Stephenz's work

[stephenz]

Hi guys,

I've decided to create another thread more generally aimed at the engine that I will be building; a thread where I will posting thoughts, ideas, simulation results, etc.

I have finished the 3 regenerator simulations and I am currently exporting data and making it presentable, I should be able to post that in the coming days.

While these results make a lot of sense, the most important thing I pulled from this study is that pre-cooling a hot gas exiting the heater, or pre-heating a cool gas coming out of the cooler is pretty easy. However, being able to do both alternatively is what makes for a good regenerator. The only way to be able to do both is by designing a generator that can maintain a gradual temperature rise (or fall) across its length. In order to achieve this, the regenerator material needs to have low thermal conductivity and/or be made of sub-regenerators (i.e. shorter regenerators separated by some distance to prevent thermal conduction from one sub-regenerator to the next).

My immediate reaction was to raise concerns on the housing hosting the regenerator itself. Since I am building an Alpha to test regenerator materials, I immediately realize that my plan would have significant losses through the housing body and the piping joining the regenerator housing to the cylinder heads. These inefficiencies could be easily evaluated through simulation, but based on the results of my "disjoint meshes based regenerator" it's obvious that these losses will not be negligible.

We're not talking about just thermal loses here that could be reduced by insulation, no. The energy losses to the environment through piping/housing body can be minimized but they will not make up for applying an undesirable temperature profile to the working fluid. The thermal mass of the housing body, or pipes combined to their thermal conduction, will significantly reduce the gas temperature (considering the gas flowing from the heater to the regenerator), for a very mild increase in temperature of the walls of the body and pipes, due to the widely different thermal masses at play. Of course, assuming insulation is applied to them, after many cycles the temperature of the pipes and housing will rise to a point where the gas temperature is no longer affected, but considering the thermal conductivity of the pipes and walls will contribute to the reduction of the temperature gradient that we want in the regenerator.


I've read many times that Beta's and Gamma's were 'better' thanks to everything being "closer" (shorter pipes) and even sometimes with bodyless regenerators. But to me this was the first time data explained those statements in a pretty direct manner.

I was pretty set on going with a wobbleplate design with opposed power and displacer cylinders (power cylinders on one side of the wobble plate and displacer cylinders on the other) but now I really don't like those long lines needed to join the hot and cold cylinders... More thinking is needed.

[matt brown]

Common swashplate has a very linear piston motion vs irregular piston motion of any slider-crank, but little chance of any dwell at TDC or BDC. Another swashplate advantage for alphas is a clean cycle potential from 'only' 6 single-acting pistons with 3 hot pistons in one block and 3 cold pistons in the other block, similar this GM Frigidaire A6 (axial 6) automotive air conditioning compressor...

https://www.youtube.com/watch?v=4Ny1VfdXzyg


Most guys think the least numbers of cylinders for a clean alpha cycle is 8 like Rinia or Steele, and likely due to 90 deg phasing penchant. However, 6 cylinders is the least and yields 3:1 compression ratio vs 8 cylinders yields 2:1. Despite clean cycle potential, the major downside of multi-cylinder schemes is the large conduit volume (as you point out).

[Bumpkin]

The Stirling bros made a pressurized double-acting engine with a segmented regenerator. If I remember the literature right, they filled-in the spaces between with broken glass to minimize dead-space. The duct itself contributes to regeneration; but since it is also a short it seems that at least in one way a longer duct has the advantage of being a longer short. Not really arguing for longer passages though. In relatively high-speed engines it seems the matrix/exchange-material conductance would be proportionately less significant as gas speed and convection increases, but conduction remains the same. I think there might be significant radiant effect to/from the matrix. If the regenerator is segmented or of low conduction, I don’t see any reason to have separate heater-regenerator-cooler; so much as just different zones of the same system, externally heated, externally insulated, externally cooled. So the point would be if you have to use a long tube, try to put the whole thing to work.

Bumpkin

[VincentG]

You alpha guys should consider the 270 degree parallel twin. Very similar if not better phasing than a 90 degree v twin and a far shorter transfer port length. Also tons of cheap worn in Japanese made motors to start with.

[VincentG]

Stephenz, I randomly found this article while doing my own research and thought you may find it interesting.

https://www.techbriefs.com/component/co ... rials/2113

[stephenz]

Hey guys,

Here's a quick illustration of the data collected through these simulations.

https://file.io/MsZiGIV5zZaG

All three are obviously running the same conditions:
- 600K Helium
- Constant flow rate
- Meshes have an initial temperature of 292K

The top sample is a 0.2mm thick steel mesh with 0.4mm holes (linear pattern, non staggered)
The middle sample is a 0.1mm thick steel with 0.4mm staggered holes (i.e. about 30% more holes)
The bottom sample is a 0.1mm glass mess with 0.4mm staggered holes (only a material change from the middle sample).

I had already given my conclusions in the other threads but the main takeaway for me were:
- it doesn't take a whole lot of material for the working gas to drop their energy
- as such it is super important for the regenerator housing not to be thermally conductive as it would easily result in thermal losses (the regenerator is supposed to hold onto the energy deposited into it, to be restored half a cycle later)
- the regenerator material must be of low conductivity as well to maintain the greatest temperature gradient between its inlet and outlet
- disjoint meshes are a good way to physically insulate mesges from another
- for the temperature gradient to be the greatest, individual meshes must have a thermal mass much smaller than the overall thermal mass of the working fluid, so that each layer is only bringing the working fluid temperature down a little

[stephenz]

My second update is about my test build around a 200cc Alpha configuration.

Long story short I was not able to get it to work but I've learned a lot.
- my attempt at using teflon rings was a pretty bad idea just off the fact they need a special tool to be inserted which I didn't have. I was able to fit them on, but the resulting friction was simply too much
- I then used the original piston (and rings) which resulted with the best results. Spinning the flywheel by hand with no heat clearly showed a fair amount of compression (5-10 psi over rest position), and heating the burner clearly made things easy and it really felt like the engine wanted to start
- I had not designed it to be used with a burner but rather an electrical heater (1500W in a hot plate) to measure the amount of heat I was putting in. Sadly the heaters are not here yet so I decided to test it on a mapgas. Sadly despite the greater temperature it didn't start, so my hopes of it working off an electrically heated source are gone. Which means the sealed regenerator which I designed for 400F max is not going to work either.
- I did another test without the piston rings and the clearance between piston and my cylinders is just too much and spinning the engine by hand is effortless with minimum amount of compression on the gauge.


I had been scratching my head about how to deal with temperatures, pipe lengths, etc and it struck me that the ideal configuration for me would be to convert this Alpha engine into a beta. By moving the cooler (and regenerator) to the displacer. With this V90 configuration, the pipe joining the 2 cylinders won't be that long and it's technically a cold pipe, and such would act as cooler since it's in contact with the ambient air, so no losses there.

Of course new (ringless) pistons are needed. So I will first get new pistons with good tolerances so I can at least retest the present alpha configuration with good compression (and no temperature issues with rings scratching my fragile aluminum cylinder).

If it works great, if not, I will also start preparing parts for the gamma conversion. I always thought the displacer/regenerator all in one was an elegant thing, so I guess I'll get to try that.

[VincentG]

Thanks for sharing your failures as well Stephenz. It could save people alot of time knowing what not to do. Pics would go a long way so we know what your building.

The gamma configuration really seams to be the most diy friendly.

[stephenz]

I've attached a couple pics. That makes me realize I didn't document the assembly much.

What you see is a 3.7HP harbor freight compressor, stripped down. I had initially remade pistons but as mentioned in my previous post, the teflon rings were a bad idea and settled back on using the pistons that came with the pump (tried them with and without rings). I designed the cylinders and heads and got those made on my company's CNC.

The hot cylinder is stainless steel with a brazed copper heater (fins and channels cut to increase surface area and connect with the 6 pipes connecting tot the regenerator).

The cold cylinder is aluminum with copper cold plate with fins on both sides (the piston side is very similar to that of the heater, the other side are just extruded fins covered with a brass body for water to circulate through).

The regenerator housing is brass (was made before the analysis) - with what I learned from the analysis I would not have designed the regenerator this way at all. The inside is made of a bunch of stacked stainless steel blades.


I'm still hopeful the redesigned piston (with lower clearance) will get the engine to at least somewhat work.
I'll implement everything I learned so far in a beta conversion using the same pump as base (crankcase, crankshaft, rods, etc.).

I use this pump as pathfinder as ultimately the goal is to make a 8 cylinder wobble plate based on the design (same piston/cylinder size/design).

[VincentG]

Looks like a great start and excellent machining and fabrication. Before you give up on the alpha I would try...

-ensure piston squish clearance at TDC is at the minimum possible around .020" for a low rpm engine.(pull up on pistons to take slack out of bearings when measuring)

-you know by now the stacked ss plates have far too much thermal mass

-try to block off all but one or two of the transfer tubes to reduce dead volume to a minimum. What seems like a flow restriction I don't think will matter, and I believe will enhance the heat pump effect overall.

-same for regenerator box, minimize volume possibly with an internal passage made of an insulator. I don't see the need to make a new regen housing when it's a great modular shape to experiment with inserts.

With access to such great machining and CAD skills you are sure to have a great runner soon.

[stephenz]

Thanks Vincent.

Yes the regenerator SS stack is way oversized. I was actually thinking of removing the stack and replacing it with stainless steel wool. The thermal mass will be much less. If that still doesn't work it will probably mean too much dead volume and at this point filling some of the volume will be the way to go.

Squish is around 0.4mm - I measured it off the cold cylinder - I can't double check it on the hot cylinder (since the cylinder and heater are brazed) but by design they have the same value.

I have a feeling the new low-clearance pistons will get it going - if not, tweaking the regenerator might.

[VincentG]

From a purely machining stand point, how were the id bores of the cylinders finished, raw off the lathe or honed? If lathe what inserts were used?

[stephenz]

I can ask what lathe insert was used. I didn't specify any special surface finish, just a somewhat tight tolerance.
Parts are technically made on my machines, but I rarely do much machining work myself anymore. I have a few full time CNC operators.

[matt brown]

The time honored method for testing any alpha is running it 'backwards' where an electric motor attempts reefer effect. Many alphas (and other SE) that fail as engines often succeed as reefers (albeit poor ones). An alpha is more limited by heat driving the system to produce work than by work driving the system to produce cold. This apparent dichotomy is often reduced to: all work can produce heat, but all heat cannot produce work. Not the most accurate statement, but it gets the general idea across.

As a long time alpha guy, I'm very familiar with all the issues. Putting aside heating and regen issues, the compression ratio is the main issue. Most SE, including alphas, have a very low compression ratio whereby in ICE buzz, the backwork ratio is very high. The easiest way to lower the backwork ratio is by simply increasing the thermal ratio, a la Carnot. Most 'successful' comm'l SE use a high thermal ratio combined with a high buffer pressure (increases cycle rate). But in the DIY camp, a high thermal ratio and buffer pressure is not happening. I think a careful parametric study on compression ratios trumps regen and heating.

[stephenz]

@Vincent: we used standard high speed steel finishing inserts on both the aluminum and stainless cylinders. Not the best, but I will probably specify a decent RA next time.

The new ringless pistons will be here this week, hopefully I regain some of that compression and play with the regenerator to get it going.

- Regenerator wise I think I have found the material (and supplier) for ideal regenerator. Cordielite. Probably nothing new to any of you guys. This material is typically extruded in either square or hexagonal patterns (these extrusions need constant wall thickness, so holes aren't typically done). The conductivity is under 2 W/m.K and the specific heat above 750 J/Kg.K which makes it close to 10 times better than stainless steel from the specs alone. I am rerunning a set simulation with this structure (0.90mm squares and 0.10mm walls) but instead of a stack of meshes, I am doing a single longer extrusion, 8mm. I have the 20mm diameter. I am also going to run 5 simulations at 5 different engine regimes (100 RPM, 600 RPM, 1200 RPM, 1800 RPM and 3000 RPM). I want to see how the same regenerator design behaves under different regimes. The higher RPM means higher mass flow rates, but also much shorter transfer times, and each simulation will be run for the transition time according to the engine regime.


- Piston Sealing: when I was venturing seriously into Alpha's the practical issues of piston/cylinder temperatures were pretty serious. But with a Beta design which is what I am considering for the next prototype, both pistons technically run at temperature close enough to ambient where material requirements are too difficult to meet. Considering lubrication, I am leaning toward oil spraying/pumping inside the case and using an oring with low compression (dynamic seal specs) in the upper part of the piston and keeping a 0.01mm clearance between piston and cylinder. And because both pistons run at temperatures close enough to ambient going with dissimilar materials shouldn't create differential temperature expansion problems. So perhaps a stainless steel cylinder and a 6061 or 7075 piston.