Last week we introduced the hardware of our new DIY water injection system.
To briefly recap, it uses an ultra high pressure pump, a mains power inverter,
cartridge filter and brass/stainless steel high quality atomising nozzle.
The result is a continuous flow of very small droplets of water.
We said last week that the water injection system is ‘roughly proportional’.
By that we mean that the amount of water being added to the intake air at high
loads is greater than at low loads.
But why only ‘roughly proportional’? Why not use electronic control to
achieve a precise outcome? Well, there’s more to that than meets the eye.
At first, building an accurately proportional system seems easy. For example,
why not just pulse an injector flowing the water? However, that has three
- The injector needs to be able to handle water without corroding. The only
common injectors that meet this criterion are those designed for methanol.
- An injector system needs to be equipped with a pressure regulator – again one
that will cope with water (and we don’t know of any of those).
- An ECU will need to be used to control the system.
And it gets even more complex than that. If you’re using just a single
injector, its pulsing rate will need to be very fast. (A slow pulse adds
[ie discrete amounts]
of water to the intake. The problem with this
is that each time the intake valves of a particular cylinder open, the water may
not be present.) So what’s wrong with using a very fast pulse rate? Well, a
large injector (and methanol injectors are large) pulsed very fast with a low
duty cycle will end up not flowing anything – the opening times will be too
OK, so what about changing the speed of the pump to vary water flow? Again
that’s possible (with this 240V powered pump we’d use something like a light
dimmer), however our testing showed that once pump speed drops, water pressure
plummets. The result is that the nozzle dribbles rather than atomising.
So is there a simple and cheaper way of making the water addition
proportional to intake air flow?
There is. Firstly, to avoid the ‘cylinder missing out’ scenario, the water
injection spray needs to be continuous. Secondly, the easiest way of making the
water addition proportional is to offer the engine intake all it wants. That is,
if the spray is across the mouth of the airbox intake, a greater intake airflow
will take more droplets, a smaller intake airflow will take less, and when the
intake airflow is stopped (eg when the throttle is closed) no new droplets will
pass into the system.
But there is a negative to this approach: because the water that doesn’t flow
into the intake evaporates in the outside air, it’s much more wasteful of water.
In itself that’s probably no big deal, but it means that a larger tank will be
Having an exposed and open access nozzle also make it easier to periodically
check the nozzle (for both flow and atomisation) and to remove the nozzle filter
A system of this sort can still have sophisticated control of when it starts
and stops, so using water only when desired.
As described above, this system injects water in front of the air filter, at
the engine intake. As far as we know, no other water injection system attempts
to do this. So why do we?
Firstly, one of the problems of before-turbo water injection is potential
erosion of the compressor blades by the impingement of drops of water. When the
turbo is spinning fast, such impacts have the potential to strip tiny fragments
of metal off the blades, over time pitting it. Whether this actually occurs or
not is open to question – but it is a valid question. Therefore, if injecting
the water pre-turbo, it makes sense to keep the water droplet size to an
Another potential problem of water injection is washing the oil off the
cylinder bores, so increasing wear. There have been documented cases of this
occurring in test engines, although the amount of water and how it was being
atomised (was it being atomised?) are unknown. Therefore, it again makes
sense to keep the water droplet size to an absolute minimum.
Finally, if the system is used in a diesel it’s very important that a large
quantity of water cannot find its way into the engine. This is because the
clearance volume (the volume above the piston at Top Dead Centre) is very low –
that’s why the compression ratio of diesels is high. If enough water gets into
the engine to fill the volume above the piston, engine damage will result from
this hydraulic locking – in fact, the connecting rods will probably bend. (This
occurs because water is not compressible.) It’s therefore vital that there is no
way that sufficient water can ever enter the engine – even if something goes
wrong with the water injection system. Injecting a fine spray of droplets the
water into the intake air stream ahead of the filter makes it very unlikely that
this problem will ever occur.
At this point we need to take a look at water. Water in your bath is in
liquid form; even when water is in very small drops, it is still in liquid form.
As described is the breakout box below, it is the change of state from liquid to
a gas that has such a huge energy uptake, therefore cooling the surrounding air.
So no matter how small the droplets of water, they can still change in state –
that is, evaporate and become water vapour.
The water droplets in a cloud in the sky are in the order of 0.01mm (10
microns) in diameter. That’s pretty small – but they’re still droplets of water,
not water vapour.
So what if we can generate water droplets of this size or even smaller? In
that case, plenty will go straight through the air filter! Most fuel filters are
rated at about 10 microns, and while it’s harder to say with certainty with air
filters (most are tested using a broad spectrum dust and their filtration
ability is based on how much of the dust they catch, not how small it is), you
can be pretty confident that if you can make the drops of water sufficiently
tiny, they’ll go all the way to the engine as a fog.
But does the Spraying System nozzle that we're using produce drops as small as 10 microns? We
don’t know – we have no way of measuring drop size. However, even when running
water injection continuously at the intake (in the test case some 100cm from the
filter box), in normal driving the filter itself remains quite dry.
The other way of looking at it is to say that even if many drops are larger
than 10 microns, the filter will filter them out, with the deposited water then
evaporating over a little time. Clearly, though, testing should always
include assessing the wetness of the filter element, and if it is becoming
damp to touch, the system should be changed (eg in nozzle size or location).
From Water to Water Vapour
Evaporating a kilogram (ie a litre) of water requires 2257 kilo-joules of
energy – and that’s a lot! If the nozzle flows 400 ml/minute, and if all the
water evaporates, each minute 903 kilo-joules of energy are extracted. One joule
per second is the equivalent of 1 watt, so fully evaporating 400 ml/min of water
provides a cooling power of 15 kilowatts! Even a 130 ml/min spray provides a
potential cooling power of just under 5kW.
The key point is that to cool the intake air, the water must evaporate – it
is this change of state from water to a gas which absorbs the energy. And the
key to getting water to evaporate is to use very small drops – an atomised mist
– which dramatically increases the surface-area-to-volume-ratio of each drop,
In addition to drop size, the rate of evaporation will also depend on the
relative humidity of the air (if you like, an indicator of how much ‘room’ there
is left in the air for evaporated water) and the temperature of the intake
Post Turbo Spray?Because of the pressure the Ulka pump generates, if it’s desired, the spray nozzle
can be placed on the intake after the turbo (and intercooler, if fitted). In
other words, there will still be plenty of pressure left to create a good spray
pattern even when working into boost pressure.
The advantage of this placement is that if there’s an intercooler prior to
the spray, the ‘cooler will work better (because the temp differential between
the cooler and the outside air will be greater than if the water spray is in front of
The disadvantages are that the system will no longer be proportional in water
addition (the amount of water added at low loads will be much greater per cfm
than at high loads); water may not all evaporate before reaching the combustion
chamber; and it is more dangerous from a point of view of washing away cylinder
wall lubricants and potentially water-locking diesel engines.
As we’ve covered previously, water injection can be used to perform a myriad
of functions – from reducing the amount of fuel enrichment needed at high loads,
to decreasing NOx emissions, to acting as a form of intercooler. We’ve not
tested the system in every possible way, but we have fitted the system to two
cars, with varying results.
The water injection system was used on the hybrid Honda Insight in an attempt
to improve fuel consumption.
Already one of the world’s most economical cars, we thought it would be
interesting to see if we could improve its cruise fuel economy. The car runs a
very lean cruise (an air/fuel ratio of about 25:1!) and we wondered if with
water injection, we could also use an interceptor to advance the ignition timing
and so get even higher efficiency. (In this case the water would effectively
increase the octane rating of the fuel, allowing ignition timing advance without
As a first step we left the engine management standard and extensively tested
the fuel consumption with and without the water injection system. The
water injection system flowed about 65 ml/minute. Over numerous tests we found a
just-perceptible (but repeatable) fuel economy improvement – say about 3-5 per
cent. With the water being added prior to the air filter intake, the whole
intake system remained cold to touch, even after sitting on the freeway at 110
km/h for half an hour or more. The air filter also stayed dry.
As stated, this was with the engine management system standard; with the
timing advanced, we fully expected to get even better results.
However, a water injection system that is going to improve cruise fuel
consumption needs to be running whenever you’re cruising! And that translates to
a large consumption of water. The range of the Honda is around 1000 kilometres,
so if that was all done on the highway, you'd need something like 10 hours of
water supply (or else you need to fill the water tank more frequently than the
fuel tank). And ten hours of water is getting on for 40 litres – a lot in a little
For this reason, we decided not to pursue this approach any further. However,
we think the testing we did showed the water injection system has major potential to
reduce fuel consumption - and that must be even more the case with accompanying engine management
Peugeot 405 diesel
The other car that was tested with the high pressure water injection system
was a Peugeot 405 diesel turbo. The car runs a mechanical fuel injection system
that has been tweaked for extra turbo boost and extra performance.
Two aspects were tested with the water injection system, with again water
added in front of the air filter.
The first approach was to see if reduced smoke emissions could be achieved.
Smoke is a bugbear of the mechanical injection system because the system doesn’t
have any compensation for higher intake air temps, resulting in full-throttle
black smoke in hot conditions and no smoke in cold conditions. With 65 ml/minute
water injection running on full load acceleration, smoke emissions visually
appeared to be reduced.
The second approach was to use water injection as an intercooling agent.
Testing of its effectiveness was carried out up a long, steep hill. This hill
required that the car continuously stay on moderate boost for about 2
The intake air temp after the intercooler was monitored, and the temperature
rise between the bottom of the hill and the top of the hill assessed.
In standard form, the temperature rise was 39 degrees C (the intake air at
the bottom was 28 degrees and at the top of the hill, 68 degrees C). With water
injected ahead of the filter, the rise in intake air temp was barely any
different at 37 degrees C.
Being a diesel, we were reluctant to inject the water directly into the
engine (rather than through the filter) so no more water injection testing was
Note that in this case, much better results were obtained by using the spray
on the intercooler core – with a larger TN 1.5 nozzle, the temp increase up the
test hill was reduced to only 21 degrees C.
The above test results show that water injection is not a panacea for all
automotive ills. You need, especially in continuous flow applications, to weigh
up water consumption versus benefit, and to keep in mind that as an intercooler,
cooling an existing core might be a better way to go.
However, the described hardware provides a robust and effective way of
producing very small droplets of water. The water can be injected either before
or after the filter, and the pump provides easily enough pressure to work
against boost. The different nozzles allow a wide range of flows to be used and
the ability to inject at the mouth of the engine intake gives proportionality of