www.pulse-jets.com

I threw together a model of a Lockwood engine that Andreas Sunnhordvik posted. The basic idea was to test uflow1d against a well-known design to see if my technique for modeling the combustion ICs was sound.

It seems like a decent model. You can actually see the short arm of the Lockwood "breathe". The flow in the model seems to match this figure reasonably well.

Dear Pezman.

Have a look at the graphs posted here
http://www.pulse-jets.com/phpbb2/viewto ... 998c422a4c

It'll give you some more realistic starting values.

Graham

Actually, it looks like the model is reasonably close to this, except that the pressures are a little high, the temperatures are a little low and the all portions of tube that are not part of the combustion zone are at ambient conditions.

The model is just one pulse from a cold-start and intended to be as simple as possible. Basically, my goal was to see if uflow1d would model the reversal of flow in the short arm (which it does).

My real goal is to model a "Bruno-O" engine and my current design does not draw any air. I was just trying to sort out whether this was a design flaw or a uflow1d anomaly. Since the lockwood model works, I'm leaning towards design flaw.

BTW, would you be willing to post the Lockwood model that you created based on that data?

Dear Pezman.
Could you try and model a Lockwood type motor under steady running?
Think of the flow conditions and temperature gradients in the induction pipe and tailpipe. Start at either peak combustion pressure or minimum combustion chamber pressure.

Graham.

Not quite what was requested, but here's a "Bruno" engine that actually breathes a little. I dubbed it the "Bruno O'Lockwood" since it's derived from the Lockwood design.

The attached model just does one pulse from ambient initial conditions. The combustion occurs at the left end of the engine. Some exhaust squirts out the left end and the rest shoots over into the right chamber. The restriction in the middle seems to cause enough vaccuum in the left end to suck in a puff of air. The engine is designed so that if combustion at the right end is timed to pccur at the peak of the pressure pulse, the resulting right-to-left pulse will arrive just about when the left chamber is done gulping a puff of air. Note also that the

I think that an engine based on this design can be made to run with a bit of tweaking. My guess is that the overall length is a bit too short for the higher speeds that will occur when the air is hotter. The initial pulse gets a 2:1 compression ratio. Hopefully, subsequent pulses should give increasingly high compressions up to some limit.

I tried to use the conditions at about the 17th step in the model as ICs for the next pulse, but UFlow1d crashes every time. This makes me reluctant to invest much time in a refined model (Lockwood, Bruno or otherwise). Anyone know of a better modeling package?

mike

Dear Pezman.

Try reducing your mesh length until it Uflow stops working then go back a bit.

Graham.

Hi Mike,
Graham is right, 50mm mesh length is quite large. Otherwise the model looks interesting. I haven't seen any other freeware for this, other than some Fortran stuff available on the same site as UFLOW. It's remarkably powerful when you master it. Sometime temps can be what screws it up, so it's good to get it resonant with none and slowly add what you think are the "real" initial conditions and temps.

That worked -- apparently it's all in the wrist ;).

I tried successive runs with ambient temps and pressures and it looks like the compression tops out at about 7:1. The pulses in the combustion zone are horrific though -- 50 bar! (using 2200K/300k * 1 bar * 7). I also had to shorten the arms to almost nothing. Basically, its two washers jammed into two truncated cones that have their tips glued together.

Now that I know how to get the models stable, I'll either try to reverse engineer the IMM file format or come up with an excel program that can extract parameters from a point in the run and automatically feed them back in for another iteration.

10x for the help.

mike