As far as i know, electronic fuel injectors only know two discrete states, on or off. Under light load (ie. low rpm and throttle) the injector doesn't have to deliver much fuel. So naturally the injection time is short in order to deliver little fuel, because the fuelflow is always the same. This leads to injection only happening during a fraction of the intake stroke.

Consequently, you get a superrich part of mixture together with just air in the cylinder. These still have to mix/homogenize in the cylinder to get a good combustion. This process takes time and possibly doesn't give you the best performance. So why aren't there injectors that can vary their fuelflow? The ECU can derive the duration of the intakestroke from the rpm, and can consequently use all that time to make the injector deliver it's fuel during the whole intakestroke. This way, the mixture that enters the cylinder is almost completely homogenized already.

What do you folks think? Stupid or smart idea, or does it already exist? I understand that designing new injectors that can do this(if they don't exist already) costs money, but in the neverending battle against emissionregulations constantly demands more innovation.
Any input is welcome.

  • I'm not exactly sure where the injector fires during the intake stroke, but would bet it does it in the early part of it. This would allow for the best homogenous mixture I'd bet. @Zaid may have a definitive answer for you. Commented Oct 31, 2016 at 21:14
  • @Pᴀᴜʟsᴛᴇʀ2 at idle (~700RPM) roughly 30ms or so before the spark event. so right at the start of the intake stroke I believe. and will fire for around 3ms or so assuming fuel trim is normal.
    – Ben
    Commented Oct 31, 2016 at 23:48
  • I'm pretty sure your suggestion and what exists currently achieve the same result. I'd imagine that an injector with varying steps to the pintle would be a degree more complex and require a larger package etc...
    – Ben
    Commented Nov 1, 2016 at 0:56
  • Did you have port injection or direct injection in mind when you asked this question? In the case of port injection, the air and fuel mix before they make it to the intake valve. Direct injection systems are quite capable of running in homogenous mode, so I would say that the concern of poor mixing highlighted in your question is addressed by both scenarios.
    – Zaid
    Commented Nov 1, 2016 at 10:26
  • @Zaid I had (multi)port injection in mind, but the same might count for DI. You say they properly mix before they get past the intake valve, but that's what my doubt is. My guess would be that they still need some time to complete the process of homogenization in the cylinder, because the mixing in the intake manifold wasn't 100% complete. I base this on the fact that they also use advanced piston surface shapes to enhance swirl and homogenization. That'd meand there's actual benefit in further homogenization, otherwise they wouldn't do that.
    – Bart
    Commented Nov 1, 2016 at 10:38

4 Answers 4


Keep in mind that at high loads/WOT, the injector may be open nearly constantly - during all four strokes. (Injector Duty Cycle >90%)

In fact, many systems switch from sequential port to bank and then continuous fire (all injectors firing at the same time) when high fuel demands are called for.

The difficultly is balancing WOT needs with idle with only one injector. Big pintles and nozzle orifices mean bigger flow, but also add mass and latency to the operation. In addition, atomization efficiency is drastically affected by critical orifice size, and not as much by pressure drop. A firehose is great for fighting fires, but not as useful when adding a drop of water to your Scotch.

I'm not aware of a "variable" injector that would be cost effective in all modes. However, new GDI technology obviates a lot of those needs, and also has the benefit of not injecting on to a cold valve and destroying nice small droplet size. It also has the beauty of not caring about valve/cam lobe position. You can inject a very lean charge, and then a small pocket of stochimetric mix immediately before (and maybe even during!) ignition.

So my quasi-answer is that I can accomplish so much more with Gasoline Direct Injection, that although your idea is salient, I think those IC wizzards have already moved well beyond such ideas. In fact, they are just scratching the surface. I was invited to a webinar called "Dr. Stochiometric is Dead". I couldn't afford it, but my point is I fully expect IC efficiency to improve dramatically in the next five years. Just gotta deal with heat and NOx... Lean is the new phat!

On edit:

To clarify, I only mean that an on/off style injector that is optimized for idle may not have enough ultimate flow for WOT. Albiet WOT is rare (well, for some... I won't specify what category I serve time for) the problem is you just can't go wicked lean without terrible consequences.

A good compromise might be a small-pintle small-orifice injector for idle, coupled with another injector: my proverbial "firehose" for hole shots. This, of course, would be problematic in engineering and production expense.

This is why I'm such an advocate of GDI, which allows a huge flexibility in injection schedule, regardless of valve position or which stroke you're on.

The idea of a variable-pressure fuel rail is laudable, if it only worked. As I mentioned before, when you look at optimized atomization based on a "critical" orifice, there's some strange physics involved. Orifice size rules, and pressure drop becomes almost fixed. Once an orifice is "critical", increasing feed pressure does not have the linear effect one might hope. (Diesel injection aside; that's a wholly completely different animal...) In addition, I'm not ready to admit (as if I understood!) that injection during an entire intake stroke leads to more homogeneous mixture (after compression) than just a short spray.

Idle emissions are at an all-time low. In Los Angeles, tailpipe emissions at idle are lower NOx values than the intake air.

Look to GDI and some crazy wicked power transmission technology to change the rules. Run your IC in a sweet torque band, very lean, and allow a seamless power transmission with low losses to allocate the torque. Save stochiometric for stoplights and when I'm trying to impress a girl...

(No, I'm not buying a Nissan... In my opinion they are virtually undrivable and too frustrating to comprehend...)

  • GDI sounds amazing in a lot of ways but the only downside is the carbon build up
    – method
    Commented Nov 1, 2016 at 2:56
  • How you decribe GDI sounds the same as a combi of stratified and pilot injection to me, where you build a relative rich mixture around the sparkplug to easily ignite the leaner base mixture in the rest of the cylinder. Do these two differ in nature? I don't get your point about WOT, a situation that you don't have very often. I'd say fuel consumption and emissions at that point are inferior to getting power. It's the light load conditions that should be optimized for emissions and fuel consumption IMO.
    – Bart
    Commented Nov 1, 2016 at 9:18
  • It's very plausible that 'they' don't implement certain innovations yet, for the sake of being able to directly implement them, off the shelf, when emssion regulations get more demanding. I have the idea that's the whole actual philosophy emission management of car companies, to slow down the emission regulations as much as possible by complying without holding the slightes margin.
    – Bart
    Commented Nov 1, 2016 at 9:22

Consequently, you get a superrich part of mixture together with just air in the cylinder. These still have to mix/homogenize in the cylinder to get a good combustion.

This is already handled by both flavors of pulsed fuel injection:

  • In port injection

    Fuel and air are mixed before they enter the intake valve, so the incoming charge is already fairly homogenized.

  • In direct injection

    Under homogenous operation fuel is injected during the intake stroke and cylinder tumble and swirl does the rest.

As @SteveRacer highlights in his answer, this is a concern for port injection setups when the engine is cold, because the fuel isn't atomized well. The traditional "fix" is to run slightly richer in anticipation that not all of the fuel will combust due to poor atomization.

So why aren't there injectors that can vary their fuel flow?

More likely than not, this is for control system simplicity than anything. Compare these two scenarios:

  1. Manage injected fuel quantity through injector pulse width (time) while maintaining constant fuel rail pressure and injector flowing characteristic

  2. Manage injected fuel quantity through injector pulse width and varying fuel rail pressure/injector flowing characteristic

Both approaches will get you the same result but the latter is far more complicated to implement a control system for (not to mention the need for fancier hardware).

As a side note, there are injector manufacturers that claim to provide the ability to vary fuel flow rate through use of multiple pintles (Injector Dynamics springs to mind). I'm not in a position to validate the veracity of their claims though.

The ECU can derive the duration of the intake stroke from the rpm, and can consequently use all that time to make the injector deliver it's fuel during the whole intake stroke. This way, the mixture that enters the cylinder is almost completely homogenized already.

Fuel management system already do this implicitly by taking input from the crank-postion sensor (RPM) and combining it with information about engine load.

Note however that if the intent is to control AFR, RPM alone is not enough to determine how much fuel needs to be injected. A manual car at 2500 RPM in 3rd gear will burn fuel at a different rate than when it is in 2nd gear at the same engine speed.

  • Thanks for the clear and explaining answer. You say "the incoming charge is already fairly homogenized", this suggests that further homogenization is possible (and beneficial) right? Furhtermore, i get that variable injectors are less simple and more expensive, but the ever increasing emission regulations demand for better (more expensive) techniques anyway. This might possibly be a solution. The techniques used nowadays were also not cost effective and too complex in the 90s. We now have cylinder deactivation, which was unthinkably complex back then.
    – Bart
    Commented Nov 1, 2016 at 11:52
  • By the way, variable injection can be achieved by variable rail pressure, but also by variable injector opening/orifice. One could implement an ECO and a Power modus where fuelpressure is lower or higher, but also an injector that can open fully or partial. Whichever method is more (cost) effective.
    – Bart
    Commented Nov 1, 2016 at 11:55
  • @Bart I don't disagree with you that further homogenization is possible, but how would switching over to measuring intake stroke duration help here? It is already implicitly considered by the fuel management when it reads RPM from the crank position sensor
    – Zaid
    Commented Nov 1, 2016 at 11:55
  • Well, let's say the intakestroke lasts for 30ms, and 10uL of fuel is needed. The injector normally delivers 20uL per 30ms but if you open it for 50% it delivers 10uL per 30ms. Exactly what you need. This way you make the injector squirt the needed fuel during the entire intakestroke, giving a better homogenized charge entering the cylinder. That's what my hypothesis was.
    – Bart
    Commented Nov 1, 2016 at 12:00
  • @Bart so your objective is to have the injection occurring as long as the cylinder is undergoing its intake stroke. I tried to explain why this is inadequate in the last paragraph. You can have different quantities of air entering the cylinder for the same intake stroke duration (30 ms in your example), depending on intake manifold pressure. Injecting fuel for the same 30 ms would then result in different AFR's for different manifold pressures, which means that your AFR is not maintained.
    – Zaid
    Commented Nov 1, 2016 at 12:10

Don't forget that under stoichiometric conditions (which all engine manufacturers strive to run at), the air:fuel ratio is quite high (14:1). Injectors are commonly sized with this in mind. Very little fuel is actually required to make a big enough bang to drive the piston.
An interesting aspect is that older fuel injection systems actually would use most of the intake stroke, and the exhaust stroke, to deliver the correct amount of fuel. The 5M-GE engine in my old 1985 Supra uses this system; all 6 injectors fire at the same time, once per revolution of the engine. During one injection event, the intake valves are open and the intake air picks up the fuel as it enters the intake manifold. During the next, the valve is closed, but only for milliseconds at most (at idle, less at higher RPM). The fuel falls further in the manifold, but this probably provides a good mix when the valve opens again and the next charge enters the cylinder. Almost like seeding the lower half of the airstream, then the upper half, for each cylinder. And because the injectors are sited as close to the intake ports as possible, there is very little time for fuel to drop out of the airstream before it enters the cylinder.
New direct-injection engine designs exist that can perform multiple injection events per combustion stroke - if the engine is running at low load in a lean-burn mode and the driver steps on the throttle, rather than increasing the size of the next fuel charge, the ECU can fire the injector straight into the burning mixture for an instant increase in power. This sort of design allows the engine to burn very little fuel when not needed, but still be responsive. If memory serves, the Ford EcoBoost range has this capability.

  • Umm stoich at 40:1?... it's 14,7 to one as far as i know. Furthermore, stoich is not per se what manufacturers strive. Lean burn engines are a more and more common thing as you mentioned yourself. And yes PD systems, if that's what you mean, don't inject at the same speed at all times, but that's not really usable as variable injection i think. Your answer gave me some new perspective to look at things though, so thanks.
    – Bart
    Commented Nov 1, 2016 at 16:00
  • @Bart looks like I misheard, you're right, it's 14.7, not 40. I've corrected my answer. Also, while lean burn is desirable under low load conditions, when the engine is developing power, stoich is desirable, is what I was getting at. You're right though.
    – Gargravarr
    Commented Nov 1, 2016 at 16:52
  • I'm not entirely sure, but i believe lean burn engines still run rather lean(ie.>14.7) when power is demanded. They never get to stoich. @SteveRacer 's answer seems to confirm this, as he wanted to go to the webinar 'dr. Stoichiometric is dead'. Says enough i think! :) I guess the reason is that it'd be easier to control and take car of the emissions of a lean mixture. I heard they bring down NOx with ureum nowadays. Really interesting technique.
    – Bart
    Commented Nov 1, 2016 at 17:46

There is; they're called Carburettors. However, provided that the fuel / air is mixed appropriately at the point the spark plug fires, nothing else matters.

  • Carbs are whole another thing. Next, the question is exactly about my doubt that the mixture is appropriately mixed yet at ignition time.
    – Bart
    Commented Nov 1, 2016 at 11:42
  • The question is about fuel injected systems. Carbs are out of scope
    – Zaid
    Commented Nov 1, 2016 at 11:51
  • But a carburetor provides a constant flow of fuel / air mixed at the appropriate rate. Commented Nov 1, 2016 at 13:26
  • That is also not entirely true. A constant vacuum carb comes rather close, but a non-constant vacuum carb's AFR depends also on eg. acceleration pump effect.
    – Bart
    Commented Nov 1, 2016 at 14:27

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .