre. (BTDC). We used a low cost Hall effect switch which can be mounted on the engine such that it detects a neodymium magnet disc that has been attached to the flywheel. The Hall sensor has an open-collector output and produces a logic low signal when the pole of the magnet passes the sensing face.
The timing circuit in our case consisted of the popular Arduino microcontroller board. This was chosen because it is easy to program, low cost and it is compatible with the Gasifier Control Unit (GCU). We used a rotary potentiometer to give a variable ignition delay time. This meant that the timing could be adjusted manually when the engine is running for optimum results, without having to mechanically re-position the Hall sensor.
To fire an ingnition coil you need to give it a short pulse of current to energise the coils magnetic field, and then when you kill the current, the back EMF produced by the collapsing magnetic field produces a high voltage sufficient to fire the spark plug. For simplicity we used a TIP122 NPN Darlington power transistor which interfaces directly to the Arduino output pin, via a 1K resistor. The transistor is mounted on a small piece of aluminium plate, to act as a heatsink, if the transistor were accidently left in the on-state, it would rapidly overheat.
The ignition coil is a standard automotive type. It is connected between the positive of the 12V battery and the collector of the power transistor.
The spark plug was a long reach type which fitted perfectly through the hole in the Lister cylinder head, and was held in using a short length of steel tube and the injector hold down clamping plate supplied with the Lister.
Stripboard Prototype.
Below is the prototype spark ignition controller. It is based on the Atmel ATmega328 microcontroller - which is the same one used in the Arduino.
The schematic is here: Lister_spark_IC.pdf
The following I/O pins were used
Pin 2 Serial Rx Input from PC
Pin 3 Serial Tx Output to PC
Pin 4 Digital Input from Hall Effect sensor (Arduino Digital Input 2)
Pin 14 Digital Output to trigger power transistor (Arduino Digital Output 8).
Pin 19 Digital OUtput to Drive LED (Arduino Digital Output 13).
The board is clocked with a 16MHz crystal to retain compatability with Arduino. The firmware is developed as normal using the Arduino IDE.
The board has a 5V regulator, a rotary pot to delay the spark and a blue LED to flash in time with the firing of the power transistor. The TIP122 transistor is mounted on a small heatsink. Connections to +12V battery, ground and the low tension to the ignition coil are provided on 1/4" automotive spade terminals. The 6 way cable exiting to the top left is an FTDI USB to serial cable. This provides +5V for testing and the serial interface for programming the board and getting rpm and ignition delay data back to the laptop.
The 3 pin Hall sensor plugs into the pin-socket just below the FTDI cable header - it has the same pin order as the Honeywell sensor +5V, 0V and signal out. For testing the sensor device can be plugged straight into this connector, and later extended with a 3 way cable.
The board measures 3" x 4.5" and there is space below the heatsink to take additional circuitry. You will need bit of stripboard at least 28 holes by 32 tracks to copy this layout.
Only 4 of the I/O pins are being used in this application, leaving the remainder for other future use.
One possibility is to add a further pair of TIP122 transistors, so that this board could form the basis of the spark controller for the 3 cylinder Kubota engine. If additional TIP122 transistors are added, they need to be isolated from each other if used on a common heatsink.
From the top the spade connectors are +12V, battery ground and coil output.
As a general point, the pin-out of the Arduino (see below) and ATmega328 maps very easily onto breadboard or stripboard. This means that small Arduino compatible projects like this can be made on stripboard or even breadboard at a fraction of the cost of buying the real Arduino and building a board onto it.
Buildtime was approximately 5 hours - but that included a lot of thinking time to get the component positioning sensible. To copy this board would take a couple of hours.
Partslist.
1 ATmega328 microcontroller
1 28 0.3" pin DIL socket
1 7805 5V voltage regulator
1 16MHz HC49-4 crystal
1 10K rotary pot or trimmer pot
2 10K resistors
2 330R resistors
1 1K resistor
2 22pF ceramic capacitors
4 100nF ceramic capacitors
2 22uF 16V electrolytc capacitors
1 1N4001 diode
1 TIP122 darlington power transistor
1 Heatsink
1 Stripboard at least 28 holes x 32 strips
1 6 pin 0.1" header
1 3 pin 0.1" socket
3 0.25" pcb spade connectors
single core hook-up wire - 4 colours used
1 M3 (or equiv) bolt and nut for fixing heatsink
1 Hall sensor (Honeywell SS441R)
1 neodymium disc magnet
1 FTDI cable for programming (optional) (ATmega328 can be programmed on Arduino board and swapped across).
Layout and Construction notes.
Here is how the Arduino I/O is mapped onto the pins of the ATmega328 microcontroller. Further details about arduino are available on their website and forum.
Care was taken to ensure that the signals from the FTDI cable header (+5V, Tx and Rx) lined up with the pins 1,2 and 3 of the ATmega328.
The 10K reset pull-up resistor was fitted between +5V and pin 1 and the track cut. Rx and Tx line up directly with their "opposites".
+5V and 0V ground were linked down to pins 7 and 8 respectively. A 100nF decoupling capacitor was fitted between pins 7 and 8.
The crystal was fitted between pins 9 and 10 and the two 22pF loading capacitors linked back to pin 8 ground.
A 10K pull-up resistor for the Hall sensor input was fitted between pin 4 (Arduino Digital Input 2) and pin 7 +5V.
The 3 pin socket was fitted for the Hall sensor connection picking up +5V, 0V and the yellow link to pin 4.
The voltage regulator, the reverse polarity protection diode and the two electrolytic capacitors were fitted - noting the capacitors polarity - to the middle ground pin of the regulator.
Links were fitted to connect the voltage regulator pins 2 and 3 to the 0V and +5V pins of the FTDI header.
On the right hand side of the chip a 100nF decoupling capacitor was fitted between pins 20 and 22 - decoupling AVCC to ground.
A 330R resistor was fitted from the +5V output of the voltage regulator to pin 20 of the IC.
The rotary pot was fitted such that its wiper aligns with analogue input 0 Pin 23, and the low end of it's track to ground on pin 22.
A link from +5V was conneted to the other end of the potentiometer track. Pin 24 had a track cut to isolate it from this +5V link.
An LED was fitted with anode to pin 19 (Digital 13) and a 330R resistor from the cathode to ground.
The power transistor was fitted with its leads as close as practical to the RHS of the board. The collector and emitter leads could carry as much as 5 amps so the connections to the 1/4" spades need to be kept short, and the emitter is connected to it's spade terminal with a short link of stout wire - shown green in the picture.
Fit a 1K resistor from the base of the TIP122 and up to Pin 14 (Digital output 8).
A link was fitted to ensure that there is continuous ground from the microcontroller to the emitter of the power transistor and the 1/4" spade.
On the underside of the board all tracks between the microcontroller socket pins were cut with a drill bit and a single link was fitted between the ground pins 8 and 22.
Before powering up use a test meter to check for power rail shorts and power continuity of the +5V and 0V rails. Check carefully for very fine shorts where tracks have been cut with a drill bit.
Programming.
The board is programmed using the usual Arduino IDE, with the FTDI cable being recognised as a serial port.
The FTDI programming cable does not use the automatic reset function used on the standard Arduino. A reset switch could be fitted between pin 1 of the micro and 0V, but it is easy to use a piece of hook-up wire to make a temporary short between Pin 1 and ground.
If you can't readily obtain an FTDI cable, Sparkfun Electronics can supply a FTDI chip with USB socket on a small break-out board. This can be used on with a standard USB cable in a similar fashion.
In Use.
The board interfaces to the Hall effect sensor which is mounted on a bracket such that it is triggered by a small disc magnet once per revolution. The Hall sensor specified is unipolar, which means that it will only be triggered by one pole of the magnet. Make sure you get the right face lined up - otherwise turn either the magnet or the Hall sensor around.
Using the simple code below - The code is written such that it will produce 5mS firing pulses to the coil for as long as the magnet is within sensing lange of the Hall sensor. If you manually swipe the magnet past the sensor it will produce a string of pulses and the LED will appear to stay on until the magnet has passed.
This has the added advantage that at very low rpm, the coil and sparkplug will be fired many times which will help to ignite the wood gas when the engine is being hand cranked at about 60rpm. This is effectively a multi-burst discharge.
If the magnet is fitted to the flywheel boss - which is about 4" diameter, an 3/8" magnet represents an arc of about 1/10th of a revolution. At 600rpm this equates to about 10mS of time. The plug will fire once and possibly twice. When hand cranking at 60rpm or less the plug will fire possibly 10 times.
The rotary potentiometer allows up to 16.4mS of delay with the standard code - which represents 60 degrees of rotation at 60rpm. This is quite a gross range of control and it can be changed to a shorter overall delay by editing the line
sensorValue = (sensorValue) * 16;
If you make this a multiplication by 8 it will half the maximum delay and give the pot control a finer overall resolution. This scaling factor can be optimised following some experimentation with the real set-up.
sensorValue = (sensorValue) * 8;
By placing the sensor magnet on the inside rim of the flywheel, it will have a higher angular velocity and allow pulse timing to be made with greater precision. Ideally the magnet should be placed something like 25 degrees before top dead centre and this will allow the 30 degrees of control to cover any likely timing requirements.
Connecting Up
The prototype board should be initially bench tested, using the FTDI lead to supply power and communications from the laptop.
1. Noting the picture above - plug in the FTDI lead such that the black wire is towards the top of the board as shown. The blue LED next to the potentiometer should give a brief flash.
2. The Hall effect sensor plugs into the 3 pin socket just below the FTDI cable. The photo above shows the Hall Sensor temporarily inserted directly into this 3 pin socket. Note the orientation of the Hall Sensor. It's "trapezoidal" front face is directed towards the left. The connections from the top are +5V, 0V and Hall output. The output has a 10K pull-up resistor fitted on board. Make up a suitable length of 3 wire extension lead terminating in a 3 pin male header - so that the Hall sensor can be plugged in.
3. Test the circuit with a magnet. The Blue LED will flash each time the magnet passes the Hall sensor. If the magnet pauses by the Hall sensor - the LED will appear to stay on - but it is actually pulsing 5mS on, 5mS off. The LED mimics the signal to the power transistor directly. Running a terminal program on the laptop (or the Arduino monitor window) will show a calculated rpm reading, plus a delay timing in the range of 1 - 16384. This is the delay in microseconds from the Hall sensor triggered to the power transistor being fired.
4. Now wire up the ingnition coil to the spark controller board. The low end of the coil is connected to the bottom 0.25" spade terminal on the right hand side of the board, and the "hot" end of the coil connects directly to the +12V terminal of the battery. The 0V terminal of the battery connects to the middle 0.25" spade terminal. If these connections are made, the board should fire the spark plug - even though the board is still powered from the USB supply from the laptop.
5. If the above is successful, an additional +12V feed can be taken from the battery +ve terminal to the top 0.25" spade terminal. This will now supply the controller board with a regulated +5V supply from the 7805 voltage regulator.
6. The board should now be put into a waterproof enclosure and fitted to the Lister engine.
7. The engine top dead centre should be located and marked clearly with white tape on the flat surface of the flywheel rim. Noting the 24" diameter of the flywheel, 1 degree of timing adjustment equates to 0.2" at this radius. The Hall sensor should be located such that it triggers approximately 25 degrees (5" on rim) before TDC. This will give plenty of adjustment range on the potentiometer timing delay control.
8. When the engine runs, the rpm and timing delay parameter are shown on the laptop for every revolution.
Schematic.
The schematic of the Arduino based ignition circuit is in the above pdf. This shows the details of the Arduino I/O pins that were used for the various functions. A LED is made to flash in synchronism with the spark plug firing. This provides a useful diagnostic to confirm that the Hall sensor pick-up is working correctly.
Arduino Code.
Two versions of Arduino code have been produced. The first one allows simple firing of the spark plug on receipt of a low going pulse from the Hall sensor.
A second version uses interrupt driven code to trigger the firing of the spark plug, which allows the Arduino time to perform other monitoring or control tasks in addition to firing the plug. These tasks might include monitoring the key temperatures in the GEK, adjusting throttle settings or general engine management. The interrupt driven code is still under development with the intention of releasing it as a general low-cost Arduino based engine/gasifier control system.
The simple code has been reproduced below - anyone with an elementary understanding of microcontrollers should be able to see how we are doing things.
lister_spark_3.pde - Simple Code
_2chan_tach_Lister.pde - Interrupt driven Code
/*
//=========================================================================================
// Lister Iginition Coil Trigger
// Wood Gas Lister Spark Ignition Conversion
// by Ken Boak, for All Power Labs, Feb 22nd 2010 with new features October 1st 2010
// Gives 125uS to 16.64mS delay following trigger from Hall sensor
// Equivalent to 0 to 60 degrees iginition shift at 600rpm
// A 10mm diameter magnet is about 1.8 degrees wide on the flywheel rim but about 11.5 degrees on the 4" flywheel boss
// 522uS on the rim and about 3.2mS on the boss
//=========================================================================================
*/
// I/O Pin Allocation
int LED_Pin = 13; // on board LED connected to digital pin 13 to ground.
int Coil_Pin = 8; // TIP122 base connected to digital output pin 8
int Hall_Pin = 2; // Hall switch on input pin 2 (can be used later as interrupt)
int sensorPin = 0; // select the input pin for the potentiometer Analogue Zero
int HallState = 0;
int Pot_Pos = 0;
int rev_start = 0; // Start time of revolution
int rev_start_old = 0;
int rev_stop = 0; // Stop time of revolution
unsigned int rpm = 0; // engine rpm
unsigned int sensorValue = 0; // variable to store the value coming from the sensor
// The setup() method runs once, when the sketch starts
void setup() {
// initialize the digital pin as an output:
Serial.begin(9600);
pinMode(Coil_Pin, OUTPUT);
pinMode(LED_Pin, OUTPUT);
pinMode(Hall_Pin, INPUT);
}
void loop(){
while (HallState == HIGH) { // Keep looping here until Hall sensor input goes high again
HallState = digitalRead(Hall_Pin);
}
//***********************************************************************************************
// check if the Hall sensor is Low, - magnet passing sensor
// if it is, the HallState is LOW:
// Delay for a given number of microseconds then continue to fire plug whilst low.
// 277uS is 1 degree at 600 rpm, 255uS at 650 rpm
// 30 degrees is 8333uS
// Multiply sensor value by 8 to give 30 degrees of ignition variation (+/- 15 degrees)
// Multiply sensor value by 16 to give 60 degrees of ignition variation (+/- 30 degrees)
// read the value from the potentiometer:
//***********************************************************************************************
sensorValue = analogRead(sensorPin);
sensorValue = (sensorValue) * 16;
if (sensorValue == 0) {
sensorValue = 1; // set to 1 as delay microseconds does not like 0
}
if (sensorValue == 16384) {
sensorValue = 16383;
}
delayMicroseconds(sensorValue); // delay the spark timing by the setting from the pot 0 - 30 degrees
while (HallState == LOW) {
digitalWrite(LED_Pin, HIGH);
digitalWrite(Coil_Pin, HIGH); // Turn the Coil on
delay(4); // wait 4mS - Change this to alter multiburst frequency
digitalWrite(Coil_Pin, LOW); // Turn the Coil off
digitalWrite(LED_Pin, LOW); // Turn the LED off
delay(4); // wait 4mS - Change this to alter multiburst frequency
HallState = digitalRead(Hall_Pin); // Is the Hall sensor still triggered by magnet?
}
// Now calculate and display the engine rpm and print out the spark delay in uS
rev_start = millis(); // count the milliseconds per revolution
rpm = 65535 -( 60000/(rev_start_old - rev_start));
rev_start_old=rev_start;
Serial.print(sensorValue); // print out the spark delay in uS
Serial.print(",");
Serial.println(rpm);
}
//******************************************End******************************************************
Timing Pulses.
The photo below shows the output signal of the spark ignition on the blue trace, having been triggered by the signal in the red trace. This was done using another Arduino to generate the 600rpm (10Hz) trigger pulses - effectively simulating the output of the Hall effect sensor. The output trace (blue) is a 4mS wide pulse, which is currently at about 8.4mS after the start of the triggering pulse. *.4mS is about 30 degrees of crank angle on the Lister running at 600rpm. Having a stable pulse generator for bench testing, makes the system a lot easier to debug without having to actually deal with real running engines. The code running on the Arduino allows the width of the coil firing pulse, and the delay to be varied quite easily. It is also possible to generate a multiple series of short firing pulses to spark the plug several times each revolution.
Video.
A short video was made of the first trial run of the engine after the spark conversion was completed. Further runs produced much more smoother running, once the wood gas mixture and timing adjustments had been optimised.
The Team.
The team working on the Lister spark conversion project have come from an international community of wood gas enthusiasts. Special thanks should go to the following individuals whose efforts have made this happen, plus to the wider wood gas community and the staff at APL.
Mike Anthony
Ken Boak
Darrel Licks
Marcus Hardwick
Jay Hasty
Bear Kaufman
Jim Mason
Ron Ohler
February 2012 Update
I have finally made some opening progress on the Lister wood-gas
conversion project.
At sundown yesterday we fork-lifted the Listeroid from where it was stored on
top of a shipping container, and got it down into a position where I
could work on it.
I pulled the head and removed the extra gaskets that we had fitted in
2010 in order to lower the compression ratio.
I cleaned up the head and cylinder block and fitted the stock gasket.
This should return the CR to close to 17:1.
I re-adjusted the valve clearances - and then check the oil - or
should I say mayonnaise.
I cleaned all the emulsified goop out of the crankcase and refilled
with fresh lube oil.
A fast spin of the starting handle and she fired right up. This
engine has not been run since October 2010.
I spent the evening test running the engine alternator on various test
loads and at 50Hz and 60Hz. The intention is to get the engine and
alternator characterised on diesel, so that when we switch over to
wood gas (possibly next weekend) we have representative figures for
maximum electrical output on diesel - so that comparative
measurements can be made.
For those of you interested in the whole Lister/woodgas saga - I will start blogging it, and make my findings
available on the APL Wiki.
I took a few videos this evening - and have initially posted them up on my Youtube channel.
We are hosting a weekend workshop session on March 31, April 1,2
where the "wood gasification" of the diesel Lister will be one of the
projects. We will be adding the Level 4 GEK and Totti to the Lister
generator skid, and logging its performance.
Out of this work, we hope to show how a gasifier can be combined with
a low speed diesel and some simple open source electronics (Arduino +
shield) to make a reliable biomass fueled genset - of direct
relevance to the developing countries - where Listers are still very
much commonplace.
We want to start on diesel - to jump start the
gasifier - as we have thermal and electrical power available, then
seamlessly fade across to woodgas, thus minimising the overall diesel
consumption.
This is not dual fueling, it's a diesel-start, woodgas run - and I
hope to see approaching 95% diesel replacement.
For those interested - contact me at All Power Labs
Progress Report - March 10th/11th
Original team member, Marcus Hardwick came down to All Power Labs this Friday, and we were also joined by local newcomer but enthusiast Marlin Watson.
We set about doing some of the jobs on the Lister that had previously done in the panic of a Weekend Workshop - and now we had the luxury of time and willing hands to get some things sorted out.
The Lister had been leaking water from around the head gasket - so we fitted a stock Lister gasket - and sealed it around the water channels with high temperature silicone sealant (RTV - as we call it here at APL).
There had also been a leak around the injector - and we discovered that there was no copper washer! A little more high temperature silicone solved that problem.
The Lister is now capable of retaining nearly all of it's bodily fluids - well almost!
Marlin and I made some J - ties (candy canes) from 3/4" rebar. It was fun trying to remember what I learned 25 years ago about the oxy-acetylene torch - so that we could do some localised heating and bending of the J- bars.
These were hammered through the asphalt and stop the Lister skid walking about. It still shakes - but at least it does not wander off anywhere!
Marcus has made up a new mild steel platform for mounting the throttle servo, the decompressor solenoid and the governor linkage on. It brackets onto two of the cam-housing cover bolts and one bolt at the injector pump end. We also have a new steel tray for mounting the electronic ignition system and a small 12V battery.
We positioned the GEK at the end of the 72" long skid - but not on it. Whilst the Lister shakes like a tree - it may not be fully beneficial to have the grate being constantly "riddled".
Marcus has installed a 3/4" gate valve in the cold feed from the 55gal drum water hopper - so if we ever need to pull the head again - we don't need to drain the hopper - just close the valve. Ultimately I want to move away from the 55gal cooling hopper - and exploit some of the heat energy from the coolant.
We have a plate heat exchanger here - and in readines for the CHP experiments - we have installed a 1200litre (300 US gal) insulated stainless steel tank next to the Lister. For those of you who know the layout here at the Shipyard, the tank is immediately outside the laundry room and bathroom - so we hope to benefit from some hot water in these two rooms. There are already holes through the wall to accept insulated pipes - so the plan is to turn the tank into a thermal dump load - for when we commission Power Pallets - as well as a research facility for engine exhaust and coolant heat-exchanger experiments.
With 1.2 tonnes of water in the tank - we can sink 100kWh of thermal energy - which is 20+ hours for the Lister or 5+ hours for our 10kW Power Pallet.
This afternoon I jet-washed all the oily grime off the Lister - its looking a lot cleaner, but a patchwork of the red paint and the original dark green - but we may have a new paint-job by the time of the Workshop.
I spent time today re-visiting the electronic ignition module - that I built up in the UK for the October 2010 workshop. It still works - despite months outdoors, and I am rehousing it into a neat new box. I spent time doing some power frequency versus rpm measurements - and the figures are 653.5 rpm for 60Hz power and 544.6 rpm for 50Hz power. Admittedly the Lister is a lot happier at 50Hz, - 60Hz seems to hit an annoying resonance of the skid.
This evening I took advantage of the extra hour of daylight - it's 7pm and the sun is till up (just - its sets at 19:14) and ran the Lister on biodiesel for a couple of hours with a 3kW load - as a shake down test.
All seemed to be well - we had sorted the fuel and water leaks - but the oil seems to be emulsion again - possibly because of an earlier water into oil leak - or me being over-zealous with the pressure washer.
So that's this weekend's progress. Next weekend Marlin and I will be building a GEK and TOTTI up together - with the intention of having it running on woodgas, either towards the end of next weekend or the week later. This still leaves us a week of "tinkering time" prior to the upcoming Workshop which begins on March 30th.
If you are coming on the 30th - and specifically want to work on the Lister project - please email me. This write-up will also appear on the APL GEK Community Blog.
More Ideas
Anyone who's ever worked with Listers will be aware that they tend to vibrate when skid mounted - they really need to be bolted down to about 2 cubic yards of concrete. This shaking could be used, selectively to our advantage - to shake the grate and encourage fuel flow down the hopper and pyrocoil.
I had an idea tonight of a solenoid acting as an electromagnetic coupling between the GEK and a tie-bar from the top of the Lister. Energise the solenoid and the grate is shaken at 5.5 Hz by the pounding of the Lister. This would eliminate the grate shaker motor and housing - and replace with a $15 solenoid. More experimentation needed.
I like the idea of adding an extra annular skin around the upper 3/4 of the outer gas cowling of the GEK. This would form an annular air preheater - in addition to the air heater tubes. By sizing the diameter of this outer skin correctly, the air for combustion would be brought into close contact with the gas cowling - create swirl and turbulence and allow for good heat transfer. The question is whether this would shift more heat out of the woodgas and into the air - than just having the gas cowling open to the air - and lose heat by radiating.
There is almost an argument for having a second Pyrocoil. With diesel start, the lower pyrocoil would be heated by the hot engine exhaust and be used to quickly get the fuel up to pyrolysis temperatures. The second Pyrocoil would be heated by the exist gas from the gas cowling, and this acts as a means of shedding heat from the gas into the biomass higher up the "stack" for fuel drying and torrification. This second pyro would be slower to heat up and reach thermal equilibrium with the fuel - but advantageous on long duration runs.
Adding a 45 degree cone to the base of the 55 gallon drum provides a slippery slope for encouraging fuel flow - but also adds terrific surface area - especially useful if the cone is being used as the "drying bucket" and heated with the outgoing woodgas. A cone would offer about 0.36m2 of surface area - double that of the existing drying bucket. And thus I will christen this concept the "PyroCone"
It is estimated that the Lister will run for between 8 and 10 hours on a full 55 gal hopper, with a typical electrical load of 2.5 to 3kW plus any that is in the stack. This is an acceptable time between refuels for a genset of this application.
Pulsing Power
Anyone who has stood next to an unsilenced Lister will know that there is a terific pulse to the exhaust - about 5 times per second. The power of this pulsating gas could be utilised in a number of ways. By tapping off say 5% to 15% of the exhaust gas flow - it could be used to run an ejector or wind sieve - a bit like the Kalle Gasifier. Secondly by expanding this gas into a vessel - eg a small drum, you could use the natural pulsation to produce a shaking motion - which could be coupled to the fuel hopper of the gasifier to encourage fuel flow.
With the shaking of the Lister - it should be possible to achieve good gravity fuel flow - without a fuel feed - auger. This would be a further cost reduction and simplification for deployment in cost sensitive markets.
