Purpose:

The purpose of this section is to describe a fuel injection system that is being developed for the Austin A series engine. I have picked the A series because, 1) I have one, and 2) it has some particular quirks that make it difficult to inject. I know of one injection system for the A series engine, the one from Rover. There does not seem to be any aftermarket systems that will handle the A series. This is not surprising as the issues are large, the solutions are complex, and the market is small.

This system should be applicable to the Austin/Morris B series, some Triumphs, and some Studebaker engines as well, since they have the same issues. The issue is the design of the inlet ports in the head of the engine.

The issue:

The A series engine has the feature of a "siamesed" inlet port. A siamesed port means that a single port is used to feed two cylinders. The A series has two intake ports for four cylinders. They are grouped 1-2 and 3-4.

The problem:

The problem with the Austin A series engine is that the head has siamesed inlets. It causes scavenging of the inlet charge by the other cylinder in the pair.

Many will attempt to convince you that this is not a problem since 1) they have not actually thought the problem through, and 2) fuel injection systems are made of JPBM (just plain black magic), they can defy the laws of physics, feed the hungry, and give you an infinite amount of power. They will say that aftermarket fuel injection systems manufacturers are smarter than you are, they have thought this all through, and you're just not smart enough to realize it. BULL $#!+.

Let's explore the problem and try to understand it.

The intake order is just like the firing order, on the A series this is 1-3-4-2. The intake order has to be the same as the firing order, it just precedes the firing order by about 360 crank shaft degrees. So, let's take a look at what the order tells us, first 1 intakes, then 3, then 4, then 2, then back to 1. Big deal! That means that the 1-2 port intakes, then the 3-4, then the 3-4, then the 1-2, then back to the 1-2. Another way to look at it is 1-2, 1-2, 3-4, 3-4, which is just the same sequence 180 degrees later. I call this an AABB pattern. The problem comes up when you look at the intake order as 2-1-3-4 (which is the same as 1-3-4-2 just starting at 2 instead of 1). We see that 2 is on intake just before 1 is on intake.

2 and 1 share the same intake. Therefore you have two intake cycles on one port, then 2 intake cycles on the other port.

One thing that people miss is the fact that most fuel injection systems store fuel mixture in the intake runner while the valve is closed. Think about it, the maximum injector duty cycle (percentage of time that the injector is open) is 80%. How are you going to even come close to that when the valve is only open about 260 degrees out of 720? That is only about 37%. The answer lies in storing fuel in the intake runner. When the valve opens, it takes in the mixture that was stored in the runner as well as what is coming from the injector currently.

Now, look at what happens in the A series – if we store the charge in the intake port, valve 2 opens (it precedes 1), and 2 takes in the stored charge. Next, 2 closes and 1 opens, there is nothing stored in the intake runner, only what the injector is currently spraying. The same thing happens with 3 leading 4, 3 takes the stored charge and 4 gets very little. If we tune the injection based on 2 and 3, 1 and 4 will be very lean. If we tune for 1 and 4, 2 and 3 will be very rich.

The solution:

The solution to this situation is diverse, 1) to agree that there is a problem, 2) have a homogeneous mixture available at all times so that any mixture will be of the proper mix, and 3) only inject when the valve is open and don't store charge in the intake runner.

2) above is known as running a carburetor or throttle body fuel injection. All of the air in the manifold is mixed with fuel so that it is homogeneous. This leads to a situation known as a wet manifold. The problems with this approach are well known, manifold icing from the evaporation, and poor mixture predictability due to mixture condensation on the manifold walls. The solution to icing and condensation is manifold heat with the charge density decreases that come along.

3) above is a form of sequential fuel injection with a large twist, you cannot store charge in the inlet runners. Though storing charge for the inner cylinders is possible since it will take everything stored, but you cannot store charge for the outer cylinders. The big difference between classical sequential injection and what is needed here is timing of the injection to occur only when the valve is open.

The following is a little bit of simple math to guide you through the problem:

The amount of fuel that is required to be injected in each intake cycle is related to the air/fuel ratio. For gasoline, the stoichiometrically correct air/fuel ratio (the ratio where all the fuel is consumed by all of the air without any surplus) is 14.7:1. This means that you have 14.7 grams of air for each gram of fuel. (use grams, kilograms, or pounds).

Let's assume a four cylinder, 1000cc engine for the next part. Each cylinder (assuming a 100% volumetric efficiency) would require 250cc of mixture. Air has an agreed upon density of about 1.2929 kg/m**3. 1 m**3 is 1000 litres ((100cm/m)**3) / (1000 cm**3 / L) So, air has a density of 1.2929 kg / 1000L. 250cc of air is 0.25L. 1.2929kg / 1000L = x / 0.25L solve for x. x = 0.000323225kg this would be 0.323225 grams of air per cylinder.

So, one cylinder of air weighs 0.323225 grams. To consume that air we would need 0.323225/14.7 grams of gasoline, or 0.02198 grams of fuel.

Fuel has a density of 803 kg/m**3. Which is:

803kg / 1000 L

0.803 kg / L

803 g / L

0.803 g / cm**3

So 0.803 g / 1 cm**3 = 0.02198 g / x solve for x. x = 0.02738 cm**3

To fire our 250cc cylinder, it would take 0.323225 grams of air and 0.02738 ccs of fuel.

The engine fires the cylinder once every OTHER rotation of the engine. So at 1000 RPM the injector injects 500 times per minute.

At idle, say 500 RPM, we inject 250 times per minute. 250 rotations/ minute * 0.02738 cc / rotation = 6.85cc/minute. At high speed, say 8000 RPM, we inject 4000 times per minute. 4000 r/m * 0.02738 cc/r = 109.53 cc/minute.

Both of these numbers are well within the capabilities of regular injectors.

But if we only have one injector per intake port, this would mean one injector doing two cylinders, which would mean 13.69 cc/minute to 219.06 cc/minute, which is still doable.

So throttle body placement with 2 250cc/minute injectors would probably work, and would be able to handle about 100HP.

For sequential port fuel injection, all of the fuel has to be injected while the valve is open. If we assume a cam with a 266 degree opening time, an instantaneous opening, and being non-restrictive, we see the following:

At idle, 500 RPM again, the engine goes through 360 degrees in 1/500 of a minute or 0.12 seconds. A 720 degree cycle goes by in 0.24 seconds. Our 266 degree window went by in 720/0.24 = 266/x x = 0.0886 seconds.

We have to inject 0.02738 ccs of fuel in that time.

0.02738 cc/ 0.0886 sec = 0.31 cc/sec = 18.53 cc/minute. Quite doable.

At high speed, 8000 RPM, the engine does 360 degrees in 1/8000 of a minute or 0.0075 seconds, 720 degree cycle in 0.015 seconds. Our 266 degree window is 720/0.015 = 266/x x = 0.00554 seconds.

Inject 0.02738 ccs. 0.02738 cc / 0.00554 sec = 4.94 cc/sec = 296.47 cc/minute.

But this is for one cylinder. Two cylinders, 37 cc/minute at idle, 593 cc/minute at high speed. That's a big injector.

The amount of fuel injected per cycle is pretty much constant, since it all has to be injected while the valve is open, as long as the injection time is less than the valve open time, and greater than the time that it takes to open the injector it should work.

The other restriction is that the injector must have a duty cycle of less than 80% (according to FI lore). At idle the open time is 0.0055 seconds every 0.24 seconds or 2.3%. At high speed the open time is 0.0055 seconds every 0.015 seconds or 37%.

So much for theoretical engines, in real life the injector takes about 1ms. to open, the cam does not open the valve immediately, we don't drive around with the throttle wide open all of the time, and volumetric efficiency is not 100%.

Volumetric efficiency less than 100% and throttle less than 100% will reduce the fuel requirements, so that will make it easier.

At idle, there is more than enough time to open the injector since there is a 89 ms. window to do the injection. At high speed, the 1 ms. opening time cuts into the available 5.5 ms. injection cycle. If we subtract the opening time from the total time that is available for injection, we have to alter the calculation above so that the 0.02738 cc charge is injected in 266 degrees at 8000 rpm MINUS 1 ms for opening. 0.02738cc / 0.00454 sec = 6.03 cc/sec = 361.75 cc/minute.

That delay in the opening time just boosted the required flow by 22%!

At 8000 rpm, in that same 1ms the crank rotated by 96 degrees, the cam by 48 degrees.

We are starting with 250cc of air then adding fuel, doesn't that make more than 250cc? Well yes, but the difference is 0.01% difference. Remember we are talking about 0.02738cc of fuel in 250cc of air. The difference is ignorable.

To avoid charge stealing, I propose to have the injector open as early as possible on the leading cylinder (2 and 3) and as late as possible on the trailing cylinders (1 and 4). This avoids the possibility of charge stealing during overlap, and gives the injector time to close and reopen.

By storing charge for the leading cylinders, we can avoid the problem of an open-close-open cycle in quick succession. This gives the injector plenty of time to close before it needs to be reopened for the trailing cylinder (which must be injected while the trailing valve is open).

Conclusion:

So, there's the problem, the siamesed A series head forces us to inject the complete intake charge while the inlet valve is open AND do it with two injectors. The result is that the injector needs to be more than 6 times as large as one would expect.

Other options would be to use 4 injectors but still use the idea of injecting the complete charge while the inlet is open. This would not reduce the size of the injectors, just alter the opening pattern.