|
|
Post by Roger on Jan 23, 2021 22:19:55 GMT
What is the lower pressure you are aiming hoping for? Personally I wouldn’t worry about an injector that won’t run below 40psi. In some respects it’s an advantage; we have all done it. Come to a halt with a great fire going put on the injector to make use of that heat and then got distracted or into a natter about something. Next thing you know you have flooded the boiler! However, Speedy boiler is a good size and I often find the pressure drops before you over fill. Seeing the injector shooting water can be a reminder that you have just done something daft! Now with 40 psi and decent fire you can turn that around and no one will know. To be honest, even 50psi would be plenty low enough. I mean, you can't really drive it until there's 60psi on the clock, so it's only to get out of trouble that it would be of much value. However, there's no argument against an injector that runs down to very low pressures. It would appear that I can acheive a clean pick up at higher boiler pressures, even with really hot water as long as there's a slight positive head. The delivery carries on fine as the pressure drops, but with an ever increasing amount of water from the overflow with the setup as it currently is. It does seem to keep feeding though, even when there's a lot coming from the overflow. I'm still not convinced that the overflow valve is big enough and light enough. It does seem better than before though. Trying to get solid data isn't easy, because one parameter affects others. I may revisit some of the cones I've made and run another batch of tests to establish their performance at fixed conditions. Ideally I'd like it to have less overflow over a wider pressure range, say from 60psi to 90psi. It's close to being good enough, but I think I can make it better. |
|
|
|
Post by Roger on Jan 23, 2021 22:21:35 GMT
One of my injectors still has a tap so I can restrict the water on that one. The one without the tap is also good for my youngest son. Now I have them working better, and a steam valve further away from the hot boiler and manifold, that side is easier for him to operate. Just one control and less chance of touching something hot. If my pressure gets down to 40psi, I have to pull over into a siding to get i back together anyway, I'm not a good enough driver to do it on the run most times, esp with passengers. That means if I have any reasonable amount of water I can bring pressure back and then use the injector. If I was filling the boiler at 40psi, so lowering the pressure, I might not have enough steam for the blower to be effective so would be finished anyway with a dead fire. That all makes perfect sense. I tend to think of 40psi operation to just get some water in the boiler in case of emergency. |
|
|
|
Post by Roger on Jan 25, 2021 0:02:25 GMT
It's been an interesting day and a few conclusions reached. One of those is that it looks like this 0.8mm radius on the inlet to the Condensing Cone appears to help the low pressure performance. I'll make any further cones without it first, and then add it to make sure I'm not adding unnecessary complexity.  20210124_123903 by Georgia Montgomery, on Flickr I've gone some way towards the setup Julian suggested ie making a header tank that approximates to the Locomotive's tanks. To cut a long story short, I've had quite a few cones that didn't pick up, and I've blamed the cones unfairly. What's been happening is that if the syphon hasn't started, there can be enough heat in the Injector body to turn the water arriving into steam. That bounces the water back before enough has got through to cool the Injector. With a little positive pressure, this doesn't tent to happen. I'd like to have more suction, so I'll be aiming towards that. However, in the meantime I can avoid unfairly condemning cones when this happens, because once started they work fine.  20210124_210511 by Georgia Montgomery, on Flickr Hopefully the new water valve will arrive tomorrow, this one is too coarse. So this video is a test of a modified Cone 5 which has the following... Combining Cone oveflow holes 1 row of 5 x 0.6mm and1 row of 5 x 0.5mm holes offset radially Delivery overflow holes 1 row of 8 x 0.5mm The middle O-ring support has been machined away and the fillet added to the inlet. I don't think I need to experiment with separate overflows any more, it works just the same with one. Again, I was being misled by the action of the syphon as explained above.  20210124_212113 by Georgia Montgomery, on Flickr Another conclusion I've come to is this. You can add more holes along the body, but that turns out to unfavourably change the characteristics as the pressure drops. If you add holes there, it doesn't cut out abruptly when the pressure drops beyond a certain point like a conventional Injector does. Instead, it progressively dribbles more and more out of the overflow and doesn't drop out completely until much lower pressure. However, it does mean that you would have to manually regulate the water flow so it wasn't wasteful, and that's not satisfactory. So the last of the batch will be as close as I can get to how the original design would be if it was made as three separate cones. There are 8 x 0.75mm holes in both positions, simulating the two gaps. Yeah, I probably should have tried this first, but that wouldn't have been half as much fun and I wouldn't have learned much.  Cone12 by Georgia Montgomery, on Flickr The next batch of cones will be with a slightly modified body as shown below. I've thinned out the O-ring supports at the delivery end to make room for the new Blue O-ring. I've modified the end cap to have a complete ring around the end to press against it. One of the things I didn't like about this design was trying to get a dead fit with a spacer. The problem is that the outlet holes in the sides of both end caps have to line up when they are screwed in tightly. That of itself isn't an issue, but then there has to be a custom spacer which should be a exactly the right thickenss. Too thick and the cap won't screw on to the correct angular position, and too thin risks the regulation gap being compromised. So the solution is to compress an O-ring axially to accommodate small differences in length. I'm only trying to allow for say +/-0.1mm at the most, but this will make life much easier. The force trying to push the cone is only 4 lb. If I make the O-ring size for size in the bore, it will expand and grip the parts too. I don't think there's any way it will move. I haven't shown the new style flat valve here either, but that's probably what I'll fit.  Injector assembly with axial O-ring by Georgia Montgomery, on Flickr |
|
jma1009
Elder Statesman
Posts: 5,922
|
Post by jma1009 on Jan 25, 2021 0:59:30 GMT
As an aside, Roger, your test bed injector body is quite a big lump!
Can you slim it down a bit?
If you have a slower start with the your cones, you need to avoid the injector body getting too hot.
Just a thought!
|
|
|
|
Post by Roger on Jan 25, 2021 8:40:13 GMT
As an aside, Roger, your test bed injector body is quite a big lump! Can you slim it down a bit? If you have a slower start with the your cones, you need to avoid the injector body getting too hot. Just a thought! Hi Julian, That's a very good point, and yes, I certainly could make a slimmer version. Slimmer is definitely better, due to the amount of energy it stores. That's great when it's cold, but definitely not when it's hot. I think this is part of the problem if I don't close the Steam valve tightly and there's a slight wisp of steam heating the body up. The same thing happens if it doesn't pick up right away for whatever reason. In both of those situations, the body is getting roasted by the Steam, and it's turning it into a boiler. That's why the feed water is getting flashed into Steam, recoiling after it's been heated at bit, and then does the same again, but this time it's even hotter. It's never going to start like that. However, having a really good Ejector section ought to make it more able to cope with this extra Steam volume and that may well be why the gap in the Combining Cone is bigger than would appear to be necessary on the face of it. For the moment I'll press on with what I have, but in the knowledge that if it works in this body, a slimmer one ought to work better from this point of view. The scale body is very slender and has very little mass at all. |
|
|
|
Post by delaplume on Jan 25, 2021 11:52:13 GMT
Hi Roger,
yes, injector body temp. is something that has to be considered........full-size ones being placed in as near a cooling airstream as possible eg}---- just inboard of the cab steps for a tender loco........although the position for a tank engine would appear to contradict this at first glance...
Usual trick for the crew would be to leave the feedwater valve slightly cracked open to maintain a small flow of liquid, but only if you were sure where your next watering point was going to be..
maybe you could incorporate those two ideas in your test rig ??
There was a third method sometimes resorted to, which the more seasoned members of this Forum might remember.....and that was the crew to urinate on the errant injector !!...............
Photos of the first two methods being applied by yourself would be welcome,......BUT---------------!!!! LoL....
Have a nice day
Alan
|
|
|
|
Post by Roger on Jan 25, 2021 11:58:31 GMT
Hi Roger, yes, injector body temp. is something that has to be considered........full-size ones being placed in as near a cooling airstream as possible eg}---- just inboard of the cab steps for a tender loco........although the position for a tank engine would appear to contradict this at first glance... Usual trick for the crew would be to leave the feedwater valve slightly cracked open to maintain a small flow of liquid, but only if you were sure where your next watering point was going to be.. maybe you could incorporate those two ideas in your test rig ?? There was a third method sometimes resorted to, which the more seasoned members of this Forum might remember.....and that was the crew to urinate on the errant injector !!............... Photos of the first two methods being applied by yourself would be welcome,......BUT---------------!!!! LoL.... Have a nice day Alan Hi Alan, I imagine the 8x and 10x Injectors only dribble a bit when the water is left open since there's a non-return valve on the overflow. I suspect this is how I'll be running mine. Fortunately, the LH Injector is bolted to a short metal elbow which then bolts to the valve which is screwed to the tank. Not only is the path very short, but the tank ought to act as a heatsink which can only help. I'll try to remember to keep a full bladder though, just in case of emergencies. |
|
|
|
Post by andyhigham on Jan 25, 2021 12:16:28 GMT
So that means that a full size injector will work at 37 Deg C
|
|
|
|
Post by Roger on Jan 25, 2021 14:13:56 GMT
So that means that a full size injector will work at 37 Deg C I can't see why one shouldn't work even in the low 40's Celcius as long as there's an initial flow through the Injector and there's a non-return valve on the overflow. |
|
baldric
E-xcellent poster
Posts: 211
|
Post by baldric on Jan 25, 2021 17:34:29 GMT
I imagine the 8x and 10x Injectors only dribble a bit when the water is left open since there's a non-return valve on the overflow. I suspect this is how I'll be running mine. On full size if you open the water valve, water will flow out of the overflow, I do not have a drawing to hand, but from memory I thought the overflow valve will open if there is water in the body to get out, but stop air getting back in the injector when there is a vacuum, that seems to go against what you are saying here. Have I misunderstood what you are saying or is your overflow non-return valve backwards? Baldric. |
|
|
|
Post by Roger on Jan 25, 2021 18:11:32 GMT
I imagine the 8x and 10x Injectors only dribble a bit when the water is left open since there's a non-return valve on the overflow. I suspect this is how I'll be running mine. On full size if you open the water valve, water will flow out of the overflow, I do not have a drawing to hand, but from memory I thought the overflow valve will open if there is water in the body to get out, but stop air getting back in the injector when there is a vacuum, that seems to go against what you are saying here. Have I misunderstood what you are saying or is your overflow non-return valve backwards? Baldric. Hi Baldric, I've got the Works Drawings of the overflow on the 8X and 10X Injectors, and it's clearly shown with a spring that forces the valve upwards against the seat so seal air from going back up into the injector while still allowing water to escape downwards. The opening is very small, there's only a 3/8"gap when it's fully open. There's a chain attached to the valve to enable it to be opened so that the boiler can be back fed to fill it with water. If water comes out of the overflow when the water is on, it must either be leaking past the valve or there's enough head to displace the valve. |
|
baldric
E-xcellent poster
Posts: 211
|
Post by baldric on Jan 25, 2021 18:24:40 GMT
Glad I misunderstood your statement about only a bit coming out when valve is open, it doesn't take a lot to open the valve, it is easy to empty a tank through an injector, even a Pannier with only the injector on the bottom of the tank.
It is interesting to see your progress with your tests and how you are going about it.
Baldric
|
|
|
|
Post by Roger on Jan 25, 2021 22:34:27 GMT
This is the cone that's as true to the dimensions I gleaned from Bob Bramson's book, and it's proved to be a bit disappointing. It certainly creates a much stronger suction than any of the ones I've made so far, but the downside is that it isn't as good as the Cone 5 I modified and retested. It doesn't create a really clean overflow as the pressure begins to drop. This might be because of the suction being so good that it overwhelms the regulation gap with too much water. This design also has a very large overflow gap at the delivery end. So rather than this being the complete answer, it will take further iterations to home in on the best performance. I added the large fluted entry, and that didn't make any appreciable difference.  20210125_165351 by Georgia Montgomery, on Flickr Anyway, I've used all 12 of the cones I'd made, so it's time to make some more. Here are some of the 3D printed covers that protect the fragile tools  20210125_130527 by Georgia Montgomery, on Flickr This time, all I had to do was clock up the tooling location with the quill extended 50mm and the knee 1mm from the top of its travel as recorded in my setup notes. I keep notes of everything because my memory can't be trusted.  20210125_131457 by Georgia Montgomery, on Flickr so it's happily churning out a load of these while I play with the boiler.  20210125_220947 by Georgia Montgomery, on Flickr I've actually made seven now, and another is on the way. It's certainly been worth grappling with Fusion 360 and the Post Processor to make this easy. I finished the first one on the lathe, and the second one got put into the chuck the wrong way round, so that has the bar end mess instead of a nice clean face. Still, it's good enough for a test. These are the new modified design that will fit the scale body using an axial O-ring for the end load.  20210125_220916 by Georgia Montgomery, on Flickr |
|
|
|
Post by Roger on Jan 26, 2021 13:54:19 GMT
While I'm mass producing the new batch of cone bodies, 12 made so far, I've been chewing over the direction the experiments need to take. This one is a development of Cone 12, aimed at discovering whether the dribbling as the pressure drops is due to the size of the Delivery overflow. the Combining overflow remains the same.  Cone14 by Georgia Montgomery, on Flickr This is a development of the modified Cone 5 which is the best so far. Cone 5 has 8 x 0.5mm holes at the Delivery overflow, so I've reduced those to just 4 x 0.5mm holes. Changing from 8 to 4 holes doesn't reduce the area that much because they overlap considerably near the axis when there are 8 holes. However, having 4 holes does leave a narrow portion where the flow can pass straight across, and either side of that the gap is effectively narrower that 0.5mm Quite how the flow reacts to going over a series of holes is beyond my understanding. Maybe one day someone with a decent FEA CAD package will be able to model the flows, but it's a really difficult problem because you've got Steam that's condensing and mixing. Doubtless it's possible, but that's way beyond me.  Cone13 by Georgia Montgomery, on Flickr Here are 18 new style cones ready for the next tranche of tests.  20210126_135051 by Georgia Montgomery, on Flickr |
|
|
|
Post by Roger on Jan 26, 2021 20:30:17 GMT
This is Cone 14 which dribbles... Cone14 by Georgia Montgomery, on Flickr ... So I've reduced the Combining hole diameters from 8 x 0.75mm with 10 x 0.6mm to still make them overlap in Cone 15. The idea is to simulate a straight gap between cones while reducing the volume of the overflow and also the gap it needs to jump.  Cone15 by Georgia Montgomery, on Flickr This is Cone 13 which is as good as Cone 5, either of which are probably acceptable. Cone13 by Georgia Montgomery, on Flickr Here I've kept the same diameters and rows for the Combining holes, but I've reduce their number from 5 holes to 4 in each row. There aim is to lower the pressure at which dribbling starts to below 60psi. I need to see if this is better overall or worse. The danger is that I'll reduce the volume too much and then it won't draw enough to overcome the issue when the body is hotl  Cone16 by Georgia Montgomery, on Flickr |
|
|
|
Post by Roger on Jan 26, 2021 22:51:30 GMT
I've just done a quick test on Cone 15, ie the single row design, and that's much better than it was with the larger holes in the combining cone. It's dry from 90psi to 60psi with 21C water, and it can be made to run dry with a bit of adjustment of the water down below 40psi.
So this is a huge improvement.
I've also taken to roasting the Injector with steam for ten seconds before turning on the water as an additional test. This passed the test by picking up straight away.
It's too late to do tests on Cone 16 tonight, but this is great progress in as much as the designs are using fewer 0.5mm and 0.6mm holes which are easy to drill. Hopefully I can boil this down to something that's really simple.
|
|
jma1009
Elder Statesman
Posts: 5,922
|
Post by jma1009 on Jan 26, 2021 23:23:58 GMT
Hi Roger,
On the subject of dribbling at lower pressures, the amount of water that gets condensed, and also the delivery rate, falls.
Don't avoid the bigger picture that includes the steam cone and the annular gap!
Cheers,
Julian
|
|
|
|
Post by Roger on Jan 26, 2021 23:40:47 GMT
Hi Roger, On the subject of dribbling at lower pressures, the amount of water that gets condensed, and also the delivery rate, falls. Don't avoid the bigger picture that includes the steam cone and the annular gap! Cheers, Julian Hi Julian, I've got the Steam Cone throat taken from the graph in D.A.G Brown's book for the starting point, ie it's calculated to work up to 90PSI. I've also made a Steam Cone with 0.05mm (2 thou) bigger which might prove to be better for the lower pressure end as long as it's not too much for the hight end. In correspondance which Ed had with Bob Bramson, he states that he sets the Regulation Gap to just make the overflow dry at the maximum working pressure. I've played around with this extensively, and all of the cones I've made work at about 1/4 to 1/3 of a turn back from the Steam Cone being hard against the face of the Condensing Cone. With 0.75mm pitch, that resolves to 0.188mm to 0.25mm. There's not much room for adjustment, you can see it pick up sharply when the gap reaches that minimum, and you need to go a little more to make it stable and for the pressure to rise to a maximum. Any more gap than that results in a gradually reducing pressure with no benefit. Leaving it larger than necessary means it passes too much water at lower pressures. Remember I'm not using Annular Regulation, I'm using End Regulation where there's a flat part on both cones that controls the flow. This makes it really easy to know what the gap is and to experiment with it while it's running. So far it looks like it ought to be possible to interchange cones from one Injector to another, it's a very repeatable setup. Anyway, there's still a lot of tests to perform and other variations to try to subtly home in on the best performance. So far I'm happily surprised that it's working this well after a slow start. |
|
|
|
Post by Roger on Jan 27, 2021 14:57:53 GMT
I've performed a lot of tests on the last two cones today, mostly at temperatures 30-35C and they perform pretty much the same as far as I can tell. I've also added the 0.8mm radius to the entry of the Combining Cone, and frankly I'm now now convinced it makes a difference. Anyway, I've decided to make another couple of cones that are subtly different to those to see if any further improvements can be made. Cone 18 now has 6 x 0.6 holes instead of 10 so that they don't overlap any more. That might mean there isn't enough area, but it's worth a try.  Cone18 by Georgia Montgomery, on Flickr Cone 17 now has 0.5mm holes in the first row instead of 0.6mm. This simplifies making it because all of the holes are now the same size. The reduction in area isn't huge, but you can't tell how important that is, you just have to try it. It's not just about the size of the holes, it's the way the flow passes over and between them. Obviously a smaller diameter first row ought to interfere less with the following staggered row, but nothing it certain in the world of Injectors!  Cone17 by Georgia Montgomery, on Flickr Anyway, I've come to the conclusion that many different patterns work, although some are definitely better than others. There's not a lot to choose between quite a few of them. I can see having to make some more accurate comparative tests with better control of the water temperature and starting conditions to be able to tell them apart. It's almost impossible to get a dry overflow at elevated water temperatures of 30C plus. It's also easy to kid yourself that something works better than it does if you use water that's 15C. I thinks stress testing at 30C is definitely worth doing since it really does need to feed when it's hot, even if it's not with a clean overflow. Another thing I've noticed is that the characteristics of the Injector are not the same when feeding the boiler compared to when using the balanced clack valve. I think this has to do with the fact that there's no mass to damp any fluctuations in the flow when using the balanced clack valve, there's just a ball and a light plunger. It's definitely smoother and more likely to pick up and keep going when feeding to the boiler. |
|
|
|
Post by Roger on Jan 27, 2021 21:40:28 GMT
Ok, I've re-tested some of the more promising Cones, and Cone 13 seems to be the most resilient when it comes to starting in difficult conditions. I guess this is due to the generous overflow area, achieved with the largest holes being 0.6mm The double row certainly seems to help with controlling the disruption to the flow, while giving a large area. Just reducing that first row of holes to 0.5mm makes it much worse. I think I need to spend some time trying this one over a wide range of conditions to see if this one is satisfactory. I can be really horrible to it, and it still picks up. At 90psi and water at 23C, I can turn the water off and let the Steam roast the body, and it still picks up again when I turn the water back on. Hopefully this will get me close to where I need to be, and then I can try it in the scale body and see if it still works. Cone13 by Georgia Montgomery, on Flickr |
|
