AI Safeguards
 

while it is tempting to think of AI as a free lunch know that it isn't. yes, w the addition of a $250 to $500+ AI system you can buzz around town at 25 psi on pump melting your tires and scaring yourself half to death you also assume major engine risk.

if the system malfuntions and you have "tuned up" (meaning leaned out your pump fuel map, advanced the timing and boost) it may be new motor time.

so, unless you haven't touched your non AI tune and are running AI just for cooling, you bet your motor.

given the high quality/reliability of some AI systems i am willing to make the bet.

the first rule w regard to AI "safeguards" is to make your first selection parameter system intergrity.

the bold is there for emphasis. i do not believe that saving a hundred dollars on system X should be your first consideration. you are betting your motor.

you want it bulletproof first.

O K.... so you picked a system in which you have confidence and will install it very carefully w regard to electrical connections, AI injectant connections etc. you will read up on tuning... my suggestion if you are using alcohol is Turbobuick w an unending amount of tuning threads. (btw, no significant differences w the rotary).

you can layer on various "safeguards" before we get into them let me make a point.

if you blow a pump fuse things happen very fast as to engine failure. the engine sees no injectant, goes lean, probably at higher boost than straight pump, probably more timing than straight pump. guess how long it takes to break an apex seal etc...

some of the "safeguards" such as opening the WG to cut boost to spring pressure, opening a (greddy) pop-off valve (better idea), depend on MECHANICAL events. how long do they take to effect?

my guess is TOO long. when the motor detonates how many compression strokes occur before damage?

i don't know and hopefully someone does and will contribute to our understanding.

because i am skeptical as to the speed of mechanical safeguards i run a J&S knock detector as my safeguard. it is the newer model that only arms under a settable boost. it retards the leading spark (IGL) up to 10 degrees. since it works on the electrics it works at the speed of light. literally. that's fast enough to retard on the next rotor face! whether taking 10 degrees out of the IGL is enough to save my motor i don't know. it is a help.

Richard L, key guy at AI powerhouse Aquamist, wrote an introductory piece on Safeguards for the Misub site (Evolutionm.net) that is a nice starter-upper piece....

here's Richard...


"I think we have discussed most of the systems available and their capabilties. It is a good moment to see what failsafe mechanisms are available from the simpest to the most complex. I will update the list when more contributions are posted, if any.

Direct detection:

1. Fluid tank level sensor :
Advance warning but only good for early warning but little use for system power failure or blocked nozzle.

2. Loss of system power:
If power is interrupted to the AI system, it will de-engergise a relay and performs "Plan B", but will not detect blocked jet or cut pipe or no fluid flow.

3. Inline pressure gauge:
This is very cost effective but require user paying attention.

4. Inline single pressure switch:
If the system looses fliud pressure during activation, an inline preset pressure switch can action a boost drop etc.

5. Two inline pressure switches:
Detects blocked nozzle (over pressure) and "lost of fluid pressure" (under pressure). This method is quite cost effective and within DIY capability.

6. Flow switch:
In inline flow switch, comprised of spring-load magnetic plunger inside a tube where the water flows through. The rate of spring determines the promixity of a reed switch again the pluger deflectiion. Output: on/off. - cost: low

7. Turbine flow sensor:
Good for progressive system but requires complex electronics circuitry to report a differnce between the "actual" and "planned" flow in real time.


More..?

In-direct detection:

1. Knock detection:
2. AFR tracking:
3. EGT tracking:
4. Inlet temperature tracking:

I leave it as it is until the this topic attracts more interest. It is important to know how "safe" is a "failsafe" .

Richard"


a follow-on...

"Shiv is working on a function on the XEDE that would work as a great as a water/alky injection failsafe.

From Shivs post,

1) Lean Run Protect
You define the leanest allowable on-boost AFR curve (5 RPM points) and decide how long it needs to be triggered for (0.5 sec is long enough to not induce false triggers caused by sensor lag) and let the XEDE do the rest. By "rest" I mean a sudden boost drop. If it still remains leaner than that threshold at the lower boost (say another 1 second), it'll do a hard fuel cut. The user can change the conditional logic (allowable time durations, actions, etc,.) This is critical for Methanol use where a pump failor or a jet clogging will destroy a motor in no time. Also good for protection in the case of fuel pump failure or anything else that could induce leanrun."


another follow on from Engine Runup/vendor
"Along with the direct port injection, I also have added a safety feature. The engine, without a boost controller, boosts around 15 psi. In the last picture, you will see a bleed hooked up to a solenoid, that is t'd into the line running to the wga. When the water/meth injection activates, the nozzle pressure sensor between the water solenoid and the injectors will trigger, which in turn does a couple of things. The first is it will light a small led in the cabin that will be mounted in my gauge pod. The second, and most important, is that it will activate the solenoid with the bleed, instantly increasing my boost to 20-21 psi. The best part is, no pressure to the injectors, then the boost will stay at 15 (peak). But if the water injection system is working correctly, the boost will increase."



and...

"If one of the four nozzles gets clogged the psi sensor wont notice.

The only way to have a system like that totally failsafe would be four separate flow sensors."


so you can see there lots of ways to backstop the AI risk... let's hear from all on the subject.

howard coleman

to be quite honest there is no safeguards you can put into place when running that kind of boost. problem is in rotary applications the intake charge heats up severely under high boost, combine that with already sky high rotary engine internal temps and you are rewarded with pre-ignition of pump fuels. what that means is even a knock box full retarding the timing will not be able to control the spark. the only thing you can do as a safeguard is monitor the fused circuit controlling the AI, if it fails it would have to turn on another system in parallel which to me seems like overkill but to each their own.

i just consider the risk vs the benefit, failures will be rare but it's a gamble you take in the quest for power. best suggestion i have is choose the most reliable and promoted(by other users) product out there on the market.

things happen so fast at those boost levels that almost all safeguards you can think of will just react far too slowly.

Aquamist DDS3

Problem solved

Aquamist DDS3 rocks Ditto mine has arrived in the post last week

Scott do you fancy me visiting you and giving me a hand fitting the thing :) ?

Im installing the Custom Coolingmist Shurlfo water tank as we speak

We will be releasing our safety/tuning device shortly.

http://www.coolingmist.com/products/...isafelarge.jpg

http://www.aquamist.co.uk/forum/gall...8/intro-gr.gif

I use a line presure switch that reverts to WG spring if a relay fails to open, or there is not enough pressure in the line post relay. I have mine set a 80lbs. It has saved me a few times.

I left my unassisted waste gate row (12psi) fat with fuel.

For the higher electronic boost controlled rows I use a relay to energize the Profec B Spec II solenoid. No AI pressure, no increase in boost.

The control for the relay is the Coolingmist clog switch. See diagram.

Barry

We have several answers to the failsafe issue. Option #1 below is by far the most flexible failsafe system on the market with the ability to activate/deactivate devices based on any or multiple inputs including user defined. Also it has a nice gauge based datlog system.



#1 http://www.coolingmist.com/detailmain.aspx?pid=1099
http://www.coolingmist.com/products/.../1099large.jpg
Documentation to our smart injection is here: 28 page PDF

#2 http://www.coolingmist.com/detailmain.aspx?pid=fs1000
http://www.coolingmist.net/products/...onfailsafe.jpg

Option #2 has an integrated flow sensor. We will have these in stock shortly. This one can function as a basic failsafe on its own, or it can be used in conjunction with our smart injection controller. A 0-5 V output can be sent to our smart injection unit and you can set the exact flow you want the system to activate the failsafe (what ever that would be).

http://www.coolingmist.com/images/gauges1.jpg

Has there been any progress or updates in the failsafe field? I'm planning my methanol injection system and I want to make it as safe as possible with current tech. I use my car for road racing so it definitely gets a lot of punishment.

thewird

^+1

I would like more info also. If you have followed my thread, you'll know my AI system failed and took the engine with it. The primary cause we think was uncontrollable boost and a blown AI fuse. Richard can correct me on this, but I think the Aquamist system is supposed to shut down boost if it detects a fault, but as my boost was uncontrollable, this didn't happen. Richard has an idea on why the fuse keeps blowing in my system so I'll speak to him about that later.

My big newbie question is how exactly does one use AI for safety? I don't care about maximum HP as my 3 rotor will make plenty anyway. Gotham Racing (now defunct) talked me into installing a custom Aquamist system as a safeguard on the car. They said they wouldn't use the system for HP gains, but only as an extra safety factor. I thought this happened by tuning the car first without the AI system, and then going back and adding in AI. Is this how it works? Any advice?

The new shop, Kilo Racing, wants to remove the AI system, claiming they don't need it. What should I do?

I really recommend reading about our smart controller. Its the most customizable failsafe on the market. Its clearly more technical than others on the market, but its potential is unlimited.

smart controller details:
http://www.coolingmist.com/info.aspx?key=smartc

Here is an advanced diagram that a customer did with his G35. It shows you its full potential. He even has backup fuel pumps being managed by it.

http://www.coolingmist.com/diagrams/upgradediagram.pdf

David,

The Aquamist Kit works as a failsafe as you design it to. What I mean to say is, when it detects a fault (low fluid level, clog, etc) it puts out a signal (I forget off the top of my head whether it energizes an output or grounds it; 6 and one half dozen) that can be used to open a wastegate, switch fuel/ignition maps (if you've got a Haltech or other ECU that accommodates this), activate a relay that cuts power to your coils or injectors, or anything at all.

While I'm not familiar with the Coolingmist kit, I'd imagine that it does something very similar.

Personally, because I'm not pushing for an in any way ridiculous power goal, I'll tune completely without AI, and then just inject water on top of that at the cost of some power. That way if it fails, I can just keep on boosting until I fix it (hopefully ;))

I'm doing my own system from scratch. The current plan is to have a single pressure switch and have the injector and pump go through a pre programed routine to check line integrity then nozzle flow. Pump pumps up the line with no injection, waits a few seconds to see if pressure holds, then injector flows and the time until the low pressure switch is measured. If the time is too high then the injector/filter is clogged. The routine can be interrupted if WI is needed. This assumes either a small accumulator or the lines expand and hold some pressure. If you're going to use your own controller this might not be a bad way to do it.

Or you can just buy an AEM water/meth failsafe which detects both low flow and over flow situations.

Nice old thread bump.

thewird

"you can just buy an AEM water/meth failsafe which detects both low flow and over flow situations."

thanks for the helpful suggestion. as i see it the new AEM safeguard program is exactly what is needed.

many safeguards key off some sort of peak metric, generally pressure, sometimes flow. the AEM system monitors the delivery curve...

curve as in starting from first delivery. if you wish you can stipulate what is a fail

AT ANY POINT ON THE CURVE

so you get ahead of the curve. (sorry couldn't resist). but seriously, this can generate an engine saving action ahead of many other systems.

check it out. and thanks thewird.

hc

Delete