Engine Pressurization

[stephenz]

It seems you picked a good starting point for regenerator size then. I think 600rpm is realistic for a LTD style engine as well.

Yes, basic calorimetry can do a lot. q=m.c.DT - at least in terms of figuring out scales.

The simulation tool I am using can't do moving walls, so I won't be able to simulate the gas movement off the moving piston. Too bad as I would have liked trying to run something to try to replicate some of the effects discussed here.
But for approximations of the heat transfer within regenerator/heater/cooler it will help.

The main thing I am looking for in the next steps is the average gas temperature across a plane past each meshes.
Ultimately when you isolate the renegerator itself, the temperature of the gas exiting it will be the same temperature as the last mesh in the stack.
If you consider the gas flowing from the heater into the regenerator, it (the regenerator) is supposed to pre-cool the gas before entering the cooler. In this direction, the cooler the gas is leaving the regenerator the better.

[stephenz]

Hmmm, this made me realize I botched a recent post. As a gamma (but similar for beta) I assume your 4 strokes are,
and as always, considered as distinct ideal events with 300-600k cycle and regen:

(1) displacer TDC-BDC where TDC hot space (clearance volume) is minimal, but BDC space is noticeable, this event ends with Pmax thruout engine, 600k in hot space, and 300k in cold space

(2) piston TDC-BDC expansion lowers P thruout engine and further lowers T thruout cold space (and piston cylinder)

(3) displacer BDC-TDC ???

(4) piston BDC-TDC compression increases P thruout engine

If correct, then note that during expansion (2) T is <300k unless heated by "ambient" 300k which would add to Wpos. Of course, as an isolated event, an isothermal expansion following by an isothermal compression is a zero sum game, just like a matching pairs of adiabatic processes. The question is what happens during (3) ???

With just a single working 'oscillating' piston at a constant temperature, the only Wnet is via PV=mRT where that little "m" represents dP due to a change in density. This akin MEP (mean effective pressure) in the ICE world. Since piston cylinder volume is part of the cold space, and this cold space is ideally considered a constant temperature, a little quiet reflection should illuminate MEP limitations. Yep, since high compression appears impossible, the only way to achieve large pressure swings is with multi bar buffer pressures.

As a sidebar, note that the above sequence is 4 'clean' 180 deg strokes, and no alpha ever had it this easy.

I gave this some thoughts in the past, drawing diagrams, I remember getting stuck but felt like I should be able to overcome it.
The basic idea is a 4-stroke alpha (with 2 real power pistons) is to synchronize piston strokes, i.e. 180 degree phase shift: while 1 compressing, the other is expanding at the same time, with valve closed, so the two cylinders don't communicate, this step is followed and preceded by a transfer step at constant volume, since both pistons are opposing, if the valves are open, then fluid is displaced, the last step is the one that was giving me trouble. I'll eventually get back to that concept. I'll mention that when I was thinking about this concept I actually was constraining myself to hacking a commercial 4stroke 2-cylinder engine (or 2 one cylinder) to have a solid base for prototyping purposes. Just to be clear, I have not given any thought about what cycle this would be the closest to, but it all started off trying to improve upon an Alpha design.

[matt brown]

It seems you picked a good starting point for regenerator size then. I think 600rpm is realistic for a LTD style engine as well.

I agree, great reference points Stephenz, perfect for 1-2 HP. I consider 600k the standard DIY limit and focus on 600 rpm performance, but allow for 1000-1200 rpm when valid. My wildcard is buffer pressure with 5-10 bar on the low side preference.

BTW the Ericsson cycle doesn't have most of the regen issues that plague the Stirling cycle. Stirling is usually considered closed cycle & valveless vs Ericsson is usually considered open cycle & valved. The major Ericsson disadvantages are (1) Cp regen vs Cv magnifies regen inefficiency (2) volume ratio relative thermal ratio...only Otto & Stirling have thermal ratios unrelated to volume ratios. The first issue is so well known that the US gov't restricts Cp engines (few know this) but the second issue passes by under the radar. I have a limited pain threshold for Stirling (phasing, regen, isothermal) so I mainly chase Otto ECE solutions.

[matt brown]

As a sidebar, note that the above sequence is 4 'clean' 180 deg strokes, and no alpha ever had it this easy.

I gave this some thoughts in the past, drawing diagrams, I remember getting stuck but felt like I should be able to overcome it.

Outstanding !!! It sounds like my 20 year old widget design (as I call it) that can run any major cycle at any compression ratio. The bottom line is that there's better (and simpler) ways to achieve isobaric Ericsson & Brayton, and it's rather lame for Otto. Oddly, if you figure this out, there's a similar way to further simplify this same scheme but Stirling only, and just as difficult to discover. In recent threads, you may have noticed that I mention sitting around for 15 yrs trying to solve a regen issue. Yep, I was focused on phasing until I solved this, but then a major regen issue appeared that I didn't resolve until last year (far worse then anything I've seen here lately). I'm still considering getting a patent, just so everyone can see the 'impossible' and no one will waste anymore time chasing it.

Here's a clue: western algebraic equations are typically yadda-yadda-yadda=0, but this bugger reduces to yadda-yadda-yadda=1, typical Egyptian and Greek math.

The common idealized 'clean' alpha with 4 distinct events has 2 equal wait states (90 deg after each transfer), but also 2 unequal transfers (90 deg and 180 deg) which mess up everything. However, the gamma sequence that I posted has 4 equal 180 deg events, thanks to the low pressure 'reservoir' (cold space) that the piston compresses into and simplifies scheming (a work in progress).

[stephenz]

The next simulation is more representative of a single stroke, with ramps up and down.This way I'll have a better idea of the final temperatures for the gas and mesh.

Cooling-Mesh-Settings.png
Cooling-Mesh-Ramps.png

Have you guys every try to run basic calorimetry analysis to size your regenerators (or coolers/heaters for that matter)?

Here's a look at the temperature profile at T = 3.3ms under the conditions listed above.
The temperature profile across the tube shows the dramatic impact of the mesh on the fluid.

I'm pretty 3-5 meshes, under the same conditions should be enough to bring the temperature of the gas down to within 50K of the initial mesh temperature.

If there is anything worth understanding from this, is if a 0.2 gram steel mesh like this one is capable of affecting the gas temperature this much, it is real clear that thermal losses are a nightmare to avoid for a gas.

T3.3ms-temperature-profiles.png

[VincentG]

Thanks Stephen. This definitely illustrates the importance of material selections for internal components. Insulation should be a key design goal at locations where we want to maintain gas temp.

[stephenz]

Thanks Stephen. This definitely illustrates the importance of material selections for internal components. Insulation should be a key design goal at locations where we want to maintain gas temp.

Yes but not just insulation - I think this also illustrates why a thin and low conductivity material (as regenerator material) is absolutely needed:

Conduction within the regenerator material will cause it to even out its temperature.
What we want is a material that has a clear temperature gradient between its inlet and outlet (remember the graph I hacked adding hysteresis to it).

The thinner (cross section of the wire making up a mesh as an example), and the lower conductivity, the better - as both of these parameters will reduce conduction within the material itself.

That's also why some papers favor a finite number of meshes spaced out by a small amount. In this arrangement, even if the mesh has some thermal conduction within itself, it will still have a average distinct temperature different than the next mesh and so on. With enough of these in series, you create some sort of thermal ladder.

[VincentG]

Agreed. Curious if your software can show a cross section analysis as well?

[stephenz]

the view on the right is a cross section of the tube/mesh
I am saving some animations with a zoomed in on the mesh cross section if that's what you're looking for.

[VincentG]

Thanks yes thats what I was hoping for.

[stephenz]

Thanks yes thats what I was hoping for.

I'm still exporting the temperature of just the mesh, but here's the one you were interested in:

both fluid and solid temps are shown here, with the same scale.
https://file.io/Nj3ClUq4v1BT

after that I'll start a simulation with 4-5 of these meshes.

and finally I'll do one with way thinner meshes, as it's only way to get the mesh to retain heat in the form of high temperature. Right now, the hotter the mesh can get is about 360K which is really far from the 600K. That's because the thermal mass of this mesh is just too high, i.e. it can absorb a significant proportion of the heat available from the gas without significant temperature rise.

[stephenz]

steps.jpg

just started on the new sim with meshes and it's early on but figured I would share as it really illustrates the "steps" I was talking about earlier.

pardon the quality of the screenshot, I am home and just screenshot my remote desktop.

[stephenz]

and of course your can see the clear benefits of putting disconnected meshes in series.
5 layers are the gas temperature is already close to Tcold.

next improvement is making those meshes significantly thinner to reduce thermal mass, and of course adding more of them to compensate so that the gas exiting the last mesh is close to the target temperature (in this case, Tcold).

[stephenz]

can't edit my post, I realized I forgot a word. That new simu has 5 meshes in series, 2mm apart. The dimension of the mesh is the same as the single mesh simulation just before that.

[VincentG]

With a regenerator that effective, the standard phasing may be even more of a problem with beta/gamma. And with alpha, some kind of dual one way flow control timing valve might be of great benefit.