Quick update at roughly 1.5 s of simulated physical time.
It hasn't reached steady state yet and the amplitude of the temperature swings on the cooler and heater are getting much closer in size.
Thinking more about it I think they should eventually converge to being the same amplitude.
Stephenz's work
-
stephenz
Re: Stephenz's work
Here are the final numbers for the full 3 second of simulation at 600 RPM:
-
stephenz
Re: Stephenz's work
Now that I have somewhat of a decent workflow established and a pretty good understanding of the parameters that have a good influence on the regenerator efficiency I can start some sort of iterative design process based on available materials and processes.
My first attempt will be to revisit stainless steel meshes with reduced material thicknesses, stamped to shape and stacked to created a displacer/regenerator combo. I really think this gamma configuration is the best for what I am trying to do.
My first attempt will be to revisit stainless steel meshes with reduced material thicknesses, stamped to shape and stacked to created a displacer/regenerator combo. I really think this gamma configuration is the best for what I am trying to do.
Re: Stephenz's work
Have you looked at open-cell metal foam? If it wasn’t for the cost I’d have already checked it out myself. Plus the worry of melting what I consider the most desirable aluminum version — says it’s good for 1000 f though. But there are other metals available.
https://ergaerospace.com/metal-foam-material/
Bumpkin
https://ergaerospace.com/metal-foam-material/
Bumpkin
-
matt brown
- Posts: 488
- Joined: Thu Feb 10, 2022 11:25 pm
Re: Stephenz's work
Excellent work Stephenz, a winner !!! I always consider 600rpm the sweet spot for DIY piston ECE. Maybe provide for 1000-1200rpm, but optimize for 600rpm. For the average guy who happens upon this thread, the main takeawy is how small a good regenerator should be, not how big.
-
stephenz
Re: Stephenz's work
I've used copper foam before in several unrelated application, it's a pretty amazing stuff. Neither copper or aluminum are good regenerator materials due to their high thermal conductivity. But there are stainless steel foam as well which I am still considering. For now I am ignoring them as I have no way of quantifying their behavior.
I spent a couple weeks trying to model porous materials in CAD to no avail. Neither the software are hardware are capable of meshing this kind of structure. I might still end up using them once I figure out the parameters: material properties, wall thickness, aperture, and of course pressure drop. Say I figure out the size of the mesh I need, I can easily interpolate the mesh structure into a metal foam equivalent (mostly looking at pressure drop, overall average density of the regenerator).
I spent a couple weeks trying to model porous materials in CAD to no avail. Neither the software are hardware are capable of meshing this kind of structure. I might still end up using them once I figure out the parameters: material properties, wall thickness, aperture, and of course pressure drop. Say I figure out the size of the mesh I need, I can easily interpolate the mesh structure into a metal foam equivalent (mostly looking at pressure drop, overall average density of the regenerator).
-
stephenz
Re: Stephenz's work
I've decided to re-run that long simulation (0-3s) and only change the length of the regenerator.
Instead of going to a completely different geometry/design I thought it would be good to see if the length of the regenerator had any effect on the amplitude of the heater/cooler temperature swings.
So I've increased the length of the regenerator sample by 50%, which also increases its mass (and probably the settling time as well); but I also have a feeling it will probably reduce those temperature swings, here's why:
- take this scenario where I am only changing the length of the regenerator
- the RPM, hence the transfer time are unchanged
- the amount of fluid being transferred is also unchanged
- since the regenerator is longer, it will hold a higher capacity, and the fluid will be in contact with the regenerator for a longer time, hence greater heat transfer per transfer, i.e higher rate of heat transfer
- as such, the temperatures at the ends of the regenerator should now be closer to the cooler/heater temperatures
- which means the temperature swings of the inlet/outlet should be reduced
If there is a meaningful improvement I think it will be worth repeating that simulation with say 3 or 4 additional lengths to see if the relationship is linear or not.
Instead of going to a completely different geometry/design I thought it would be good to see if the length of the regenerator had any effect on the amplitude of the heater/cooler temperature swings.
So I've increased the length of the regenerator sample by 50%, which also increases its mass (and probably the settling time as well); but I also have a feeling it will probably reduce those temperature swings, here's why:
- take this scenario where I am only changing the length of the regenerator
- the RPM, hence the transfer time are unchanged
- the amount of fluid being transferred is also unchanged
- since the regenerator is longer, it will hold a higher capacity, and the fluid will be in contact with the regenerator for a longer time, hence greater heat transfer per transfer, i.e higher rate of heat transfer
- as such, the temperatures at the ends of the regenerator should now be closer to the cooler/heater temperatures
- which means the temperature swings of the inlet/outlet should be reduced
If there is a meaningful improvement I think it will be worth repeating that simulation with say 3 or 4 additional lengths to see if the relationship is linear or not.
-
stephenz
Re: Stephenz's work
I finished the simulation with a longer (60 mm vs 40 mm) regenerator.
The same 25mm diameter and the same 0.9 (hole) x 0.1 (wall) structure.
I was a bit surprised by the results but it looks like 40 mm was more optimized.
60 mm is giving inferior results with temperature swings on both cold and end slightly greater than that of 40 mm
here is the 60 mm plot:
In order to visualize better, I put both the 40 and 60 mm plots on the same graph, but only kept the temperatures for the heater and cooler surfaces (I removed the temperatures of the regenerator ends for clarity):
There you can clearly see how the temperature swings are slightly worse.
The next step was to start a simulation with a shorter regenerator: 20 mm (everything still unchanged), and without spoiling it, the results will also be worse (quite a bit worse this time) than the 40 mm version.
That means I will probably need to do a 30 mm, maybe a 50 mm version as well.
I also means 40 mm is probably pretty close to the sweet spot in this particularly configuration.
I wish those simulations weren't as time consuming.
I will need to revisit the impact of larger diameters regenerators and also play with the wall thickness as well.
The same 25mm diameter and the same 0.9 (hole) x 0.1 (wall) structure.
I was a bit surprised by the results but it looks like 40 mm was more optimized.
60 mm is giving inferior results with temperature swings on both cold and end slightly greater than that of 40 mm
here is the 60 mm plot:
In order to visualize better, I put both the 40 and 60 mm plots on the same graph, but only kept the temperatures for the heater and cooler surfaces (I removed the temperatures of the regenerator ends for clarity):
There you can clearly see how the temperature swings are slightly worse.
The next step was to start a simulation with a shorter regenerator: 20 mm (everything still unchanged), and without spoiling it, the results will also be worse (quite a bit worse this time) than the 40 mm version.
That means I will probably need to do a 30 mm, maybe a 50 mm version as well.
I also means 40 mm is probably pretty close to the sweet spot in this particularly configuration.
I wish those simulations weren't as time consuming.
I will need to revisit the impact of larger diameters regenerators and also play with the wall thickness as well.
-
stephenz
Re: Stephenz's work
Here's the data with the shorter (20 mm) regenerator.
The performance loss is pretty big.
The performance loss is pretty big.
-
stephenz
Re: Stephenz's work
Here's an update on the Alpha test build I was putting together.
I have tried a few other things and the short version is that I cannot get it to start. You clearly feel it wants to start when the butane torch is applied to it but the main issue I believe is the fact I have to start applying heat when the temperature on the regenerator reaches 200C or the oring will melt.
Issues:
- temperature limited by design - in fairness I didn't think the engine would need such high temperature but it's pretty big...
- I don't have a very good piston seal despite the reduce gapping
- the regenerator housing is conducting heat way too well
Not holding pressure:
the new pistons I made gave me a gapping of 0.01 to 0.02 mm. And with good compressor oil lubrication it spins pretty smoothly while still being able to provide good compression. Turning the flywheel by hand gives me a highest pressure point of about 5 psig but within 2 seconds this drops down to 0 psig. I have checked for leaks by pressurizing the system to 30 psi and it holds pressure well. So it leaks at the piston/cylinder interface, I finished designed a new piston/cylinder design using the same crankcase and crankshaft that has active oil lubrication and oring seals. More on that later.
Regenerator housing issue:
it's brass material transfers heat too well which makes the cold end of the regenerator pretty hot.
This basically mean a lot of heat is transferred thermally from the heater to the cooler rather being transferred to the fluid to create work.
A quick note: I did change the regenerator material to fine stainless steel wool, and to nothing without any difference with the heavy stainless steel blade stack.
There isn't much else I can try here so I am about to finish design of a second test engine. The main thing I want to do is convert this from an Alpha to a Gamma.
Which brings me to the questions I wanted to ask here:
When looking at this compressor, you see "alpha" written all over it. I said "converting to Gamma" above but I see 2 ways of isolating the heater from the cooler, which is one of the main issue I am facing with my current prototype.
Arguably the easiest way to isolate the heater would be piggy back a displacer on top of what used to be the hot piston in the Alpha design. This displacer would be also the regenerator. The entire thing would be covered by a spinned stainless steel cap. With this modification, the hot piston is technically still there and while the displacer is technically able to displace fluid, the hot piston is still changing the internal volume of the engine, hence affecting the pressure of the system while it's moving. I am not sure if this configuration has a name but to me it's an alpha, and not a gamma either.
Anyway, a second way of going about this would be to keep the hot piston but attach a rod in the middle of it and cap the hot piston. Of course you can't close the volume between the top of the piston and the cap otherwise pressure would build up, so that piston would have vent holes communicating with the crank. Now the volume of the engine does not change as the hot piston moves up and down, and of course the rod is connecting to a displacer/rengenerator combo. And this now is technically a gamma configuration. This is obviously more work design wise and arguably with more potential for friction (the rod is another moving part that needs sealing, additional friction) but this gives more flexibility in sizing the displacer where it can have a diameter significantly different (smaller) than that of the hot/cold pistons.
What would you guys recommend?
I have tried a few other things and the short version is that I cannot get it to start. You clearly feel it wants to start when the butane torch is applied to it but the main issue I believe is the fact I have to start applying heat when the temperature on the regenerator reaches 200C or the oring will melt.
Issues:
- temperature limited by design - in fairness I didn't think the engine would need such high temperature but it's pretty big...
- I don't have a very good piston seal despite the reduce gapping
- the regenerator housing is conducting heat way too well
Not holding pressure:
the new pistons I made gave me a gapping of 0.01 to 0.02 mm. And with good compressor oil lubrication it spins pretty smoothly while still being able to provide good compression. Turning the flywheel by hand gives me a highest pressure point of about 5 psig but within 2 seconds this drops down to 0 psig. I have checked for leaks by pressurizing the system to 30 psi and it holds pressure well. So it leaks at the piston/cylinder interface, I finished designed a new piston/cylinder design using the same crankcase and crankshaft that has active oil lubrication and oring seals. More on that later.
Regenerator housing issue:
it's brass material transfers heat too well which makes the cold end of the regenerator pretty hot.
This basically mean a lot of heat is transferred thermally from the heater to the cooler rather being transferred to the fluid to create work.
A quick note: I did change the regenerator material to fine stainless steel wool, and to nothing without any difference with the heavy stainless steel blade stack.
There isn't much else I can try here so I am about to finish design of a second test engine. The main thing I want to do is convert this from an Alpha to a Gamma.
Which brings me to the questions I wanted to ask here:
When looking at this compressor, you see "alpha" written all over it. I said "converting to Gamma" above but I see 2 ways of isolating the heater from the cooler, which is one of the main issue I am facing with my current prototype.
Arguably the easiest way to isolate the heater would be piggy back a displacer on top of what used to be the hot piston in the Alpha design. This displacer would be also the regenerator. The entire thing would be covered by a spinned stainless steel cap. With this modification, the hot piston is technically still there and while the displacer is technically able to displace fluid, the hot piston is still changing the internal volume of the engine, hence affecting the pressure of the system while it's moving. I am not sure if this configuration has a name but to me it's an alpha, and not a gamma either.
Anyway, a second way of going about this would be to keep the hot piston but attach a rod in the middle of it and cap the hot piston. Of course you can't close the volume between the top of the piston and the cap otherwise pressure would build up, so that piston would have vent holes communicating with the crank. Now the volume of the engine does not change as the hot piston moves up and down, and of course the rod is connecting to a displacer/rengenerator combo. And this now is technically a gamma configuration. This is obviously more work design wise and arguably with more potential for friction (the rod is another moving part that needs sealing, additional friction) but this gives more flexibility in sizing the displacer where it can have a diameter significantly different (smaller) than that of the hot/cold pistons.
What would you guys recommend?
Re: Stephenz's work
I meant to respond to this sooner. Maybe go look at this alpha design for some insight. https://www.youtube.com/watch?v=3y5X11MXsRc
I think what you are battling with, besides excessive heat transfer, is the volume of the transfer port in relation to the working volume. Perhaps you could make an insulated insert to reduce volume and heat transfer within the regenerator housing. You would probably be surprised by how small of a port would still be sufficient for low rpm operation. After that you could try to space the cylinder heads up to increase the volume of air in each cylinder.
I think what you are battling with, besides excessive heat transfer, is the volume of the transfer port in relation to the working volume. Perhaps you could make an insulated insert to reduce volume and heat transfer within the regenerator housing. You would probably be surprised by how small of a port would still be sufficient for low rpm operation. After that you could try to space the cylinder heads up to increase the volume of air in each cylinder.
-
stephenz
Re: Stephenz's work
I don't have a whole lot of dead volume but you're absolutely right about the size of the ports needed to transfer the working fluid, they don't need to be that big.
Since my last post, I decided to ignore the temperature of the regenerator and cylinder and just heat up, expecting the orings to melt, etc.
Turns out it got to a point where it almost started (10 or so revolutions) before it locked up. I'm still trying to figure out why, since both the cylinder and the piston were made of stainless steel, which should rule out differential temperature expansion.
After letting it cool off, I disassembled and found nothing abnormal on the hot cylinder. Strangely the cold piston/cylinder is the one that locked up. Both are made of aluminum and without active lubrication, oil eventually disappears.
Quick Update on the new design:
I settled on a new design and started making parts already: crankcase, cold cylinder and piston. I am going back to the idea of o-rings on the cold cylinder/piston since there is no high temperature requirements. I designed in a way that I should be able to fine tune the o-ring size to find the sweet spot when between compression and friction. The big change with this new design is an active lubrication system. I lengthened the piston (no change to diameter or travel) to allow the mid line of the stroke to always be located between the top and bottom of the piston regardless of the crank angle. This allows me to locate 2 holes on each side of the cylinder to circulate compressor oil through a V-groove channel. A small electric pump will circulate the oil, either continuously or on a timer/as needed. Since there are 2 holes (in and out) I can technically run it continuously without having to worry about oil pressure building up and making its way through the orings. As is I am hopeful that the film of oil will remain capped by the upper and lower orings.
The biggest change of the design is the Gamma conversion. Cooler and Heater are now on the same cylinder. The regenerator is annular.
Cooler is liquid cooled copper with internal fins, similarly to what the cooler of the Sunpower engine looks like: https://youtu.be/adLZIDxM8tQ?t=158
The displacer itself is mounted on a 8 mm shaft connected to yoke articulated to the connecting rod pin. So there is no piston/cylinder seal to worry about in the displacer. I am taking care of the displacer shaft seal using a PTFE dynamic seal. I have ordered the seal and would like to get a feel for them with an 8mm shaft before I start machining more parts.
As for the regenerator, I've gathered enough CFD results to be comfortable with its size and geometry. Which brings me to:
Quick Update on the regenerator study:
While I'm technically still running simulations, I have learned a lot of things by doing these alternating pulses.
I've done a number of steady states analysis to evaluate the boundary conditions static pressures needed to achieve the volumetric flow rates matching that 192 cc transfer. As it turns out the mass flow rate going from cold to hot versus those going from hot to cold are not equal. Maybe it sounds absurd as you would think mass is conserved, and it is. However, the transfer is done by displacement of a given physical volume. And because the temperatures on either side of the regenerator are different, the densities are also different and consequently the mass flow rates are different as well.
It took me a bit a time to figure this one out, as it wasn't that intuitive. Aside from this being pretty critical to getting good analysis, I discovered something pretty interesting: the heat transfer between fluid to regenerator is increasingly different than from regenerator to fluid.
As you know I did all my analysis with Helium pressurized to 10 bar.
When the difference in densities between 300K and 600K at 10 bar are actually significant. I put this graph together using ideal gas law.
Those densities differences make the cold blow to be much "stronger" cooling than that the heating caused by the hot blow.
The consequences from this I am not too sure yet.
I am running a simulation currently, which I started with an ideal temperature gradient across the regenerator (by starting with that ideal gradient it greatly reduces the time it takes the system to reach steady state); what I am seeing is the hot end of the regenerator is not able to hold something close to the 600K from the hot blow. The other end of the regenerator is basically at 300K. In other words, the cold blows are gradually winning over the hot blows. Based on the mass of the regenerator I expect I will each steady state after roughly 12-15s of physical time, I am currently at ~3 seconds.
This graphs shows the mass flow rates differences between the cold and hot blows.
And this graph here shows you, how starting from an ideal gradient (hot end of regenerator starts at 600K, cold end starts at 293.2K) the cold end remains pretty much unchanged, while the hot end is gradually dropping.
The second observation (likely for the same reason) is how the heater temperature is also dropping much faster on cold blows than the cooler temperature is climbing on hot blows.
Since my last post, I decided to ignore the temperature of the regenerator and cylinder and just heat up, expecting the orings to melt, etc.
Turns out it got to a point where it almost started (10 or so revolutions) before it locked up. I'm still trying to figure out why, since both the cylinder and the piston were made of stainless steel, which should rule out differential temperature expansion.
After letting it cool off, I disassembled and found nothing abnormal on the hot cylinder. Strangely the cold piston/cylinder is the one that locked up. Both are made of aluminum and without active lubrication, oil eventually disappears.
Quick Update on the new design:
I settled on a new design and started making parts already: crankcase, cold cylinder and piston. I am going back to the idea of o-rings on the cold cylinder/piston since there is no high temperature requirements. I designed in a way that I should be able to fine tune the o-ring size to find the sweet spot when between compression and friction. The big change with this new design is an active lubrication system. I lengthened the piston (no change to diameter or travel) to allow the mid line of the stroke to always be located between the top and bottom of the piston regardless of the crank angle. This allows me to locate 2 holes on each side of the cylinder to circulate compressor oil through a V-groove channel. A small electric pump will circulate the oil, either continuously or on a timer/as needed. Since there are 2 holes (in and out) I can technically run it continuously without having to worry about oil pressure building up and making its way through the orings. As is I am hopeful that the film of oil will remain capped by the upper and lower orings.
The biggest change of the design is the Gamma conversion. Cooler and Heater are now on the same cylinder. The regenerator is annular.
Cooler is liquid cooled copper with internal fins, similarly to what the cooler of the Sunpower engine looks like: https://youtu.be/adLZIDxM8tQ?t=158
The displacer itself is mounted on a 8 mm shaft connected to yoke articulated to the connecting rod pin. So there is no piston/cylinder seal to worry about in the displacer. I am taking care of the displacer shaft seal using a PTFE dynamic seal. I have ordered the seal and would like to get a feel for them with an 8mm shaft before I start machining more parts.
As for the regenerator, I've gathered enough CFD results to be comfortable with its size and geometry. Which brings me to:
Quick Update on the regenerator study:
While I'm technically still running simulations, I have learned a lot of things by doing these alternating pulses.
I've done a number of steady states analysis to evaluate the boundary conditions static pressures needed to achieve the volumetric flow rates matching that 192 cc transfer. As it turns out the mass flow rate going from cold to hot versus those going from hot to cold are not equal. Maybe it sounds absurd as you would think mass is conserved, and it is. However, the transfer is done by displacement of a given physical volume. And because the temperatures on either side of the regenerator are different, the densities are also different and consequently the mass flow rates are different as well.
It took me a bit a time to figure this one out, as it wasn't that intuitive. Aside from this being pretty critical to getting good analysis, I discovered something pretty interesting: the heat transfer between fluid to regenerator is increasingly different than from regenerator to fluid.
As you know I did all my analysis with Helium pressurized to 10 bar.
When the difference in densities between 300K and 600K at 10 bar are actually significant. I put this graph together using ideal gas law.
Those densities differences make the cold blow to be much "stronger" cooling than that the heating caused by the hot blow.
The consequences from this I am not too sure yet.
I am running a simulation currently, which I started with an ideal temperature gradient across the regenerator (by starting with that ideal gradient it greatly reduces the time it takes the system to reach steady state); what I am seeing is the hot end of the regenerator is not able to hold something close to the 600K from the hot blow. The other end of the regenerator is basically at 300K. In other words, the cold blows are gradually winning over the hot blows. Based on the mass of the regenerator I expect I will each steady state after roughly 12-15s of physical time, I am currently at ~3 seconds.
This graphs shows the mass flow rates differences between the cold and hot blows.
And this graph here shows you, how starting from an ideal gradient (hot end of regenerator starts at 600K, cold end starts at 293.2K) the cold end remains pretty much unchanged, while the hot end is gradually dropping.
The second observation (likely for the same reason) is how the heater temperature is also dropping much faster on cold blows than the cooler temperature is climbing on hot blows.
Re: Stephenz's work
Hello ! Some active user with active thread detected.
Also have this china made air compressor as a starter - but making new case is not the hardest thing, compared to other parts, especially hx. But it may allow to make other parts easier and better + better geometry.
My first engine which run, was made with ptfe lip seals. And hard anodised aluminium cylinders.
https://m.youtube.com/watch?v=RcxW9fIGPDo
And 90 degree is not optimal for our lower heat input. We need in a 120-130 degre for Alpha ones. Or go to alphagamma side.
Also have this china made air compressor as a starter - but making new case is not the hardest thing, compared to other parts, especially hx. But it may allow to make other parts easier and better + better geometry.
My first engine which run, was made with ptfe lip seals. And hard anodised aluminium cylinders.
https://m.youtube.com/watch?v=RcxW9fIGPDo
And 90 degree is not optimal for our lower heat input. We need in a 120-130 degre for Alpha ones. Or go to alphagamma side.
Re: Stephenz's work
Sorry if I missed an earlier post, but as this appears to be a "new" theory/design "alpha/gamma?
In your photo on the pressurization thread you have a long copper transfer tube. It leaves the copper cooling jacket of the displacer(?) Cylinder. Then goes over to the cold power cylinder?
Are you still using a standard 90° alpha type offset?, that is, the connecting rods attach to the crank the same as in a typical Alpha?
I'm just trying to get a picture in my mind of how this arrangement functions, or theory of operation.
I think there was a thread on "alpha/gamma" was that yours? Seems like long ago, before the world went mad. Some Interesting topics sometimes get "lost". But one good thing about this forum is old topics are never (almost never anyway) closed, archived or deactivated so they can always be revived. Trying to actually find them can be a challenge though.
In your photo on the pressurization thread you have a long copper transfer tube. It leaves the copper cooling jacket of the displacer(?) Cylinder. Then goes over to the cold power cylinder?
Are you still using a standard 90° alpha type offset?, that is, the connecting rods attach to the crank the same as in a typical Alpha?
I'm just trying to get a picture in my mind of how this arrangement functions, or theory of operation.
I think there was a thread on "alpha/gamma" was that yours? Seems like long ago, before the world went mad. Some Interesting topics sometimes get "lost". But one good thing about this forum is old topics are never (almost never anyway) closed, archived or deactivated so they can always be revived. Trying to actually find them can be a challenge though.
Re: Stephenz's work
Looks like a Harbor Freight compressor conversion?
I think I have the same one if it is. Maybe like 4 of them actually, but three are still working as actual compressors.
I haven't had the crankcase open yet though.
- IMG_5757.JPG (272.08 KiB) Viewed 1074 times
I haven't had the crankcase open yet though.