Early experiment prototype, redundant equipment archives #13

NOTE: Some/most of these posts are discontinued or redundant but are here for reference.

Settling vessel prototype

Always there's foreign matter that settles out from the retort produced fuel and to facilitate the process, I made a small settling tank with a tapered cone so that particles would collect at the lowest point and be easily drained off through the ball-valve. Pic shows the tank (left) plus an optional stand (right) for running settled oil directly into a filter bag. Otherwise this vessel could be fastened to a wall or suspended from the ceiling.
 Should the vessel work well, I'll make a bigger model.

Update 1 April 2013:
After testing, I found that particles stuck to the sides of the tapered section.
I was surprised as the angle is steep so pre-straining was found to be necessary. It separated water out as expected.
Overall it is a useful tool so I may make a bigger version from a 200L drum with a larger drain valve and sight glass. It will be supported on a pipe frame and fixed to the wall for safety.
Thanks for viewing.

Bubbler design made easy

It doesn't get much simpler than this bubbler idea and it's very cheap/free to make, so anyone can afford.
I have use this type with early prototypes and it has proved quite effective. It's efficient and is transparent so status can be viewed at a glance.

PE 20 liter container


Caution: incoming gases should be at a reasonable temperature so the plastic doesn't melt, therefore it should be placed well downstream and when the gases are sufficiently cooled. Target temperature about 20C.

Reflux condenser

This reflux vessel is above the retort so that vapor that condenses in the reflux can run back by gravity and thereby be re-cracked to a lighter fraction. Once the vapor rises up as a light fraction, it will exit the reflux, heading toward the condenser array. The temperature is therefore important, as too high a temperature will mean heavier fractions will escape to the condensers, whereas too cooler temps will mean the vapors will get cracked down to lighter and lighter fractions.
To put this another way, heavy fractions can condense at higher temps while lighter fractions require lower temps. This is why temperature control of the reflux vessel is so important because those vapors that condense there, will run back to the retort.

The reflux vessel is half filled with broken terracotta and lathe swarf. This dramatically increases the surface area for the reflux. There is a bung for filling and inspection. A thermocouple probe port is positioned topside for temperature monitoring.

Please remember that this is a experimental prototype and intended for learning purposes. Much of what will be learned may be used for subsequent design improvements.


Temperature probe thermocouple equipment:

The temperature probes used are k-type thermocouples which are favored for their range of high readings. The handheld style of display was bought from HongKong via Ebay along with 4 probes. The adaptor as shown was made from a BSP blanking plug and a length of 6mm tube which was brazed to the plug and crimped/welded at the lower end. These are essential tools for thermal cracking of hydrocarbons.
Ebay thermocouple seller

Flame trap / flash back arrestor:

I made this flame trap so that non-condensable gases could be fed to the turk burner with an increased margin of safety. These are the gases that exit the retort and don't condense in the condensers or the bubbler. The bubbler is a very good check valve but why not have a 2nd safety feature?
The cannister is moderately packed with lathe swarf allowing a small gap at the top and bottom. Important is to have the inlet at the top and the outlet (to turk) at the bottom. The gases add significant extra fuel to the turk, improving efficiency.


Injectors to fire the turk:

These are the injectors that squirt the fuel as the heat source. They have done 2 -3 hours use and are looking reasonably clean. (The upper one was in position but not connected with an injector line). The worry was that they might carbonize however, the fuel was not WMO rather it was clean used oil/gasoline blend. The injectors improved the control of heat and also saved fuel.
They are from a Toyota 2L diesel engine, probably about 25 years old. They are well used but performed satisfactorily. Hope this helps.


Condenser, water cooled:

The condenser is an early iteration soon to be updated by a more efficient heat exchanger-condenser. This vessel was cheap and easy to construct. It is an old refrigerant bottle, while the outer is an expired LPG cylinder. A water hose is connected and overflow is the larger white hose to the right. An adjustable valve maintains the water temperature and so too the vessel temperature. The water flow needs to be such that it quenches the hot vapor flowing from the retort, thereby condensing it to liquid. Currently, I'm aiming to maintain 80C in this vessel, so that any lighter fraction that has a lower boiling point will evaporate and travel to the next condenser. There is a thermocouple temperature probe in the top though at only 80C, a simple thermometer would suffice.
There's a drain tap in the bottom to allow fuel removal.

Heat expansion considerations:

Heat of the turk-retort and reflux means that allowances need to be made for expansion of the vessels and pipework, otherwise the risk of fracture and cracks compromise safety. In the first pic is the support for the reflux and pipe leading to the 1st condenser. It is adjustable for height to take the weight but does not hold it totally captive.
The turk-retort holding down structure allows for expansion with springs.


3 stage forced draft fan:

A simple but reasonably powerful kitchen extractor fan was found and adapted for the air delivery needs of the turk. It has 3 speeds and runs on mains 240v. Also there is a butterfly in the 3 inch air tube leading to the turk however I've not found the need to use it. The plastic fan body was such a shape that it needed mounting to properly direct air flow, so an old PC casing (purple item) was screwed to the base. This fan also has a cover which is the other part of the PC casing though in the next revision, I'll move the condenser from being directly above.

Turk burner design example:

This is an example of a turk burner. The outlet can be at the top or side. What's critical is the feed should be such that the flame becomes turbulent inside the housing and resulting in efficient combustion which saves fuel. These can howl and big ones can generate quite a bit of volume

Carbon build up: 

Picture of the temperature probe after a run. The carbon could potentially block pipework so care must be taken to clear deposits.

Sight glass:

Sight glasses are a wonderful way of monitoring drip rate. The one in the pic is an example I have used though I regard as somewhat small. In the revision under construction, I want at least a 2 inch diameter. As boil-over is a constant enemy, sight glasses are important indicators of flow/drip rate. Flow would be one gauge of evaporation/condensation. Dirty fuel flowing through would signal a boil-over condition. Depending on location in the vapor stream, a good choice of glass material is said to be Pyrex (Borosilicate).

Redundant electrics:

 Turk inlet spark plugs.(redundant)

 Fan, MOSFET circuit breaker and Omron timer.(redundant)

Flasher unit to provide a separate circuit for spark via 12v coil.(redundant)
Update 28Mar2014
The burners and servo featured here are now redundant.

The non-condensable gases plus gases that evaporate from the diesel reservoir will be burned as heat for the retort. To facilitate  gas control, a simple ball valve will be controlled by a servo which in turn is under Arduino/PID management. The ball valve has the interference fit adjusted so the ball can turn with little effort. I dismantled the valve itself and added a shim washer to achieve the desired fit. Even so, the servo can deliver up 11kg per cm force.

A simple 5v-7v servo commonly used in RC helicopters is mounted carefully and some insulator board recycled from an old power meter box is used to insulate heat between servo and piping.

MG995R servo Toward Pro
Compliant with most standard receiver connector: Futaba, Hitec, Sanwa, GWS etc. .
  • Power Supply: Through External Adapter.
  • Stable and Shock Proof
  • Operating Speed : 0.17sec / 60 degrees (4.8V no load)
  • Operating Speed : 0.14sec / 60 degrees (6.0V no load)
  • Stall Torque : 9.4 kg-cm at 4.8V
  • Stall Torque : 11 kg-cm at 6V
  • Operation Voltage : 4.8 - 7.2Volts
  • Gear Type: All Metal Gears
  • Temperature range: 0 ℃-55 ℃
  • Dead band width: 5us
  • Servo Plug: JR (Fits JR and Futaba)
  • Color: Black
  • Item size: 40.7*19.7*42.9mm
  • Servo weight: 55g

  • 27Aug2013 An arm for the valve was fabricated from 20mm x 3mm flat steel bar. The connecting rod is made from stainless welding wire. It has been tested running under Arduino-PID control. Yellow paint marks indicate F=(full open) and C=(closed), 90' of travel.
    There remains some finishing off work e.g. Trimming of the valve arm, clips for the rod and possibly a mounting bracket.

    31Aug2013 The above pic is the servo valve mounted on the turk intake. I fitted a brass plumbing inline Y-strainer as a flame trap and an isolating valve for emergency stops. Also some water traps to drain the troublesome bubbler water.

     A PC power supply (PSU) is fitted so that 12v, 5v and 3v are available.

    A bright LED gauge for the reflux thermocouple provides at-a-glance temperature readings and replaces a handheld unit. It is powered by 12v from the PSU 12v rail. It was bought from Canadian based procon-products via eBay.

    Burner nozzle (redundant)
    26Jan2014 A new design of gas burner is being trialed using the venture principle. The tapered section of tube was recycled from an old table leg. The small bellmouth section was lathed from solid round bar.

    A hosetail was lathed directly onto the jet tube and (currently) a .5mm hole was bored as the orifice. Adjustment for air is facilitated using the threaded nut/washer.

    Inspiration and ideas came from this website: http://www.duncanshearer.co.nz/kilnplans/burners.html
    PID controller:
    This PID controller has been discontinued and is no longer used on the Orion plant. The arduino it is based on will be used for a system of alarms and warning buzzers.

    The temperature control process needs are handled by Arduino and Front-End V_03 on a Java framework. A PC is used to monitor and adjust the various parameters & setpoints  which is connected via serial com7 and USB. A K type thermocouple (TC) is used for temperature signal via Max31855 breakout board. This feeds a digital signal to the Arduino mega 2650 where PID calculates the response in a 0-255 PWM value. This instructs the servo which it turn controls the gas feed valve.
    The servo uses 3 wires: 5v+, ground and signal from Arduino pin9.

    Sample of Front End V_03 interface
    These are the links to get the software and components required to run PID.
    Front End V_03 http://playground.arduino.cc/Code/PIDLibrary Java framework
    Processing framework http://processing.org/download/  Stable version 2.0b7 , use 32bit only
    Max31855 breakout board http://www.adafruit.com/products/269 For the thermocouple (TC)
    A proportional-integral-derivative controller (PID controller) is a generic control loop feedback mechanism controller widely used in industrial control systems. A PID controller calculates an "error" value as the difference between a measured process variable and a desired setpoint. The controller attempts to minimize the error by adjusting the process control inputs.

    The software is currently causing the XP PC to freeze. In short Java.exe uses 100%cpu caused by msvcr.dll. I'm going to try a Windows7 PC in the hope for a workaround.
    1 May 2013
    I'm making excellent progress. I have imported two servos from Hong Kong via Ebay.
    I have one working on PID control on the bench using the TC signal.

    Here is Arduino with the servo. The thermocouple amplifier is mounted on strip board now. I have left room for two more thermocouple amplifiers for datalogging purposes. The leads that run away to the lower left are the two thermocouple wires. The servo will require its' own power supply otherwise it can run directly from Arduino/PID
     7Sept2013 The gas jar was completed today. It uses the displaced water principle. Effective capacity is about 150liters. I'll wait till the concept has proven itself, then look at whether a bigger one is worthwhile. Alternatively, ganging up a series of these would increase capacity.

    21April2014 Here is a sneak preview of the prototype retort level indicator device.
    Made from 8mm steel tubing with brass fittings and taper seat valve. A 12mm bolt was through-bored and tube was brazed in. A matching 12mm nut was welded to the retort flange and sealed with a copper washer. The tube protrudes 150mm down into the retort at the desired "full level" point. There is no rubber or heat perishable parts.

    As I write it's untested though the principle is that either gas or liquid will squirt from the tube from residual retort pressure with the valve open. The oil feed pump will be adding to back pressure also a full gasjar with add approx .5 PSI. Raw liquid feedstock ejecting will signify a level up to the mark. Gas vapor will signify otherwise... I hope.
    Am planning a run of about 12hr tomorrow so we'll see how it works.
    30April2014 update: The device performance was somewhat less than satisfactory with confusing results. It seems that vapor exiting the pipe, turns to liquid in the pipe itself. I need to improve this or make something else.
    A study of refractory is currently underway. There are quite a few options and I prefer refractory over earlier prototypes where I used kaowool.

    True to form, I'll probably elect to go with the most affordable solution that gives good results.

    Vermiculite, perlite, pumice, recycled firebrick and clay are on the shortlist. Also binders sodium silicate and calcium aluminate. I like the idea of making the brick as they can be made crescent shaped to fit the turk lining. Otherwise dry fill refractory insulation captive between sheet steel layers.

    18June2013 I have located a good cheap supply of perlite and vermiculite so this will be the insulation, sandwiched between 300mm diameter inner and 550mm outer. Therefore I reason that the refractory insulation will be a layer of approx 125mm.
    I propose to use additional insulation material around the 550mm outer. Because the vermiculite will take the brunt of the heat, subsequent insulation can be a lesser grade product such as the fiberglass wool material recycled from old stoves.

    24July2013 I'm getting closer to needing the refractory insulation. It's a day trip to pick up so I'll tie it in with some other jobs to do in the area.

    22Aug2013 I flashed off the turk inlet and exhaust ports where they go through the outer layer. The main task of the flashing is keep the loose refractory insulation from spilling out.
    23Aug2013 The tabs to hold the two drums central to each other. There are four. I could have simply welded the two drum together but this ensures the structure can be more easily dismantled should the need arise. (Update: In hindsight, not welding these drums together was a good idea because the inner section expands more than does the outer. The drums pull apart a few millimeters)
    30Aug2013 I bought 3 x 100liter bags of grade 4 vermiculite insulation. My calculation of 267liters will leave some left over which may work as catalyst. Cost was about $140
    19Sept2013 I filled the insulation layer with the vermiculite and discovered that there was a shortage. I was going to need at least another bag to finish. I believe settling of the bags' contents has reduced the volume, so to finish the assembly I used all my kaowool from earlier prototypes. This kaowool was suffering the effects of heat and was hard and baked in places. So, the retort is ready.....
    11June2014 I found some old weights from a vintage water pump in the garden which I plan to use as counter balance. I cleaned and painted them.
    9June2014 The decision was made to make a leveraged counterbalance scale so the retort level could be determined by weight. The proposed system will use a hanging balance on the steelyard principle with a leverage ratio of about 10:1.
    10Jun2014 After testing the sensitivity, I find that a lesser ratio gives better results. I'm planning to use about 20 - 25kg balance weight at about 6 - 8:1 ratio. Should the device work nicely, possible refinements could be a microswitch to operate the feed pump. Additionally, a sliding weight set up could be used to identify a full and empty retort. As I write, it's early days for the concept to prove its' worth however it appears that the 3 main pipes coming to/from the retort/reflux do not significantly interfere with the new balance scale.

    Reflux lid with adapter for counterbalance arm.

    Weights, stacker rod and counterbalance arm.
    Counterbalance arm, made from a bicycle crank and old water pipe.
    This will solve a lot of problems if it works.....
    4July2014 More pics of the counter balance scale...

    K-type thermocouple (Update: grounded type so not suitable for the electronics used on the Arduino controlled servo temperature controller, also doesn't work with LCD displays).
    5July2013. The retort flange and endcap were welded on. I borrowed a more powerful welder and used 3.15mm mild steel electrodes. The turk air/gas inlet was prepared for welding. This is angled to encourage the swirl of the hot flame front in typical turk style. Note that the spiral finning at the lower end is cut short so the incoming flame can swirl around the vessel as many times as it wants.
    Before adding the insulation layer I will run some preliminary tests to establish how effective the design is likely to be. Tests seem good so I'm proceeding.

    18July 2013. The turk powered retort progressed to the "standing on its' own feet" stage. 75mm inlet and exhaust (above) are 180degree oriented, albeit 1.5m apart. Next is the cladding and vermiculite insulation.

    25July2018 More important steps made since last update. The outer cladding is being constructed using 200L drums parts.

    Upper cover
    26May2014 This bucket lot of terracotta catalyst is how it comes out of the reflux tower. It is completely coated with carbon. It appears that it helps to clean the diesel perhaps as a molecular sieve.
    This is what the terracotta/earthenware/brick pieces look like before going in.
    13Dec 2013
    The pic shows a pipe-in-pipe which forms the final heat exchanger run, then the feedstock drops into the reflux, falling downward to the retort.
    It became apparent that a major limiting factor to production volume was the oil in-feed temperature.
    A turk style burner will preheat up to 20 liters. A gear pump will be used to pump the preheated oil to the retort in batches.

    The unit is made from recycled 9kg LPG cylinders. These are degased with hot water and soap suds before cutting. The inlet and flue are 3 inch. WVO will be used to fire them.

    11May2014 The preheater was ready for testing. It's fitted with an adjustable drip feed and uses a forced draft fan. It was run up to 170°C dry.

    25May2014 A simple flashback arrestor was recycled from an old fuel strainer. The gauze (top left) was packed with pot-scrub. This will be used on the vapor the preheater produces before being fed to the burner.

    30May2014 Reduction gearing for the gear-pump giving approximately ratio of 28:1
    This equates to about 50rpm on the assumption that the motor is 1400rpm. The object is to match inflow of preheated oil to retorts' "speed of evaporation".
    The motor pulley is a standard vee belt type driving to a 16" bicycle wheel. The bicycles fixed sprocket pairs up to an old pedal crank. Photos shows the set up with the chain guard removed.

    11June2014 The preheater was a disappointment and I don't think it's a good use of the energy. The pump also failed to deliver on at least one occasion. This is probably because the pump itself was not low enough when compared to the preheater. In other words it didn't have a reliable enough head.
    The main problem keeps coming back to knowing when the retort is full or not so the new counterbalance scale will be tried along with a converted electric water heater in place of the preheater.
    Hope to be testing the new theory and equipment later next week.
    My centrifuge in the making started out as a torque converter. I've parted it on the lathe and hollowed the internals and mounted it on a spindle. I plan a v-belt and pulley drive set up powered by single phase electric motor. The pulley ratio will be adjusted till good results come but will be a compromise between speed, g-force and vibration..

    The conversion table shows the g-force using the calculation of rpm and rotor radius. The rotor measures 133mm (5.25") radius.
    The 2850rpm motor kept tripping the overload while trying to wind up to speed using 3:1, 2:1 & 1.33:1 ratios. Another 1400 rpm motor tested out ok though rotor revs was 2000rpm. The main parts are made.
    Higher gearing didn't work with the first two motors so I fitted a 2hp 2880rpm motor initially with 1.28:1 ratio increase giving 3700rpm equating to 2000g.
    I've decided to mount the centrifuge on a 200L drum as it will provide a tank for the processed oil plus eliminate the need to make legs.

    11May2013 I test run the centrifuge yesterday. There were a few problems to solve.

    The waste oil drain wasn't big enough and it overflowed when the motor was shut down. This covered the drive belts in oil and made a mess with the 1 -2liters of oil. The solution was to decrease the size of the 2 drain holes in the rotor plus make another two drain ports in the outer housing, each with their own hose connection to the waste tank.

    The small outer 'fins' on the rotor were ground off smooth to help reduce air turbulence.

    The oil feed tube and fittings were enlarged by drilling. I may regret using only 8mm but diesel weight oil may well flow a lot more. Another bigger feed line will be added should extra flow be required. The tube was shortened due to vibration affecting it.

    13May2013. I had another re-think of the project and have fitted a 1400rpm motor due to the rotor being too far out of balance. It was a shame because the dry balance was good but once the 2 liter oil payload is added, the vibration is too much. The unit is now much quieter and smooth. I feel Ok leaving it for an hour unattended. Thinking about failsafe, if the motor burnt out, threw a belt or power cut, oil feed would be contained in the dirty oil catch container so spills are avoided.

    A sample of sludge from running the centrifuge !!

    24May2013 I'm not satisfied with the balance so I'm now going to remake the rotor from large steel pipe and plate. Once fabricated, I'll spin the new rotor on the lathe to ensure it is dead true.

    7June2013 In a recent run of the centrifuge, I found that the device is much more stable and vibrates less when it is plumb-dead-level. It seems so obvious to me now that I've had time to think it through.
    I still plan to press on with a remake of the rotor as soon as some suitable plate steel comes to hand.

    28Oct2013 I'm fitting a new drum as the clean-diesel-reservoir. The centrifuge will sit on top as before though now the fuel will pass through a 1micron bag as it falls into the drum. The filter should last a long time given that the oil is stripped of particles trapped with the centrifuge. In addition I want to process the diesel while it's still warm so to increase the centrifugal effect on the residue. A short stand has been made to raise the drum so a drain valve can be accessed. The pic shows the valve and threaded nipple brazed to the underside.

    This sludgy stuff came out of the diesel!!

    21Dec2013 A remake of the rotor is partially complete and it's proving to be quite a mission. It weighs about 8kg at present. Pics to follow.

    10Mar2014 Ok, it took a while but the new rotor is complete. It is fabricated from steel plate and pipe. It was turned in my home workshop lathe to be true. My fuel is being cleaned as I write...!

    1April2014 I've been using the 'fuge and now I realize I need a much more solid platform for mounting it. The relatively flimsy framework creates vibration harmonics which will fatigue the unit unless I address the problem. Also the vibes upset the rotor slightly to the point where it can "spill" product prematurely. I will make a very heavy foundation with facility to accurately level the rotor.

    2March2015 The centrifuge project has been officially abandoned in favor of extended settling using an upflow-settler system. Choosing settling over centrifuging came down to effort, energy, effectiveness and economics. The centrifuge required multiple passes consuming a sizeable amount of power though it had the benefit of being fast turnaround. The upflow settler uses no energy but takes up to 2months. Either method was deemed effective if it did not plug downstream filters.
    I compare the 2 systems like this:
    Centrifuge exerts say 3,000 times gravity for a few minutes.
    Settling uses 1 x gravity but applies it 30,000 times longer..