Oscillating Engine Simulator

(My apologies if this is not the correct section of the forum, I couldn't work out a better fit).

I've been working on a simulator for oscillating engines that I would like to share in case anyone is interested, you can use it online: https://incoherency.co.uk/oscillating-engine/

The colour in the cylinder is an indication of the pressure (red is high, blue is low).

The colours shown on the flywheel indicate the port timing (red means inlet port open, blue means exhaust port open, the thickness tells you how wide it is open).

The plot on the right-hand side is a pressure-volume diagram, with the blue and red horizontal lines representing atmospheric pressure and inlet pressure respectively.

I think it should mostly be physically accurate, the main weaknesses are:

1. it assumes the supply pressure in the inlet does not momentarily drop when the port is open, even if you make the hole really large.

2. I don't have a decent model for the inherent friction losses in an engine, so you might find that the simulated engine ticks over substantially faster or slower than a real one with the same parameters - in this case please let me know and I'd be keen to integrate your data into the loss model!

If you want to know more or need help using it just ask. I also have written a blog post that goes into more technical details: https://incoherency.co.uk/blog/stories/oscillating-engine-simulator.html

Edited By James Stanley on 20/03/2023 14:16:37

Edited By James Stanley on 20/03/2023 14:16:51

Interesting addition - Dockstader treatment for oscillators! Well done.

I am intrigued by the compression loop and even more intrigued by the expansion loop. So what is going on there?

I suspect managing those loops could be key to getting good running.

Martin

Fantastic. Can we persuade you to do one for a compound steam engine?

I would have replied sooner but have had to drag myself away from reading more of your website. I haven't seen anyone mention Lisp for 30 years and didn't realise people still made their own CPUs. I am possibly the only person who still has a set of the newsletters of the Amateur Computer Club that operated in the seventies. It includes the club design State Machine CPU (type Nand2Tetris made I think). In '76 a friend of mine built a CPU using a 6 in square matrix of diodes for the microcode so it was limited to 100kHz clock.

I'm impressed too: not just the simulator but James' other activities. More fascinating reading when I should be working. Anyone who introduces himself with 'The writing is bad and the stories are not very interesting' is my kind of guy!

More please!

Dave

Excellent, keep up the good work. Of course you now need to confirm results by experiment. Not stupendously difficult, pressure sensor on cylinder, slotted wheel on crankshaft and a data logger.

Thanks for the kind words! And Bazyle I don't think very many people make their own CPU any more, but if you are interested there is a great page at **LINK**listing some of them.

> I am intrigued by the compression loop and even more intrigued by the expansion loop. So what is going on there?

If you assume that the inlet and exhaust ports are symmetrical, then it is always the case that either the port timings overlap each other, or there is a range of positions, crossing top dead centre and bottom dead centre, where neither port is open. This means there is no way to avoid at least a little bit of unwanted compression and expansion, without letting both ports open simultaneously.

I don't have a good intuition for whether you would be able to solve this by making the port locations or sizes asymmetrical - my guess is no, but it might be interesting to fiddle with.

I don't think this unwanted compression/expansion actually costs any *efficiency*, because the work extracted in one cycle of the engine is given by the volume enclosed by the pressure-volume diagram. The pressure spikes enclose no volume, so they waste no energy (i.e., the energy you spent compressing the air on the approach to top dead centre is given back when the air expands after you pass top dead centre). It does however slow down the engine: compare rolling a ball up and down a frictionless hill to rolling it along a flat plane. The ball loses kinetic energy going up the hill and regains it coming down the hill, but it still takes longer to travel any given distance because the average speed is lower.

> Can we persuade you to do one for a compound steam engine?

Quite possibly! There are 2 different directions I want to take this. One is to make it simulate more than just simple single-cylinder single-acting oscillating engines. It would be relatively straightforward to split out the part that computes port overlap areas, and turn it into a pluggable "valve operation" module, where you can select what type of valving the engine has. Making multiple-expansion engines should be relatively easy as well - you'd probably model it as a fixed-volume vessel connecting the exhaust port of one cylinder to the inlet port of the next, and all the air flow calculations remain pretty straightforward. Double-acting cylinders would be similarly straightforward.

Probably the most annoying parts of supporting different types of engine would be making the visualisation and the user interface handle all the different possibilities.

The other direction I want to take it is to have it take some constraints and goals, and then it would automatically fiddle with the engine parameters to try to optimise the engine. For example, with displacement below 5cc, with 20 psi air supply, find the most efficient engine that can supply at least 1 Watt, below 1500 rpm.

> Not stupendously difficult, pressure sensor on cylinder, slotted wheel on crankshaft and a data logger.

Yes, I think this would be very interesting to do. Sadly I'm much less competent as a machinist than as a programmer, so my engine build is progressing much more slowly than my simulator did. But I definitely want to try to plot a pressure-volume diagram from a real engine. I suspect the limiting factor will be finding an inexpensive high-frequency pressure sensor.

This will be useful to me. I have a steam engine with several parts missing, including the piston and cylinder. I can use this to work out how to make the best replacements. 🙂👍

I'd be really interested to hear how that turns out. I'm not sure how applicable the results will be if your engine is not an oscillator however.

The latest version now:

• can simulate double-acting engines (but still only oscillators)
• has a new air flow method option, from a paper written by the late Bill Hall, sent to me by Duncan Webster
• plots torque curve charts
• plots pressure-speed charts

Thanks for the clarification James. There is certainly more to these simple little engines than meets the eye.

Martin