Wednesday, January 14, 2015

Gas mixing prototype v2: Not a success. But not a total failure.

Gas mixing prototype v2 was somewhat more successful than v1, but I think in the end, I am going to abandon it.

Main differences with the v2 prototype:
  • Using a metering pipe and adjustable regulator for measuring propane.  The idea is to use a calculated length of pipe (4" in this case) at a particular pressure to measure out a volume of propane.  Knowing this and the size of the ignition chamber allows you to create a relatively precise fuel/air ratio.  The adjustable regulator allowed me to experiment with different ratios by adjusting the pressure feeding into the pipe.
  • I used an electric pump (got it off of Amazon I think) to circulate the fuel/air after measuring.
  • I substituted the piezo spark gap ignitor (which I didn't think was highly reliable) with an electric one from Sparkfun.  (It appears they've since retired it, probably because it's kind of dangerous.  But it works great.)
Here's the v2 prototype:

Here's the v2 prototype in action:

As you can see, it works.  We get ignition, and you can also see that my makeshift pressure relief valve works exactly as intended: it's just a piece of metal held onto the top of the tube with rubberbands.  The top of the tube is rimmed with silicone caulk to make a good seal.  When the pressure rises in the tube from ignition, it blows the top off releasing the pressure.

The problem with the v2 prototype is that, no matter how far down I pushed the fuel/air ratio, I wasn't able to get substantially slower flame propagation.  I titrated pressure from 80 PSI (which should have been the optimal fuel/air ratio) down to around 30 PSI (which should have been below the fuel/air ratio required for combustion.)  At the bottom end, I titrated in 1 - 2 PSI increments, and basically found that below about 33 PSI, I got no ignition at all.  Above 33 PSI, I would get ignition, but the flame front propagation is essentially instantaneous to the naked eye.  At very low fuel/air ratios, the pressure produced by the ignition was definitely lower: instead of popping the top off, it would simply escape from the tube with a hiss.  But the flame front propagation never got anywhere close to 10 cm/sec.

My hypothesis here is that 10 cm/sec is the limiting rate for deflagration in free space at atmospheric pressure.  In my experiments, the ignition chamber is highly anisotropic, and my hypothesis is that the driving factor for flame front propagation isn't heat diffusion (as it would be with deflagration in free space) but is being pressure driven by the space constraints.  Essentially, the flame front is getting pushed upwards rapidly by the pressure.  (In a sense, this is similar to the physics of detonation, as opposed to combustion, in which the pressure wave drives the reaction).

As it turns out, computer modeling combustion is really hard, and there are no open source tools for doing it, because presumably if you've developed a tool to do it, it's worth a lot.  There are a bunch of proprietary ones on the market.  It's not outside of the realm of possibility that I could write one myself, at least a basic one, but it requires simultaneous modeling of chemical reaction, chemical diffusion, and heat diffusion rates and dynamics.  Suffice it to say, this would be ambitious, and it's not clear that it would work well enough to help me design something that works.  I may try some further experiments with some configurations of air/fuel mixing in various ways (adding holes to the tube in various ways, etc.)  But it would be mostly to help my intuition about combustion.  I doubt I'm going to come up with a clever way to get this to work just by kind of cogitating on it.

Fortunately, posting the blog to my burning man group resulted in some really great feedback from my campmates about ideas for a mechanical solution.  I had been thinking about mechanical solutions before, but the ideas I had weren't very good.  The new design that I'm working on based on the ideas that I got from my campmates are much better, and I think this is going to be the primary direction of my new attempts.  I'll post some schematics and thoughts soon.

No comments:

Post a Comment