The Magic of Electronic Fuel Injection, Fuel Economy

and the Amazing Rotax 912iS Engine

Why Rotax?

There has been much talk of how to address the issues raised by the new swathe of fuel injected

engines for light aviation. Some are modified automotive systems and others are very specialised,

specific aircraft engine designs. The Rotax offering is firmly based on the rock solid bloodlines of the

912UL and ULS engines a great heritage and a serious engine, albeit at a premium. Personally, I fly in

the humanitarian aviation sector and consider the cost of the engine as more than just the purchase

price. I consider acceptability by my Civil Aviation Authority, fuel consumption (especially when good

fuel is hard to come by) and reliability, especially since our operations are approved for operations down

to 200 for aerial supply of health related materials to rural communities in somewhat challenging

terrain. So, for me, the only way I would consider a fuel injected engine was to be on the basis of high

redundancy. I wanted the usual double ignition, plus double injectors, double pumps, double

generators/alternators, double rectifiers, double ECU and double the confidence in the engine. Rotax

made just that! Then I wanted an engine monitoring unit that could take the heat and the dust, and the

millions of insects that try to make their homes inside any electronic device we install. The Stock

Aerospace EMU912iS is practically hermetically sealed and ideal for our applications so we had the

solution to try for some advantages with Electronic Fuel Injection with the 912iS and the Stock

EMU912iS.

What did we do?

We worked on the retrofit of our Rotax 912iS to a

Zenith CH701 airframe (the first Zenith to fly with

the 912iS worldwide). It has been an exciting

journey and we have learned a lot, there is more

to learn still, but we would like to share our

thinking so far, based on practical, in the field,

and in the air, testing.

First of all we need to understand that electronic

fuel injection requires higher fuel pressures (45psi

or 3bar is the 912iS pressure), tends to result in

lower fuel consumption (much lower in the case

of the 912iS), coupled with higher fuel flows.

For example, with two pumps running on the

Rotax 912iS there can be as much as 120litres per

hour running along the fuel rail, but consumption

in the cruise can be around 12litres. In the descent that drops to 2 or 3 litres per hour, and in the climb

rises to a potential of around 25 litres per hour (Wide Open Throttle, POWER mode). Consequently, the

non-consumed fuel must make its way back to the fuel tank from whence it came! You cannot return

the fuel to another tank! So, we have up to 118lph of fuel running over the engine and back to the tank

from whence it came.

ONE TANK SOLUTION:

If you only have one fuel tank, you simply return the fuel to the one tank you have. It is generally a good

idea to return the fuel BELOW the level of the fuel already in the tank and if possible diffuse it. This

helps to cool the fuel and to reduce fuel vapour with subsequent losses.

We did this with the 912iS. We set up a single tank installation, as shown below.

However, we made the return line run

internally to a diffuser (finger screen used in

reverse) at the bottom of the tank.

We put in around 50 hours on this system, and

never missed a beat. But a single tank is not a

valid solution for most aircraft especially if

you want to go places in the developing

nations (our area of operations) where 4 hours

between aerodromes and fuel supply

challenges (such as NONE) are common.

TWO TANK SOLUTION:

For such a system, traditional thinking suggests that you

use a duplex fuel selector, such as the FS 2020 (shown left)

from Andair, and when you switch tanks you connect the

feed line of the given tank to the fuel pumps and the return

line to the relevant tank. This is wonderful and makes a lot

of sense if you only have TWO fuel tanks, and can happily

return to each one as you select it.

Single Tank 40 litres

To fuel pumps and engine

from engine

However, we wanted to run a FOUR tank system for our applications. There are not any 4-way duplex

selectors on the market, and if there were it would almost certainly be a weak point in fuel management

systems, and outrageously expensive, so we had to find a different solution.

We worked through a long set of discussions about how this could be done, with as many people as we

could. Special thanks must go to the team at Rotax and Zenith who exhibited amazing patience with our

oddball concepts, finally, it all pointed to a header tank.

HEADER TANKS:

There are great arguments for a header tank in ALL aircraft fuel installations it ensures a supply of fuel

from tanks that may be prone to unporting or sucking air in uncoordinated flight, or long descents,

and ensures that, even in turbulent conditions with fuel sloshing around your tanks, pretty much

whatever the (non-aerobatic) attitude, fuel should be readily available to the engine.

As is the case in many flat bottomed fuel tanks without baffles, the CH701/750/801 fuel tanks in the

high-wing can get easily get unported during long descents, especially with low tanks. We know, we

have done it.

Voice of Experience:

It was in the 701 with 4 tanks, running in pairs; inboards and outboards. We were running on

outboards, with the 912UL 80Hp carb engine, flying over rain forest. We were on a photo run, and as

we set up for a quiet passage over a sensitive area (we try not to alarm the inhabitants of areas we

survey, for which the Rotax is a great, quiet, engine), as we got to about 800ft agl the engine changed

tone and rpm started to flutter, we looked at the flow in the transparent fuel filter (we always visually

monitor our fuel flow) and saw that we were frothing fuel a quick look at the Skydrive fuel pressure

gauge showed a needle flicking about like a conductor during one of Beethovens more exciting

symphonies! A quick change of attitude to nose up, and a change to inboard tanks and it was nothing

more than a moment of what if a BIG what if. But we learned to manage our fuel along with attitude.

Later we experimented with it, and reproduced it at altitude. The fuel outlet is at the back of the tanks,

and in the descent it can unport but with a header tank that would not matter, because there would

be fuel that cannot unport so easily. Without a header tank, the time from unport to engine hiccough is

about 2 minutes... in a carb engine (Had we been in an injected engine, we would probably not have

been able to restart in the time from engine quit to the rain forest canopy tickling our toes.)

We then did some tests on the ground at various engine settings, shaking the plane left and right, up

and down, and forcing air into the lines in ways that no pilot could ever imagine or probably manage

and we got the engine power to drop from 5500rpm to 3500rpm and back again as we forced air into

the lines. That was in a carb engine, that air can vent at the carbs, but all the same, we have decided

that we need a header tank for our operations in extreme conditions for both carb and injected engines.

UNPORTING IN AN EFI ENGINE:

Clearly, EFi engines have bigger problems with unporting. The problem can be expressed in fuel flow.

Our 80Hp Carb engines flow about 16lph against the maximum of 120lph on the example 912iS. That

gives us a 120/16 risk increase or 750% increase. This has to be actively responded to in the design

for safety reasons, after all we do not want to bend an aeroplane through a poor fuel system. Due to

the way the fuel is run along the rail, anything more than a small amount of air in the system is a

problem. Air is not vented in the firewall forward area (unlike in the carbs of the carb engine) it has to

return to the tanks VIA the rail and return line (unless you install a limited flow bypass line, which we

will discuss later) and since air is compressible, it can cause changes (that is LOSSES) in pressure, it needs

to purge the air because any air in the rail can, and does cause SUDDEN STOPPAGE in a matter of

seconds, especially at high revs. It also takes longer to prime in case air causes an engine stoppage in

flight (this is improved by a limited flow bypass line, if installed). Therefore, in any installation that

could unport, it is, in our opinion, essential to have a header tank we have discovered that there are

other advantages to the header tank too.

BACK TO THE TWO TANKS AND SOME TESTS:

So, let us consider a two tank, selectable system, with a header tank: The header tank is fed from the

main (generally wing) tanks, and is also the point of return of the unused fuel from across the rail. Then,

all you need to do is to select the wing tanks supply to the header tank. Yes, you still need a return to

the tank, to take air out (you need a duplex

fuel selector). The air seems to be lazy and

will take the line of least resistance back to

the tank. We proved this with a single tank

connected to a header tank.

If you only have a supply line from the fuel

tank to the header, without a second line or

return, it will air lock very quickly on a

single connected tank you MUST have a

return line if you want to run the engine for

more than a few seconds at a time!

We set up our single tank to the header

with valves on BOTH supply and return lines.

Then we opened the supply line ONLY to the

header less than 0.5 litre of fuel flowed

slowly before it stopped altogether. Then we

opened the return line, instantly the air could

be heared bubbling into the fuel tank. Now

we had established a very important fact, all

by accident. The return line was ALMOST at

the same fluid head pressure as the supply line (see the earlier picture where the return is run to the

BOTTOM of the tank). We then timed the flow along the lines to fill the eight litre tank which we

modified from the Viking Honda Fit aircraft engine installation. We did this several times, and it never

took more than 10 minutes (best fill was less than 9minutes), of gravity feed. That would equate to a

gravity feed of at least 60 litres per hour, along one 3/8 line with less of a head than we have in reality

in the aircraft.

Here we asked ourselves a question. What would happen in a TWO tank system if there was air in the

header tank?

So we installed the header tank with the following configuration.

NOTE, we did not run any return lines to the wing tanks and fitted valves on each supply line. We, again,

tried filling just one tank, and selecting just one tank. Lo and behold, the fuel flow trickled and stopped,

as expected. We then put some fuel in the second tank to create a fuel barrier, and then opened the tap

to see what happened. The fuel entered from the fullest tank and the air bubbled up into the least full

tank. The header still filled in 10 minutes or less. We then ran ground and air tests all went perfectly.

What was most interesting to us was that we used the old 1/4" lines to the header tank! The theory was

a simple one. The AMOUNT of fuel needing to come from the wings tanks to feed the CONSUMPTION of

the engine was easily met by the two 1/4" inch lines. The 3/8 line running to and from the

engine/header tank has to cope with up to 120lph, but the feed lines from the wing tanks would only

need to handle an estimated 25lph between them. In fact the maths makes a lot of sense.

TIME FOR SOME CALCULATIONS:

Using Pi.r2 as the formula for the cross-sectional area of the supply line to the header, and considering

the previously proven flow and consumption along the 3/8 lines, we did the following calculation (for

8litre Header

To fuel pumps and engine

from

engine

those who forgot 22/7 is roughly Pi , a radius is half the diameter, and to convert inch to mm we

multiply by 25.4):

Area of 3/8 fuel line : (22/7) x (((3/8)x25.4)/2)2 = 71mm2

Area if TWO 1 / 4 lines: 2 x (22/7) x ((0.25 x 25.4)/2)2 = 63mm2

so with two lines pushing fuel into the header tank have close to the supply of a single tank running one

3/8 line. Furthermore, we know that 3/8 lines can gravity feed 60lph to the header tank, so the

calculation for 1/4" lines supply, if only ONE were to supply fuel would be approximately 60x(32/71) or

27lph, and based on 25lph max fuel feed if only one tank had fuel in it, thus it would supply the need.

(remember we are averaging around 12lph, so in the cruise and the descent, even fuel flowing from only

one tank, would be around 100% greater than the demand and the header itself holds 8 litres of fuel,

enough for nearly 20minutes of operations after all fuel supply from the tanks has ceased at MAX pull

or for 40minutes in ECO cruise But we still felt this was a bit on the limit side for operations, until.

REALITY IS BETTER THAN THEORY:

THEN we actually measured the 3/8 versatube lines we had been running on with our single tank

installation and there was a surprise Normally when you buy rubber hose you purchase an inside

diameter (ID) With aluminium tubes from a well-known international supplier, the 3/8 tube is OD

(Outside Diameter) has 0.9mm walls, so the ACTUAL inside diameters work out as follows:

Rubber Hose 3/8 : 9.52mm ID +/- 71mm2 area

Rubber Hose 1/4 : 6.25mm ID +/- 32mm2 area

Versatube 3/8 : 7.72mm ID +/- 47mm2 area

Thus, we recalculated, a single 1/4" line would supply 60 x (32/47) or 40 lph on gravity drip. a 60%

safety margin under 25lph demand. Then we realised that we would always have TWO supply lines in

our system, giving us up to 80lph on two tanks and 160lph on four tanks we could actually supply more

than the rail feed just from the main tanks if all four were connected, with fuel flowing, and open all

on 1/4 lines!

So, how did we do away with the return lines to the wing tanks?

ARE RETURN LINES ESSENTIAL IN A CONNECTED TWO TANK SYSTEM?

When we did the header tank test, we forgot that the return line was to the BOTTOM of the wing tank

and hence we got the wonderful bubble back, diffused fuel event and not fuel vapours being

introduced to the top of the tank in the air space. Then when we tested in the plane we also got the

same result. It appears, based on our experience, that provided that 2 or more tanks are connected and

open to the header tank at any one time, and that that the fuel system is properly purged on first fill (ie

fill from one tank until the fuel balances up into the other tank(s) and purges initial air from the system)

it should be fine.

Back to the air, try it all out, and land safely it worked and still (many hours later) works!

FOUR TANKS INTO ONE DO GO!

Now, to add the other two tanks to see if it really works with FOUR tanks connected. It should be noted

that we only have fuel gauges on our INBOARD tanks, and have in the past flown on time when

operating on the outboard tanks:

Here we get to encounter the old problem of the outboard tanks being higher than the inboard tanks.

This is not a problem provided the level in the outboard tanks is equal to or lower than the filler neck on

the inboard tanks. (with 912iS, and for us, that means a good 8 hours of fuel). It provides a no-need-to-

select-tanks option for most of our flights. For the longer flights we need to close the supply lines to

the outboards and fill them to the top. We have locked the inboard tanks OPEN to the header tank

with tie wraps (so that we can still snip the tie wraps and easily shut them off if needed under

contamination conditions, etc).

For longer flights, where we want to be able to carry up to 150 or 12 hours+ of go juice with us, since we

cannot guarantee fuel at our destinations, we would fill all tanks, closing the valves on the outboard

tanks. There we fly on the inboard tanks until the inboard fuel gauges show half full, before opening the

outboard tanks. Then, the fuel will balance (more or less, since the system will normally pull more fuel

from one side than the other) across the four tanks. At the moment of change over, the header will

replenish principally from the fuller outboard tanks and then, through some back pressure, fuel will flow

upwards towards the inboard tanks. Therefore, the fuel gauges for the inboard tanks will, after an

amount of time, give fuel availability over the 160litre maximum fill load. We will still calculate our time

8litre Header

To fuel pumps and engine

from

engine

remaining based on uplift and consumption, but now have fuel gauges across the four tanks (albeit that

the scale would not be linear and the volumes represented different when on 4 tanks)!

I CAN SEE THE FUEL!

Remember, once the

4 sight gauges are

empty we know we

still have 8litres in the

header tank, more

than thirty minutes

fuel left on board,

provided we do not

use full power!

This set of arguments

has still not convinced

some folks who want

to insist that they

MUST have large return lines to their tanks, even in a multi-tank connected solution. There is no doubt

that return lines to the tanks is a positive system and reduces risks of fuel system problems. However,

let us just consider for a moment the situation that is in place, and all of us can make our own minds up.

WHAT IS IN THE RETURN LINES? AIR OR FUEL?

Here we have the standard system with all lines in 3/8. There is only one tank connected at a time,

and any air can return to the tanks via the ESSENTIAL return line for the tank (shown red here). For the

record I have not seen a duplex selector with BOTH option since it could mean fuel could return to the

other tank, causing potential over-spilling and loss of fuel! This is an ESSENTIAL installation criteria if

running just TWO tanks, NO header and running from ONE tank at a time. Failure to comply could

results in sudden engine stoppage and even death.

To fuel pumps

and engine

from

engine

Duplex

Selector

So, let us NOW add the header tank and take away the selector and see what happens to those return

lines

Let us look at that more closely with the tanks FULL The return lines will AUTOMATICALLY fill with

fuel. Fuel will find its own level. So, the air will have to push up against the same head as the supply

lines. Now, of course, in the ideal header tank, the return line is taken out at a HIGHER position than the

supply line, but air can, and will find its way into the lines at some point.

8litre Header

To fuel pumps and engine

from

engine

8litre Header

To fuel pumps and engine

from

engine

As the flight continues, the fuel levels will drop, but there will always be the same head in each line, (if

the aircraft is perfectly level, and or balanced) since the fuel will find its own level, filling the return

venting line with fuel. This means that when air comes through the venting line it will splutter fuel and

vapour into the tank it is heading for.

We can improve on that situation by venting back

to the bottom of the tank as we did in our initial

tanks, where we proved that the air would be

forced out of the lines by the pressure of fuel

arriving from the higher pressure areas.

This led to the consideration of our system, which

currently does NOT have return lines to the

tanks, and MUST always have TWO or more tanks

open to allow venting. In the one of our tanks is

empty tank condition (remember, one will

always empty first due to geometry and other

factors!) the empty tank would provide the vent

even though the line to the tank will still contain fuel - and allow continuous fuel supply to the header

tank but in that condition it is most likely that the pilot is looking for somewhere to land or will soon

be forced to!

The answer is simple, the supply line contains fuel which will flow down to the header tank and then rise

up any other connected line, due to fluids finding their own level, and since the system is GRAVITY fed to

the header tank, even when there is an empty tank, which would have the lowest hear, the air would

8litre Header

To fuel pumps and engine

from

engine

not be sucked into the header. The least full tank would be the natural tank for air to return to,

bubbling up the lines.

WHAT IF AIR IS TRAPPED?

First of all, line installation is essential you must make sure that air cannot get trapped in a goose-nck

arrangement. Lines must provide upslope from the pumps back to the tanks. There must be no funny

shapes that allow air to get trapped or you will have problems. That is the basis of all fuel systems

installations, but needs reiterated here for completeness.

BUT if there was fuel in the tank, then we underwent uncoordinated flight, or extreme descents,

uncovering the port, allowing air into ALL of the lines, and then covering the ports by returning to high

angle of attack, now with air in the header tank what would happen. There would fuel in the tanks,

air in the lines and fuel in the header.

TEST PILOT TIME!

This could only practically be flown in a test condition and I was pleased to be able to test it!

Making sure that two tanks were really low, and two had about double the amount of fuel.

Against the rules of take off and land on the fullest tanks we took off on the low pair of tanks,

confident that the nose high attitude would feed the header and that the header contained enough

fuel for at least 20minutes Wide Open Throttle climb. Then, at a safe height, immediately after opening

the four tanks, we pushed the plane into a steep dive. The four tanks unported and we ADDED power to

suck fuel at a higher rate to get air into the lines AND the header tanks. A sustained VNE dive was

undertaken, and then we pulled up, blocking the air in the header tank and lines. Within minutes the

bubbles had dissipated and normal fuel flow was resumed. There is nothing like real life testing of a

theory! It worked in practical testing.

WHAT IF THE AIR GETS PAST THE PUMPS?

So, what about the case where you do get air all the way to the fuel pumps? That would be bad news.

It is advisable to install a a limited flow bypass line. This has been done on Rotax 912 engines for a long

time. The standard 0.35mm jet line that allows excess pressure and air away from the carbs and back to

the tanks (which is in the installation manual but not on all installations)!

It is possible to install a similar solution between the SUPPLY line (after the fuel pumps) and the return

line. The advantages are that it allows quicker priming of the engine, since the air will prefer to take the

short cut rather than take a trip across the rail, and will also permit the quicker depressurisation after

shutdown, by allowing fuel pressure to bleed back to the tank making work on the fuel system less likely

to look like fuel fountain!

Since we did not have access to a suitable 0,35mm jet, we chose to manufacture a helicoidal air bleed

plug between the pressure side, just after the fuel pump, and the return to the header. It sounds posh,

but it is actually nothing more than the thread of a bolt inserted (without any Loctite, fuel lube, or paste)

into a fuel connector that we tapped. We tested it under water to make sure that air could pass the

limited flow bypass line, and then added a filter on the pressure side (to make sure that the bypass line

would not get bunged up).

Simple solution, brought about by necessity in the African bush, where we do all of our engineering!

(we hope to get a lathe soon, but this method works!). A 0,35 mm jet would be better, and we will

change to that in due course, but we definitely wanted to have a limited flow bypass mainly to reduce

pressure in the lines after shutdown. Imagine the situation otherwise! You could open a fuel fitting to

get a spray of fuel sitting at 3bar (45psi), unless you install such a solution.

We noted that there was a need to install the bypass line VERTICALLY so that air would rise into the

return line it as it passed through the pressure line.

This leads us to the simple niceties of the overall view of our current testing installation

For us, this works, and appears to work very well indeed!

Fuel

drain

Fuel

drain Fuel Pump Module

912iS

Engine

To Header Tank

filter

Limited

flow bypass

unit

Gascolator (70micron

Teflon coated)

10

micron

filter

3/8 lines

3/8 lines

1

1/4 lines

with visual

flow

monitoring

2 3 4

WE WERE NOT SATISFIED.

As we contemplated the challenges of this system further, we considered what would happen if we had

a massive air lock. Let us imaging that somehow we got a litre of air blocked in the header tank. What

if it would be enough to block all the flow?

So, in the interest of proving a point we created a FOUR litre air lock. We closed off all four tanks, and

then opened one, and bled fuel out until the visual filter showed we had emptied the tank and its line.

Then, we bled a further FOUR litres of fuel, and closed the fuel tap. We now had a closed system with

no supply to the header and four litres of fuel with four litres of air in the header tank.

We started the engine and ran for three minutes (the engine ran fine) this warmed the fuel and created

an even greater risk of vapour lock.

Then, we opened TWO tanks with fuel in, and observed fuel flowing. At that point we opened the third

tank with fuel in. The engine continued to run faultlessly. Fuel came from the tank with highest head,

and air bubbled back through the supply lines to the other two fuel filled tanks. After about twenty

minutes, the fuel demands of the engine had been met AND the four litres of air purged, and the fuel

rose up the other two filters, as the majority of air was purged from the system.

To complete the test, we then opened the EMPTY tank, and watched fuel RUSH up into that tank, at the

same time the lowest head tank showed air in its visible filter for a few seconds before fully recovering

to four fuel filled filters.

We were satisfied that this fuel system was robust for our applications. Even though it is not ideal, it

works and if it works in this situation, it will readily work in our future, improved, solutions.

WHAT WOULD WE PREFER?

We would prefer to have 3/8 lines from the tanks to the header tank, and must admit we would like to

change the design of the header tank quite a bit, introducing a sump, positioning inlets and outlets

differently, incorporating a low tank warning light that would indicate either air lock or low fuel. (we

have seen that this company offers such a service - http://www.coyote-gear.com/8inch.html ). We

intend to ALWAYS install with at least PAIRED TANKS to ensure that fuel flows due to head differentials.

We want to keep our header tanks low and near the wing supply point of the fuel (near the wing root

in the CH7xx and CH801). But what we would like to do is to establish this in a Rotax 912iS powered,

long range equipped, CH750 later this year, funds permitting. We also plan to fly such a plane across the

Sahara to Europe! If you have found this article useful, please consider a donation towards that by

sending funds to Zenith Aircraft Company marked Ghana CH750 Project, thank you!

DISCLAIMER:

This is what we did, in our installation, it suited our purposes. We considered our INSTALLATION, our

FUEL, our LINES, our NEEDS, our level of RISK, and we worked to a solution that suited US and our

AIRFRAME. Each person must take into consideration their own conditions, fuel, equipment, flying type,

needs, airframe, lines, risks, desires, etc. and carry out suitable design and development to ensure that

they are comfortably safe. We do not offer any guarantee or implied suitability of the above systems,

but provide our results in the interest of increasing positive discussions and the promotion of simple,

safe fuel solutions. This worked for us in our installations in our workshops, where tanks, lines, bleeds,

systems, drains, etc. are installed and tested in our specific conditions for our specific applications. Any

person copying or using this information for their own purposes does so at their own risk.

OWNERSHIP:

This document was prepared by Jonathan Porter (aka Captain Yaw) and Patricia Mawuli of WAASPS Ltd.,

Ghana, West Africa, as part of their own, independent, research into fuel supply solutions for the Rotax

912iS for use in Humanitarian Aviation Logistics with Medicine on the Move, and integrated to the

training of the young rural women from Ghana, West Africa, in light aircraft build, engineering and

development at the AvTech Academy.

Contact: capt.yaw@gmail.com

Web Sites:

http://www.waasps.com

http://www.medicineonthemove.org

http://medicineonthemove.blogspot.com

http://avtechacademy.blogspot.com

March 2013

E&OE

mailto:capt.yaw@gmail.comhttp://www.waasps.com/http://www.medicineonthemove.org/http://medicineonthemove.blogspot.com/http://avtechacademy.blogspot.com/