ZX-Tex

iTrader: (15)

So I have been putting a lot of thought into this lately and am looking to start a brainstorming session on what would be decent failsafe schemes for a WI setup. The idea would be to save the engine should the WI malfunction or run out of fluid and not deliver the intended WI charge.

Examples of failure modes would be (not necessarily in order of probability): clogged jet, pump failure, hose or fitting failure, relay failure, PWM valve failure, wiring failure...

The failsafe schemes will certainly be dependent on the configuration of the WI system and what ECU is being used to control spark and timing. Also, failsafes are not intended to be a substitution for maintenance. Let's assume that the user does regular maintenance such as nozzle checks, only uses clean fluids, and did a good job on the installation in the first place.

Schemes that I have seen and/or considered:

1. The one I have now... When I turn on the WI system, I have a switch on the high side of the pump. That switch, when the system builds pressure, grounds the EBC, thus locking it at 100%, and the MBC allows a higher boost level.
Pros:
- Only allows high boost if the pump is running
- Protects against pump failure, hose or fitting failure, low water level, and wiring failure
Cons:
- Requires a hybrid EBC/MBC setup, or at least a vacuum solenoid.
- Does not protect against clogged nozzles
- Does not allow use of WI at lower boost pressures (below the MBC set point)

2. Use a pump high-side switch to signal the ECU to retard timing and/or enrich fuel if the pump is powered on and there is no pressure. Tap power from the inlet of the motor, and run it through a NC switch, then to the ECU. If the pump has power, and the high-side pressure does not build, then the NC switch sends 12V to the ECU, triggering timing retard and/or fuel enrichment. As an option, add a dash warning light in parallel with the 12V signal to the ECU (easy).
Pros:
- Works at any boost pressure (unlike scheme #1)
- Protects against pump failure, hose or fitting failure, low water level, and wiring failure
- Does not require a hybrid EBC/MBC system (unlike scheme #1)
Cons:
- Does not protect against clogged nozzles
- Requires an ECU that can adjust fueling/timing based on an input signal (hi-low). Adaptronic can do it, not sure which others can.

3. Use closed-loop WB02 feedback at all loads, including high-boost, and use something like a 50/50 water/meth mix. Tune for target lambda with WI enabled. If there is a problem causing reduced WI flow, the lambda value will climb, and the ECO compensates by adding more fuel.
Pros:
- Works at any boost pressure (unlike scheme #1)
- Protects against pump failure, hose or fitting failure, low water level, clogged nozzles, and wiring failure
- Does not require a hybrid EBC/MBC system (unlike scheme #1)
Cons:
- Requires an ECU that can provide closed-loop control at high loads, and adjust fueling, or at least provide a warning.
- Requires a functioning, calibrated WB02, but we ALL have these, right?
- Not sure about closed loop feedback at high loads. IIRC most OEM ECUs run open-loop at high loads, but then again, most OEM ECUs do not have WB02 feedback available.
- If water/meth mixture is being used to augment near or at capacity injectors, then the scheme will not work since a marginally sized injector may not be able to add enough fuel to offset the loss of the methanol.

4. Use a multiple nozzle setup. I am working on this one now. Though it is basically a port-injection setup, it will spray at least part of the WI mixture into a common open plenum. Thus, if one injector clogs, a particular cylinder will still be able to pull a partial WI charge from the open plenum (common to all runners).
Pros:
- Offers partial protection against a single clogged nozzle.
Cons:
- Only protects against one clogged injector. If sudden catastrophic system contamination clogs multiple nozzles then mitigation is reduced or eliminated.
- Does not protect against the other failure modes.
- Requires a multi-nozzle setup, with some added cost and complexity.
- May/will require extensive manifold modifications or a custom manifold.

ZX-Tex

iTrader: (15)

I will probably be dropping scheme #1, and implementing both scheme #2 and scheme #4; They operate independently of each other, and complement each other, so this should be a pretty robust approach IMO. Chime in with your thoughts.

Scheme #3 is appealing, but I am not sure how practical it is.

y8s

iTrader: (8)

you can get a high pressure switch and a low pressure switch to detect clogs or opens as long as you know what your working pressure is (or at least a range).

not sure how well the high pressure switch would work though if normal line pressure is close to pump max pressure.

ZX-Tex

iTrader: (15)

^^^ That is an interesting idea... So the scheme would be to set the switches just above and just below the normal operating pressure range I assume. If there is a rupture, a fault is triggered. If there is a clog, a fault is triggered. Easier/cheaper to implement than a flow meter.

I did not get into flow meters because I have not found an inexpensive way to implement one, especially if a proportioning system was used.

Actually, a proportioning system might screw up the high-low switch scheme too now that I think about it.

Stein

iTrader: (46)

What about something as simple as a flowmeter before the nozzle or before a manifold for multiple nozzles? Forget pressure. Once you know what the flow is on clean, fresh nozzles at various pressures you are golden.

http://news.thomasnet.com/fullstory/557944

For flow rates below 100 ml/min (hydrocarbons)
Highest sensitivity to lowest flow rates
For hydrocarbons various other media: alcohols, oils, diesel, gasoline, adhesives, silicon oils, etc.
Robust housing suitable for industrial applications
Fast response time (< 50 ms)
Superb chemical resistance, inert materials
Straight flow path made of glass, no moving parts
Outstanding repeatability of all measurements
Protection Class IP 65
Analog output 0…5V
Supply voltage 24

This would work for single as well as multiple nozzles as well as variable injection depending on boost if you can set up the logic.

Lose the pump, lose flow.
Run out of juice, lose flow.
Lose single nozzle on one nozzle system, lose flow.
Lose one even one nozzle, flow is reduced by 25% or 50% for a given pressure so should still be big enough delta to know one is plugged.
Degraded flow on all nozzles due to partial plugging, maybe enough to trip it but should be caught on routine maintenance.

Pros -

only one thing to monitor and set up
covers all upstream failures

Cons -

possibly price to find one that will monitor low flow
setting up logic to detect reduced flow vs variable delivery at low boost

EDIT:

Damn, I typed all of that while you were posting that flowmeters weren't an option.

ZX-Tex

iTrader: (15)

No man that is cool, glad you did the pros and cons. Good point too about it not needing to be on the high side of the system.

That is the main deal I see with the flow meters. The sensors themselves are not too expensive, for something like a paddle wheel or a turbine with a hall effect trigger. What jacks up the price is the signal conditioning and the logic (mapped for progressive) to recognize when there is a problem and send the appropriate signal to the ECU and/or driver.

But, I am hoping someone has a cheap, say around $50 approach. Maybe just a dash flow meter, and the driver can recognize when flow is not right with some 'training', would be better than no flow meter.

Joe Perez

iTrader: (8)

With Shurflo-style pumps, the system operates at max pressure pretty much all the time.

ZX-Tex

iTrader: (15)

Come on now guys, I was hoping this thread would spur some more discussion. Maybe it needs some pictures of hot chicks or something...

Nah...

neogenesis2004

iTrader: (12)

I have a somewhat comedic, yet serious proposal. Here it is:

Whatever you all come up with. Make it and send it to Scott to test. If it can break or fail it will happen with him. If it lasts then you know it is made of gold.

neogenesis2004

iTrader: (12)

Additionally, why not rig up something that involves a flow meter. When water pump is active and flow is present you have situation A, and when you have pump on and no flow you have situation B. You could probably write a program for a PIC cpu thats like less than 50 lines that could infinite loop to monitor the output of those 2 situations and out put an on or off signal to something. Like say a ebc bypass, or something else amazing.

GaryR

I've been experimenting with failsafe systems for WI. I've come with two potential solutions. One is the pump carries out a momentary fire at the start of the cycle (increase in MAP) and needs to see a pressure spike and fall within a certain time duration (ms) to indicate nozzle(s) clear.

I'm struggling to find chemically resistant pressure transducer at sensible price points.

Other and preferred at the moment is to use the refractive properties of water / methanol to prove the presence within the inlet tract following the nozzle posistion. Whilst it doesn't prove volume is does provide a no / go confirmation of WI presence. The Swedish pump I'm playing with is capable of 225 psig which I run with Manifold Pressure on the tank via a filter and NRV which ups the ante to around 270psig with the benefits of immediate priming on solenoid opening.

With regards solenoid this is also a problem area. I wish to use a small lightweight fast response solenoid valve near the atomiser nozzle to prevent vapour or siphoning. The only ones that I have found are from the medical dispensing industries and would make a production kit unviable.

Sparetire

The pumps being used now and the solenoids commonly for sale out there are pretty damn relaible as far as I can tell. As opposed to the Shurflos which IMHO suck in compairison.

I honestly think the best way to be safe while taking advantage of WMI is to have a good tune with a well placed nozzle and temp sensor, in additon to good knock detection and WB02.

If the system fails and temps go up, the tune compensates. If the system fails and knock occurs, timing gets pulled and/or the driver is warned. If the system fails and the car goes lean, the WB02 will show it. Or all of the above.

One thing that I think would be cool would be a feature from the WB02 makers or maybe a separate module compatible with the PLX, LC1, etc systems that would kick out a warning light/sound if AFR went leaner than an adjustable point for an adjustable amount of time at 100% throttle. So if you get leaner than 11.7:1 @ WOT for more than .5 seconds it lets you know. Not the most advanced idea, but it would work well for about 90% of us.

You dont see much in the way of 'fail-safes' for the fuel system, and problems there kill more cars than issues with WMI, even among cars running WMI as far as I can tell.

GaryR

I think in many instances whats available on the market is agricultural and / or agriculturally derived with regard WI. In many cases also carries price tags that are obscene.

I take your points but mechanisms for controlling knock / lean runs are available in there plenty. Just down to how far you go....on the wallet.

I'm trying to put forward methods that exceed other manufacturers technoligies with a focus on future development.

I have a pump which I can control via PWM (and vary the pressure) with seals that are compatable with octane boosters such toluene.

Unlike a number of pumps this one will work from a 12volt source.

I'm trying to find positive or negative feelings on suggestions of control.

Sparetire

The Shurflos are definitely ag pumps, they have been around for like 30 years. But the new Aquatecs are not, and they have seals in them specifically for methanol and are PWM controlled. Though I dount they would handle any sort of petrolium type of thing well. Handling toluene would be pretty cool