|
Post by joanlluch on Sept 14, 2016 19:16:31 GMT
|
|
|
Post by joanlluch on Sept 14, 2016 19:34:36 GMT
The cylinder blocks are separated 3 mm from the frames. Two isolating gaskets with a metal sheet in the middle are placed between each cylinder block and the frame. Junta aïllament cilindre. by joan lluch, on Flickr These are pictures of the way I have cut these gaskets. I know there are possibly better ways, but this worked for me. First, I made the holes by sandwiching the gaskets between wooden panels. Juntes aïllament cilindre. by joan lluch, on Flickr Second, I sandwiched the gaskets between the metal sheets mentioned above that have previously laser cut to size, and I have removed the excess material with a cutter. Juntes aïllament cilindre. by joan lluch, on Flickr Juntes aïllament cilindre. by joan lluch, on Flickr And this is how the gaskets look after they have been finished. Juntes aïllament cilindre. by joan lluch, on Flickr This material is similar to klingerit, but it's specifically chosen to stand high temperatures while keeping a relatively low thermal conductivity.
|
|
|
Post by Roger on Sept 15, 2016 6:03:09 GMT
You've made a superb job of that Joan, I'm very impressed. I'm pleased you're making good use of your mill, it's giving you the freedom to modify parts and make a few pieces if you need to. Well done indeed!
|
|
|
Post by joanlluch on Sept 15, 2016 7:07:44 GMT
You've made a superb job of that Joan, I'm very impressed. I'm pleased you're making good use of your mill, it's giving you the freedom to modify parts and make a few pieces if you need to. Well done indeed! Thanks for your words. I only attempt to mill parts that are easy to make manually, or do not require complicated settings. Otherwise I still rely on the CNC service for more complicated things. You know that I wouldn't be able to machine even these simple parts if it wasn't for the help and advice you gave me in private when I bought the mill. So I want to acknowledge it now in public and to say thank you for that too.
|
|
|
Post by Roger on Sept 15, 2016 9:27:32 GMT
I'm just delighted that you've gained as much confidence, you seriously underestimate your ability to make things yourself. I do understand that it's not really the thing that gives you the thrill in creating something, but I'm sure you get a ripple of excitement when you achieve something like that. Long may it continue!
|
|
|
Post by joanlluch on Sept 30, 2016 15:50:07 GMT
While I am waiting for some parts to arrive from the CNC service I investigated the way to simulate the boiler pressure on the computer. This is a photo and I also recorded a video Caldera, prova sense reforços by joan lluch, on Flickr The boiler is not yet fully drawn. Among other things, boiler stays remain to be drawn. In the simulation, the boiler is subjected to a pressure of 10 bar for this test. The red areas are the ones subjected to a mechanical stress that would break the metal. So yes, the boiler would break scandalously. But that's to be expected because the stays are not yet there and exterior wall thickness is just 3mm. So far, that was just an exercise to figure out how to actually run simulations on the computer. Simulation should allow me to optimally chose wall thickness and place stays where they are more effective to make sure that the boiler will support the required pressure with a generous safety coefficient. Forgot to mention that the deformation shown in the video is displayed at 10x the deformation that would actually happen.
|
|
|
Post by joanlluch on Sept 30, 2016 16:19:54 GMT
And the same for the firebox.
Of course, it would also break for the reasons exposed above.
|
|
|
Post by Roger on Sept 30, 2016 16:27:18 GMT
That's really interesting, and a superb example of how useful these tools can be in visualising what's going on as well as giving measurements. You have superb software, I'd love to have that, but I fear it's beyond my budget.
|
|
|
Post by joanlluch on Sept 30, 2016 18:58:12 GMT
That's really interesting, and a superb example of how useful these tools can be in visualising what's going on as well as giving measurements. You have superb software, I'd love to have that, but I fear it's beyond my budget. I guess I only know to use less than 2% of its features.
|
|
|
Post by joanlluch on Oct 1, 2016 9:05:42 GMT
Sorry guys, but I'm new to this and I can't help to get some excitation from this simulation stuff, so this is another video showing the boiler from the bottom.
This video shows the actual deformation of the metal sheets, that's why no exaggerated bulges develop on the video. As before, red colour indicates a load above the elastic limit of the metal. Pressure is still 10 bar and no stays yet.
It's quite interesting to see that the weaker parts, in other words the ones receiving first the greater loads, are the edges of the boiler. This is expected based on existing experience in boiler making. Still, I expect this to change significantly when stays are incorporated. The tubular section of the boiler does not receive any dangerous load, which of course was totally expected as well.
|
|
|
Post by Roger on Oct 1, 2016 12:02:42 GMT
Hi Joan, I'm curious to know why the stress is initially larger around the edges? Is it because the unsupported flat section can just bend, whereas the places nearer to the edges are rigidly supported? I suppose the bending of the flat panels also put a bending moment onto the corner sections too.
|
|
|
Post by joanlluch on Oct 1, 2016 15:20:10 GMT
Hi Roger, In fact max stress develops simultaneously on the corners and on the centre of the larger unsupported flat sections.
I see it the following way:
Imagine two flanges "glued" together (or welded) at 90 degress that are exactly as long as to reach the centre of the flat sections in the boiler. Then try to open these flanges (such as unfolding a book) so that the 90º angle wants to increase. The max stress will happen just at the angle.
Now imagine 4 beams forming a square and apply a uniform load from the inside out (i.e a pressure load). Since they form a square together and loads are equilibrated, every single beam will behave as fixed (or rigidly supported) at both ends. So the bending moment will be maximised at the the centre and at the ends of the beams.
My understanding is that the software does not make load adjustments based on deformation. It will display theoretical deformations based on the material elastic modulus but that's all. This is in fact a conservative measure because it basically means that the boiler may actually support more pressure than computed if it is allowed to bulge. You mentioned in your thread that you have seen boilers exhibiting bulges at the top of the backheat, so you are taking measures to prevent it. I found that comment interesting because these 'bulged' boilers have not exploded and yet they have obviously surpassed the elastic limit of copper. So this means that these boilers would have NOT passed the computer simulation to begin with.
|
|
|
Post by RGR 60130 on Oct 1, 2016 15:31:22 GMT
I'd be interested to see what the 'picture' looked like with radiused corners between the blackhead and wrapper instead of square corners.
Reg
|
|
|
Post by Roger on Oct 1, 2016 15:51:04 GMT
Hi Joan, Thanks for that explanation, it does make sense when you look at it in that way. It's no different to any 'stress raiser', it's just that I hadn't really thought of it as such. It's no different to leaving a sharp shoulder on a change in shaft diameter. What Reg says is interesting too, because a shallow curve is going to concentrate the stress less. I can see why you wouldn't want to go down that route though if you can achieve a low enough stress with a square joint. All very interesting.
|
|
|
Post by joanlluch on Oct 1, 2016 16:19:27 GMT
Hi Roger and Reg, Yes, certainly rounded corners would decrease stress, and the greater the radius the better.
My boiler will be welded stainless steel, which is stronger than copper on the SAME circumstances. I will start from laser cut sheets and will have a certified welder joining it all together. That's a simpler approach than a silver soldered copper boiler, and it is also the current way miniature boilers are made locally.
Still, I want to investigate the possibility to make a TIG welded stainless boiler with flanged rounded plates just similar to the standard practice for copper ones. There's a man in a nearby club (Terrassa) that advocates just that and he has successfully made several of them. He manages to properly bend stainless steel without breaking it and ends with a boiler with rounded corners like a copper one.
|
|
|
Post by joanlluch on Oct 2, 2016 19:30:46 GMT
Ok, I have been struggling all the afternoon at attempting to add stays to my boiler model.
The problem is that my computer seems to be already at the limit of the required resources to perform this simulation. It is not that the computer isn't good, it has plenty of free hard disk and quad core processor with 8Mb RAM. But the simulation appears to suck all the available RAM and leaves the computer in a unusable state before the end of the computation. I tried to augment the mesh size but this does not work because I guess the stays are already too thin for the mesh to property catch them and the computation fails. In fact, I had to use a smaller mesh for the simulation to give a meaningful result which only made things worse. So I seem to be stuck with this unless I upgrade my computer.
In any case, I managed to get some results after waiting a very long time for the computation to finish. To make this possible I used stays of 8mm diameter instead of the intended 6mm because this allowed me to use a slightly coarser mesh size so that the computer is able to deal with.
So this is the video:
Only the central row of stays is activated/working at the boiler sides, and no stays yet at the back. The improvement is significant.
It's interesting to note that only by adding this central row of stays, the stress on the throatplate edges has gone just only slightly above the elastic limit, as opposed to the previous situation where stress was much greater. It is to be expected that once the missing stays are activated, load will become fully tolerable.
It's also worth to mention that there is some stress developing just around the stays. This suggests me that it would be better to have say 8mm stays instead of the originally intended 6mm ones just to provide a greater support area to the flanges, regardless of what is actually needed for the stays themselves.
This may be the last simulation I am able to perform with this computer for the reasons mentioned above, but it has already given me a lot of information on the mechanical performance of the boiler.
|
|
|
Post by miketaylor on Oct 2, 2016 19:38:36 GMT
Joan,
Couldn't you get rid of most of the boiler barrel??
This doesn't really have any effect on the analysis of the problem areas and must be sucking up quite a lot of computing power for no good purpose.
Mike
|
|
|
Post by Roger on Oct 2, 2016 19:47:17 GMT
Thinking along the lines that Mike is proposing, could you cut the model in half along the centre line and just put a flat plate on the join so it's a closed volume?
|
|
|
Post by joanlluch on Oct 2, 2016 19:49:05 GMT
Hi Mike, I think this will not make much of a difference. The problem is the required mesh size around the areas where the stays are, which can not be augmented. If you look carefully at the vid or previous photos, you will realise that the mesh size of the barrel is already much coarser than the firebox and rest of the boiler. This is made on purpose to reduce computation resources and yet produce some meaningful result. Also, I observed that the barrel itself plays an important role at supporting the throatplate and lateral flanges. In fact if you look at the model you'll see that the rear part of the barrel actually supports some stresses that are directly transferred from the laterals of the boiler. So I think the barrel must be there for a more realistic simulation of the boiler.
|
|
|
Post by miketaylor on Oct 2, 2016 19:53:58 GMT
Joan,
Continued,
I don't understand what you mean when you say only the middle row of stays are functioning. Looking at the stress colours it is clear that all three rows are supporting the side plates.
When you refer to steess around the sr=tays I presume you mean the slight trace of red around each stay. I suspect that this is not importnat. Your model will not have included the effects of the welded joint at each stay where it attaches to the plates. The weld metal will form a fillet at each location and that should take care of the apparent stress concentration.
You could check it by setting up a dummy model with a couple of stays and modelling the fillet to see what effect it had.
Mike
|
|