Warm Air Intake With Steam

[cRiPpLe_rOoStEr]

Presumably, adding steam to the EGR flow could keep the EGR valve and its body cleaner, and dillute the oily vapors from the crankcase vents more accurately instead of turning it into a sticky residue that accumulates on the intake manifold walls. Even though it's not nearly as problematic as in Diesel engines, since you're using LPG it becomes harder to clean up those residues than when you're using gasoline.

Originally Posted by teoman

Any benefit to monitoring the EGR valve actuation and adding a bit of steam to it?

I thought that monitoring the EGR valve in order to keep its actuation and the addition of steam at a proportionate ratio. BTW do you have any plan to monitor the relative humidity of the intake air stream in order to keep it more constant at a sweet spot?

[teoman]

I have a dht11 temp and humidity sensor that i can use. But i want to have the controller inside the car, so not sure i should do that.

Besides i should first think about monitoring the actual steam temperature. It should not exceed 450 deg C in lpg and 250 degC in gasoline (auto ignition temperatures) just in case.

Rorence:

Just to sum up, i have absolutely no idea what i am doing. I obviously want to get my fuel consumption down.

I researched a bit and i found that the sweet spot for intake air temperature is 85 deg C (the tests i found are for a HHCI engine). I wanted to do water injection anyway so i thought why not combine the two and do steam injection.


My initial objective with the steam is to get the air to 85 deg C. That is all. After i have it in place and connected to the obd2 i can alter the injection parameters and all kinds of stuff, but initially the algorithm will be monitor intake air amount (in grams / minute ) intake air temperature (in deg C) and inject steam to bring the air temp up to 85 deg C.

If all goes as calculated, i will have air at 85 deg C and the finest droplets of water entering the engine.


(I am also actively pondering if this is the best methodology).

With what i read about water injection it should be %20 of fuel injected. My calculations equate to roughly the same mass of steam injection.

[cRiPpLe_rOoStEr]

Raising the steam temperature to the auto ignition points would be too risky.

[teoman]

Yes, i am planning on avoiding auto ignition

Was checking my intake temperatures, outside was 20-22 deg C. Intake was 60 deg in traffic and 35 deg C going 50kph.

At delta T = 85 - 60 = 25 deg C, there wont be much injection.

[teoman]

These guys have a steam pot maintained at 3 bar and 133.5 deg C. And they use a steam injector to inject %20 of fuel by mass.
https://www.researchgate.net/publica...gnition_engine

I will try to type my algorithm tonight, to compare injection rates.

[teoman]

Further pondering about this,
air at 20 deg C is 1.2kg/m^3 -> 0.83 m^3 / kg
and air at 80 deg C is 1kg/m^3 -> 1 m^3 / kg

So if all things are equal, i will have %16 less air, so %16 less fuel injected by the ECU.

But i suppose i will be pressing more on the accelerator to get the same performance, have less vacuum so it will be a relaively more pressurized air which will account for more air and more fuel.

[teoman]

#define ML2PWM 4 // not calculated yet, may need a conversion table
#define DESIRED_IAT 85 // intake air temperature
#define JpML 2230 // steam: joules / mililiter

//IAT in degC
//MAF in g/s grams/second
int calculateOutput(float IAT, float MAF) {
// I will need roughly 1J of energy to heat 1g of air by 1 degC
// More precisely 1.005 at 0 deg C and 1.009 at 80 degC
//http://www.engineeringtoolbox.com/air-properties-d_156.html

// Steam at 100 deg C has 2,230 joules/ ml or gram


float dT = DESIRED_IAT - IAT; // in deg C
float ML = dt * MAF / JpML; //mililiters per sec

return ML2PWM * ML; // send to the injector as pwm
}


Well that turned out to be much simpler than i thought.

Just a sanity check:
outside was 20 deg c, IAT was 35 deg C
MAF was reading 5 g/s at idle, and 100g/s at a significant load.

dT would be 50

5*50/2300 = 250 / 2300 -> ~1/10 ml/s 360ml/h -> 0.3L/h
100*50/2300 = 5000/2300 -> ~2 ml/s 7200ml/h -> 7.2L/h

I really need to test/calibrate my injector and pump soon. And i have to visit the pneumatics store.

[teoman]

Anyone happen to know the id's for the MB canbus?

I prefer not to use the obd2 interface as it is slower and 2 devices usually cannot coexist. But if i tap in to the real canbus as a module, i can use torque to change settings on my device.

[teoman]

My silicone hoses have arrived. But I have decided not to use them. (0.8mm ID)

I shall be using a 3mm OD 2mm ID PTFE hose instead.

My reasoning behind this is, I do not want to pressurize the system with a restive/restrictive path. I have made the assumption that the steam is at 100 deg C. And the hotter i make the steam, the less i will need to inject it. BUT I want to inject steam, at least the water component, injecting water should have a benefit (or at least I believe so, and the papers indicate so {For petrol/gasoline it helps atomization/vaporization of fuel, i am running LPG and water is still mildly beneficial for some unknown reason to me. })

Adding a restriction will increase the pressure in my vaporization chamber, which will lead to higher temperature steam.

If the temperature is at 1 atm, the water will turn to steam at 100 deg C.

Steam has a thermal conductivity of 0.0184 W/(m K) at 100 deg C.
Air has a thermal conductivity of 0.0314 W/(m K) at 100 deg C.

So... steam is a bit more than half the conductivity of air. This means that it is a pretty good insulator which means that the steam in contact with the metal will insulate the steam that is not in contact. In addition my hoses are still pretty small, so the velocity of the steam will be high, giving it not much time to dwell in the vaporization chamber.

Sanity check:
1 meter of 2mm id hose -> volume of 3.14 * 1000 mm = 3140 mm^3 = 3.14 cm^3.

30 cm of 1cm id pipe -> volume of 3.14 * 30 cm = 94.2 cm^3

Although nothing is set in stone, these appear to be the specs. So 97 cm^3, lets just call it 100 cm^3.

Now, when the system stops, this volume will be filled with steam.
Steam at 100 deg C (boiling point) has a density of 1.69 m^3 / kg. -> 1690 cm^3/gr

Soooo my 100cm^3 of volume filled with steam will condense in to 1/17th of a gram of water when the engine cools down. This is not really a big worry for me.



Lets assume that I am injecting 1/10ml of water and 2ml of water per sec.
1/10 ml of water (for water 1ml = 1gr) will occupy a volume of 169 cm^3. {about half a coke can for those that do not follow} This is amount of steam injected. So this is 1.69 time the total volume of my chamber. So every second all of the chambers contents will be ejected in to the manifold as 1.69 times its volume of new steam is generated.

At 2ml/s that is ~3400 ml of steam (3.4 liters gas volume) passing through per second. 34 times the volume of the chamber. With such high volumes, the velocity will be high and the high velocity steam may bring with it some unvaporized water. I am not at all an expert in this domain it is just a gut feeling that i will have to test out with the system if anyone does not have any insight on this.

So from these calculations i can conclude that dwell time, or the steam dicking around in the vaporization chamber and super heating will not be an issue for me.

However, with this much gas generation, keeping my vaporization chamber (pipe) up to exhaust temp may be a concern.



The original idea was to have a very small hose that would have high pressure steam therefore high exit velocity which would help with the mixing, this is not my way of thinking now.


PS: all calculations are done in my mind on the fly whilst typing, so there may be some errors.

[teoman]

I Have started to use the sparkfun canbus library, but that returns data in string format. Will try to modify it now so it returns floats or integers.

Edit: Done