Porsche 12 Cylinder Fuel Injection

[Roger B]

The possibilities for fuel injecting Dave’s Porche 917 engine have been discussed in various places at various times so I thought I would try to bring my thoughts on the various concepts together in a new discussion thread.

The full size engine had a diesel derived Bosch mechanical injection system. There were two rows of 6 helix controlled pump elements with a central control rack. This required left and right hand helixes. (picture 01) I have seen the injection pressure quoted as 17 bar.

This may be technically feasible to model but approaching a scale size would be difficult. A means of adjusting the output (helix angle) of the individual elements to ensure that the same amount of fuel was delivered to each cylinder would have to be devised.
As the injection timing is not critical as in a diesel a simpler wedge controlled pump may be a better option. This was used on an existing model Porsche 917 engine (pictures 02, 03). I have some designs for wedge controlled pumps but have not bult and tried them (picture 04). One open question is the fuel inlet system. Using a port in the side of the pump cylinder gives clean closing but requires a pressurised fuel feed to avoid vaporisation. If a ball valve is used the ball needs to be closely constrained or spring loaded to give quick and consistent closing. It may still be necessary to pressurise the fuel feed (~0.8-1.0 bar).

Another option is a separate metering pump and distributor. This in theory reduces the need for balancing the quantities to the individual injectors. A number of diesel systems have used this concept. A 6 cylinder unit for a 4 stroke engine requires 3 injections for one revolution of the crankshaft. This either requires a multilobe cam which must be very precisely made or a 1-3 step up gear. The full size units, as far as I am aware, all used a multi lobe cam system. The Lucas mechanical petrol injection system avoids this problem by generating the injection pressure (7 – 7.5 bar) with a separate electric pump and setting the fuel quantity with a shuttle valve in the middle of the distributor sleeve. (pictures 05, 06) I think that it would be possible to make a model version of this for up to 6 cylinders.

Electronically controlled injection is also possible. Daniel Hunziger has produced a running example, however finding small enough injectors is difficult.

<a href="https://www.youtube.com/watch?v=xqMk87vOMsw" target="_blank">http://www.youtube.com/watch?v=xqMk87vOMsw</a>

Roger Moore made his own injectors for a ¼ scale Ferrari Type 049 Formula 1 engine. Quite a challenge and you also have to produce some form of ECU.

<a href="https://www.youtube.com/watch?v=7JXXfkMclB4" target="_blank">http://www.youtube.com/watch?v=7JXXfkMclB4</a>

Another option is a continuous flow system like the Bosch K – Jetronic. The problem here is maintaining equal fuel distribution across all the injectors when they are all connected to one metering valve/system. Bosch appear to achieve this by very precise machining of the outlet slits in the metering valve and very precise machining and setting of the injectors.
Schillings produced a simplified version with the injectors fed from a relatively large diameter manifold and individually adjustable injectors which were set to give approximately equal exhaust temperatures. He appeared only to use this system with methanol fuel and glowplug ignition which will run with a wide range of fuel air ratios. (pictures 07, 08)

[Roger B]

The injectors for manifold injection can be quite simple. They need to incorporate a non-return valve so that the pipes remain full of fuel. A small orifice may be sufficient, (picture 09) although the 0.2mm version I have made may still be too large, otherwise a poppet style injector as used with the Lucas PI system should work (pictures 10, 11, 12).

I don’t see a need for the diesel style needle injector, which would also require a return piping system for the fuel that leaks back past the needle.

[Vixen]

Roger,

Mist spray nozzles, similar to your (picture 09) in the above post, are readily available on e-bay with nozzle diameters down to 0.1mm.

Search for '0.1mm mist spray nozzle', there are loads to chose from, including bare nozzles and others which incorporate a non-return valve. The nozzle inserts appear to be made of a ceramic. They are intended to operate from around 20Bar with flow rates from 10ml/min. No indication of how they would perform at lower pressures, hence flow rates.

It looks like you could easily modify the outside diameter to make them smaller.

Could they form the basis of a constant flow, manifold/ port injection system, similar to the Schillings design?

Cheers

Mike

[steamer]

Wow!

I would like to thank you both for your thoughtful effort!   I have been wondering about the electronic as well as the manifold/port injection scheme.  Thus far the electronic version was one I was rather keen on as it could be controlled with an Arduino which I believe is how Hunzinger is operating his.   The challenge has been size.    The fuel injection system on the 917 is rather iconic!   A snakes den of plastic lines leading to intake trumpet mounted injectors with a flat slide throttle.....It has a "Look" all it's own! and is something that kinda needs to be right to make it work visually, to say nothing of making it run!

I wish I had the skill and talent of Mr Moore,   His workshop equipment would rival any Aerospace company!   Winding 0.0007" diameter wire injection coils with his CNC Hardinge for instance is amazing!    Alas,   I seem to be missing that particular piece of equipment....

I have seen that model of the 917 before, and I've mentioned that it lacks this visual, and is also different in many fundamental points, like not having the gear driven output shaft like the real one, no cooling fan, ect.  That is one of the reasons I've not given that model my attention, though just getting a v12 to run is an amazing feat in and off itself and I commend the builder of that model for his/her fine work.

The Hunzinger and the "Schilling" solutions seem to be the path that seems plausible for my shop, and the "Look".

Or I could just go 2 carbs .....lol    well if this model comes down to which carburation I am going to use, I think I'll be doing just fine.

Thanks again!   And I'll check out those sprayer nozzles Mike!

https://www.hagerty.com/media/news/road-legal-porsche-917-in-monaco/

[steamer]

^^^^^^^
By the way ,  that beast of an engine was held in the car with 3   THREE   8mm bolts......

[Vixen]

Ha Ha!!     but don't forget the ring of bolts securing that engine to the transmission which itself is well attached to the tube chassis.

Mike

[gerritv]

To throw another injection option into the fray, the Tecalumet system as used on e.g. BMC mini's :

[Vixen]

Hello Dave,

Twelve individual injectors will dominate the engine, you just cannot make them small enough for them to disappear from sight.

I had a thought, that you could maybe consider reducing the number to four injectors, if we treat the 917 injection system as four, three cylinder engines. Each injector firing sideways into a single plenum chamber and throttle body, with three inlet trumpets above, then feeding down into a group of three cylinders. The injector pumps being based loosely on the Hunzinger and the "Schilling" solutions.

I went further and used Google Ai to do some calculations:

First, the air to fuel ratio (AFR) for methanol glow plug engines from max lean to max rich and max power. The wide FAR range of methanol fuel is amazing.
Then, the actual flow rate (quantity) of methanol fuel that needed to be delivered (at full speed) to the single injector feeding the group of three cylinders. Now have a good handle of the required ml/min.
Then, I asked about the fuel rail pressure needed to flow that quantity of fuel through 0.1mm, 0.15mm, and 0.2mm spray nozzles I mentioned in an earlier post. The smallest available 0.1mm nozzle shows a lot of promise.
Finally, I asked about the  ideal air intake throat diameter for each of the three individual inlet trumpets. the suggested 5.0mm choke diameter does not seem unreasonable for a single 6.5cc cylinder.

I am still analysing the responses; There is a lot of detailed calculations and information to check; however the use of a 0.1mm spray nozzle at about 2.0 bar fuel rail pressure looks promising at full speed and also at a fast tick-over at a reduced pressure.

The problem I now have is I cannot simply cut and paste the full Google Ai responses due to formatting issues. I may be able to summarise the findings, if there is sufficient interest on the forum ( fuel injection is not a popular flavour with our steam engine brethren).

We all know by now, how careful you need to be in phrasing the correct question to an Ai bot, I have made a first attempt.

Cheers

Mike

[Roger B]

I have translated some more about the Schillings system and it's set up procedure. His injectors fit into a M5 threaded hole in the inlet manifolds.
---------------------------------------------------------------------------------------------------------------------------------------------------------------------

This is translated from the V12 DB 603 build notes in Hubert W. Schillings book:
Boxer-, Reihen- und V-Motoren als Modell published by VTH.

Intake manifolds, injection system

Pages 59, 60.  Diagram on Page 14.

The injection system requires a bit of understanding: by way of explanation, it should be said that, in the strict sense, twelve carburettors are supplied with fuel depending on the engine speed. The wide ignition range of alcohol makes this possible. On the intake manifolds there are threaded holes for the injection nozzles. In reality, these are the adjustable main jets, because the fuel circulating in the system is sucked off here! The piping supplies the twelve nozzles via a manifold at low pressure; the pressure is set so that the nozzles are just wet - they must not drip. Two return valves ensure that the fuel does not flow back and thus allow air to enter the line system.

The system is very simple in design and works very reliably if all rules are observed and all screw connections are tight. The amount of fuel required for good running forms a curve that rises with increasing engine speed, i.e., it is not linear. A gear pump can never be so precise that it delivers a low-viscosity medium evenly from the very start. As the speed increases, the efficiency inevitably increases. In this way, a smaller amount automatically results for the idle range. The leakage therefore acts like the compensating jet in a carburettor. At higher speeds, the mixture automatically becomes richer. Thus, the running behaviour of the engine with model is purely a matter of adjustment.
The actual installation of the injection system is carried out as the final task once the engine is fully assembled. The two intake manifolds can be machined according to the drawing. The supplied nozzles do not need to be installed yet; one merely checks as a trial whether they can be screwed in.

Test Run

Pages 61,62

Before the first start, be sure to provide the engine with a firm stand so that it won’t move. The fuel container and oil tank should also be properly fixed. We know that one eagerly awaits the first pops and bangs and can hardly wait for the machine to run. However, one must restrain oneself and do everything for one’s own safety and that of the equipment. Ensuring the ignition current is also important. We did this using a transformer that we built ourselves. An on/off switch is a must. By the way: current should be measured; a visual check of whether the plugs are glowing is primitive. Everything primitive should be forgotten when doing this work. A second person is also required. Hand signals for communication are agreed upon beforehand.
You may consider these notes to be rambling; that is your business. But we would like to point out that around 20 hp are normally released. The propeller will cleanly chop off an arm if it gets in the way - quite apart from the damage to the engine.

Now for the start. First, we make sure (transparent connecting hoses are very helpful) that lubricating oil is flowing. Then the fuel system must be bled.
We did it like this: (of course with the ignition switched off) we held the fuel tank up, turned the engine over, and opened the nozzle fittings (the nozzle needles remain closed for now). Then, one after another, we tightened the fittings as soon as fuel emerged free of bubbles. Keep turning (do not stop), briefly disconnect the check valve after the fuel manifold until fuel emerges, then reconnect it. If both check valves seal properly, the system is bled you can start. To do this, all nozzle needles are opened half a turn, and the locknut is lightly tightened.

When the engine emits its first sounds, with the throttle valve only slightly open, give it a very brief burst of throttle. If the engine then dies, the nozzles must be opened another half turn. This procedure is repeated until the engine runs steadily at low throttle. Now check with your finger (this works well) whether all cylinders are running. Those that are not yet firing are given more fuel by further opening the nozzles until they run. During this procedure, the ignition naturally remains switched on. These adjustments require a great deal of sensitivity; especially opening the nozzles must be done very carefully.

We initially carried out the starting process with a flywheel, using a modified car starter. This gave the whole thing a bit more force. Later, the engine was started by hand with a three-blade propeller of about 80 cm diameter. This is very dangerous and not recommended!

[Vixen]

Air-to-Fuel Ratios (AFR) for Methanol Glow Plug Engines According to Google Ai

This may be of interest to all i/c engine owners and is not specific to just Fuel Injection


The question given to Google Ai was:

1 Create a table showing the air to fuel ratio (AFR) for a small four stroke engine with glow plug ignition running on methanol fuel. Show the AFR ratios for the following conditions:   a, max lean      b, max economy     c, max power    d, max acceleration and    e, max rich running conditions.

Google Ai came back with:

The stoichiometric Air-Fuel Ratio (AFR) for pure methanol is approximately 6.47:1

Small four-stroke engines using glow plug ignition and methanol fuel operate at significantly richer mixtures than stoichiometric. This richness is essential for internal cooling, as methanol has a very high latent heat of evaporation, and for maintaining the catalytic reaction on the glow plug element.


Condition                     Description                                      AFR (Air:Fuel) by weight

Max Lean              Limit before flameout/overheating            6.79:1

Max Economy        Slightly rich for safety/stability                 6.15:1

Max Power            Peak BMEP and critical cooling                  4.85:1

Max Acceleration    Extra enrichment for transient response   4.53:1

Max Rich               Limit of combustion (four-stroking)          3.56:1


Key Technical Considerations:

Cooling Requirements: Methanol engines rely on excess fuel to manage cylinder head temperatures. Running at the "Max Lean" limit in a small air-cooled engine can lead to rapid thermal failure.

Glow Plug Heat: The ignition timing in these engines is determined by the temperature of the glow plug. A mixture that is too lean will cause advanced ignition (knocking), while a "Max Rich" mixture can "drown" the plug, causing the engine to stall.

Lubrication: Small glow engines typically use fuel containing 15%–20% oil (castor or synthetic). The AFR values above refer to the total "fuel" (methanol + oil) mixture.


Interesting stuff   

Mike

[steamer]

Fascinating!

Thought.....maybe a bad one, but it got me thinking anyway.

What if  we used a 50CC or so fuel injector as the "metering pump" on the Shilling layout?    It could be used as a overall metering unit via PWM control and feed the 12 "injectors" .    I think it becomes its own checkvalve as well.   This would be fed by a in line 2 bar fuel pump that could be in the fuel tank.    The injector could go in a mocked up Bosch injector pump....the PWM could be controlled by a rpm sensing crank trigger and a supply curve could be adjusted by some software from an Arduino.


If the fuel lines leading to the "injectors" which are simple nozzles have a much low flow resistance than the injectors do, there should be fairly little flow resistance difference between them?    This is something that can be calculated by me.....been there done that.....

Sound too nuts?....

[Vixen]

Volume of fuel require by each three cylinder bank of the 917 engine, at full speed.

The question I gave Google AI was:

2  Create a table showing the amount of methanol fuel consumed by a 20cc displacement four stroke engine with glow plug ignition and a 8:1 CR running on methanol fuel under max power AFR conditions at 10000rpm, 5000rpm and 2500rpm. show the fuel consumption as ml/min.

Google Ai came back with:

To calculate the methanol fuel consumption for a 20cc displacement four-stroke engine, we use the maximum power Air-to-Fuel Ratio 4.85:1.

Assumptions for Calculation:
Volumetric Efficiency (VE): Estimated at 85% at 10,000rpm (peak power), and 90% at lower speeds 5,000 and 2,500 rpm due to increased cylinder filling time.
Air Density: 1.225 kg/m (0.001225  g/cc) at sea level.
Fuel Density: Methanol density is approximately 0.792 g/ml.
Four-Stroke Cycle: The engine intakes air once every two revolutions.

Methanol Fuel Consumption (ml/min)
Engine Speed (RPM)    Air Volume Flow (cc/min)     Air Mass Flow (g/min)   Fuel Mass Flow (g/min)  Fuel Consumption (ml/min)
10,000 RPM                85,000 cc                           104.1g                         21.5 g                          27.1 ml/min
5,000 RPM                 45,000 cc                            55.1 g                         11.4 g                           14.3 ml/min
2,500 RPM                 22,500 cc                            27.6 g                           5.7 g                           7.2 ml/min

Calculations:
Air Flow Calculation: Displacement (20cc) x (RPM/2) x Volumetric Efficiency. At 10,000 rpm: 20 x 5,000 x 0.85=85,000 cc/min.
Conversion to Mass: Air volume x air density (0.001225 g/cc). For 10,000 rpm: 85,000 x 0.001225=104.1 g/min.
Fuel Volume Flow: Fuel mass / methanol density (0.792 g/ml). For 10,000 rpm: 21.5/0.792=27.1 ml/min.

Other Operational Factors
Lubrication: These values represent the total liquid flow. In a glow engine, the fuel typically contains 15%–20% oil. This means the actual methanol portion is slightly lower, but the carburettor or injector must still flow the total ml/min shown above.
Compression Ratio (8:1): At this relatively low compression for methanol, the engine may be less efficient than high-compression variants (12:1), potentially requiring even richer mixtures for stable combustion at high speeds.


Now we have an idea of the maximum amount of fuel required to be injected at full speed. 

Mike

[steamer]

Thank you Mike!!

[Mike R]

The problem I now have is I cannot simply cut and paste the full Google Ai responses due to formatting issues. I may be able to summarise the findings, if there is sufficient interest on the forum ( fuel injection is not a popular flavour with our steam engine brethren).

Please keep the discussion going here.  I'm very interested in fuel injection, mostly just to learn but also I have been researching (and designing) fuel injection for model diesels and while not directly applicable all information on miniaturized fuel injection is welcome.

[Vixen]

Yes Dave,

That's exactly my thinking. We could use four 50cc bike fuel injectors or we could use four of the spray nozzles with a separate fuel pump similar to the Schillings DB 603 setup.


To continue my story, Pressure required by commercial atomizing nozzles to flow the 27.1 ml/min required by the three cylinder bank of the 917 engine at full speed.[/b] Somehow (it was late) I managed to type 21.5 ml/min instead of 27.1 ml/min but the results are close enough to give the general trend.

The question I gave Google AI was:

3   Create a table showing the fuel pressure required to flow the 21.5 ml/min of methanol fuel through an atomising nozzle of a, 0.1mm b, 0,15mm and c, 0.2mm diameters.

Google Ai came back with:

To determine the fuel pressure required to achieve a flow rate (Q) of 21.5 ml/min through various nozzle diameters(d), we use the discharge equation for a thin-plate orifice:   Q=Cd . A . sqrt of Delta P\rho

Parameters and Assumptions:
Discharge Coefficient (Cd): Typically 0.6 to 0.7 for small atomizing nozzles. We will use 0.65.
Methanol Density (rho): 792 kg/m.
Nozzle Area (A): Calculated as A=pi . (d/2)^2.
Pressure(Delta P): The pressure differential across the nozzle in Pascals (Pa), converted to Bar and PSI.

Fuel Pressure Requirements for 21.5 ml/min:
Nozzle Diameter (d)    Orifice Area (A)         Pressure (Delta P) in Bar   Pressure (Delta P) in PSI
0.10 mm                   7.85 x 10^-9 m^2      1.95 Bar                          28.3 PSI                    
0.15 mm                   1.77 x 10^-8 m^2         0.38 Bar                           5.5 PSI
0.20 mm                   3.14 x 10^-8 m^2         0.12 Bar                           1.7 PSI

Engineering Observations
Atomisation Quality: At 2,500 RPM (fast tickover) the flow rate drops to 7.2 ml/min. With a 0.2 mm nozzle, the pressure would drop below 0.2 PSI, resulting in poor atomisation (dripping) rather than a fine mist.
System Choice: The 0.1 mm nozzle is the most suitable for a high-pressure Electronic Fuel Injection (EFI) system, as it maintains high pressure even at lower RPMs, ensuring the methanol is properly atomised for the glow plug ignition.
Viscosity: Methanol has low viscosity, but if your fuel blend contains 20% castor oil, the required pressure will increase significantly due to the higher fluid density and internal friction.
                                                          -------------------


From all of this, I conclude the 0.1mm nozzle with a rail pressure of 2.0 Bar appears to put us into a good ball park, if we are using a Shillings style gear pump and pressure regulator. Alternatively, we can use a 50cc bike fuel injector with a similar fuel rail pressure.

Note, these calculations are only indicative and may contain errors (Ai systems are not infallible)    but may give a good basis and direction for future experimental work.  And above all, remember that this discussion is specifically aimed at low pressure manifold or port injection, It is not to do with direct injection diesel systems

Mike