|
Post by finiteparts on Jan 5, 2021 20:33:24 GMT -5
Hi Chris Yes and no :-( Half way down page 3 of the Paper ................ " The rotation of the outer surface of the floating bush bearing acts as an uncentered squeeze-film damper" I provided the Link to the Paper mainly because I considered it provided generalised information about turbo rotor dynamics such as the energy distribution , modes , whirl etc , rather than about SFDs. I acquired a copy of Shaw and Macks classical lubrication textbook on Analysis and Lubrication of Bearings many years ago when I was having my problems , so I'm aware of FRB requirements and used the info in some of my design/construction work at that time. LOL....thats why I ended up going back to "brass" and standard turbo components to negate the need to workout bearings , too many other problems arise when doing a build. Yep , keep your contributions coming , hopefully we'll get Patty sorted and can soon see/hear his engine in full flight :-) Cheers John
John,
I can't argue with that, given they are towering figures in rotordynamics. I am also totally in agreement that the paper was an excellent example to show the energy breakdown in the system.
I hadn't seen Shaw and Mack's book, so I looked it up and I was quite surprised by the detail with which they covered the floating ring bearing. I guess for some reason, I thought that the majority of that work was more recent. Thanks for sharing that source.
- Chris
|
|
|
Post by madpatty on Jan 5, 2021 20:46:27 GMT -5
Hi Patty Have you considered simply fitting a couple of bigger bearings ?? Cheers John Hi Racket. Yes. Bigger bearings with dampers, so that anything and everything is taken care of. Cheers.
|
|
|
Post by finiteparts on Jan 5, 2021 22:11:53 GMT -5
Patty,
Are you still trying to use o-rings as the means to soften the bearing stiffness? Those pictures made it look like the o-rings had taken a set. If so, you might want to go to Parker's site, they have a great engineering resource for o-rings that gives really good guidance on setting the compression and gland depth.
I am working to design a squirrel cage mount for the bearing and SFD in my engine, as the means to soften up the system. I am trying to move the first rigid body mode down and reduce the amount of elastic energy that goes into the shaft, which I can't do anything with. If we can get the energy transferred into the bearing/damper (as KE), then I can design the damper to do something with that. As stated previously, if we can move the mode down to a lower rpm, then we have to deal with lower overall system energy. Of course, we will likely drag the first bending mode down well into the operating range, but with the soft mounts, we will be putting less bending strain into the shaft (and thus less elastic energy into the system). This approach is working well for my me in my model to get the bearing loads reduced to acceptable levels.
Here is my current model...not as nice as Dyrobes, but it is free and it forces you to learn the actual background theory and mathematics. Now of course it goes without saying that it is a linear model and thus there are no cross-coupling effects included.
Here is the Campbell diagram... you can see that the first rigid body mode, 1 per rev crossing, is around 14.5 krpm with the first bending mode occurring around 37.3 krpm...then we are clear. The backward whirl of the second bending mode can be seen slightly at around 74 krpm, but that will not drive the rotor to instability, so we can essentially ignore that one...
Here are the mode shapes at 75 krpm...(obviously, we are not in resonance with any of these at this speed)
Finally, here is the bearing deflection over the speed sweep from 0 to 80 krpm (in Hz, not RPM)
The first rigid body mode crossing drives the bearing forces to around 7 lbf, and the first bending mode is around 23 lbf and then walks back down to around 9 lbf. This is still a work in progress, so I may put up more later when I get the design more firmed out.
Also, I see that Dyrobes can do all the calcs for the SFD, are you simulating them in your models?
Good luck!
Chris
|
|
|
Post by madpatty on Jan 5, 2021 23:14:22 GMT -5
Patty,
Are you still trying to use o-rings as the means to soften the bearing stiffness? Those pictures made it look like the o-rings had taken a set. If so, you might want to go to Parker's site, they have a great engineering resource for o-rings that gives really good guidance on setting the compression and gland depth.
I am working to design a squirrel cage mount for the bearing and SFD in my engine, as the means to soften up the system. I am trying to move the first rigid body mode down and reduce the amount of elastic energy that goes into the shaft, which I can't do anything with. If we can get the energy transferred into the bearing/damper (as KE), then I can design the damper to do something with that. As stated previously, if we can move the mode down to a lower rpm, then we have to deal with lower overall system energy. Of course, we will likely drag the first bending mode down well into the operating range, but with the soft mounts, we will be putting less bending strain into the shaft (and thus less elastic energy into the system). This approach is working well for my me in my model to get the bearing loads reduced to acceptable levels.
Here is my current model...not as nice as Dyrobes, but it is free and it forces you to learn the actual background theory and mathematics. Now of course it goes without saying that it is a linear model and thus there are no cross-coupling effects included.
Here is the Campbell diagram... you can see that the first rigid body mode, 1 per rev crossing, is around 14.5 krpm with the first bending mode occurring around 37.3 krpm...then we are clear. The backward whirl of the second bending mode can be seen slightly at around 74 krpm, but that will not drive the rotor to instability, so we can essentially ignore that one...
Here are the mode shapes at 75 krpm...(obviously, we are not in resonance with any of these at this speed)
Finally, here is the bearing deflection over the speed sweep from 0 to 80 krpm (in Hz, not RPM)
The first rigid body mode crossing drives the bearing forces to around 7 lbf, and the first bending mode is around 23 lbf and then walks back down to around 9 lbf. This is still a work in progress, so I may put up more later when I get the design more firmed out.
Also, I see that Dyrobes can do all the calcs for the SFD, are you simulating them in your models?
Good luck!
Chris
Hi Chris. Thanks for sharing your analyses. I am sort of confused at this point on whether I should use end seals(O-rings) on the damper lands or just use a leaking end damper(like in a BBT). If you'd have any resource where I can get some insights on benefits of one over the other then please let me know. I know the pressure profile will be different for both but how much it effects damping is the question. I am willing to drop the O-rings anytime just to reduce the complexity. Is it the elastic energy that kills the bearings or the KE? Also I see your bearings are experiencing very little force at all modes. Is this because of the soft mounts you are using? I haven't yet tried dyrobes for any damper calculations but was basing my rough hand calculation from a NASA paper by Dr. Gunter. Really straightforward approach to design a damper. ntrs.nasa.gov/api/citations/19750007925/downloads/19750007925.pdf?attachment=trueRegards.
|
|
|
Post by wannabebuilderuk on Jan 6, 2021 15:36:54 GMT -5
|
|
|
Post by racket on Jan 6, 2021 19:44:38 GMT -5
Hi Patty Just out of curiosity I've been looking at turbo ball bearing cartridges , the GT55 has a 16 mm shaft and a 44.1mm OD cartridge , similar to the one at the bottom of this page otsturbo.com/products/ceramic-ball-bearing-cartridge/ , now the turbine wheel on the GT55 is kinda small compared to our wheels at only ~111/102 mm , so undoubtedly considerably lighter and less able to generate the loads you are encountering ............there is a very good chance your bearings are a tad marginal , especially at the turb end Cheers John
|
|
|
Post by finiteparts on Jan 6, 2021 21:17:05 GMT -5
Hi Patty,
You need to determine what the brg system stiffness is and from that you can find the optimal stiffness.
It is neither the KE or PE (elasticity) particularly that kills the bearings...it is the total energy of the system and how that energy is dissipated to ground or it's surroundings. As I said before, when the energy is stored in the shaft as elasticity, there is nothing that we can do to help dissipate it out of the system. But, if we can soften up the bearings, then we get more movement (KE) and since dampers are just devices that oppose this movement, now we can do something to dissipate this energy as heat out of the system. So you can see the trade...if we have a stiff system, we can control the rotor eccentricity near the bearings, but this comes at the cost of increased bearing loads (due to high amplification factors). If we soften up the system, we can lower amplification factors and thus lower bearing forces, but this may come at the cost of larger centerline eccentricity.
Two additional points....first, imbalance loads go up with the square of the rotor speed, so getting the modes down to lower rpms, means that when we are moving through the modes, we are doing so at regions of lower energy. Secondly, at lower speeds, the required amount of damping is also lower, so it can help if you don't have a lot of length to stuff a damper in there.
Since you have access to Dyrobes, I would encourage you to play around with the bearing stiffness and see where you can "tune" your systems frequency response.
Good luck!
Chris
|
|
|
Post by madpatty on Jan 6, 2021 23:44:07 GMT -5
Hi Patty Just out of curiosity I've been looking at turbo ball bearing cartridges , the GT55 has a 16 mm shaft and a 44.1mm OD cartridge , similar to the one at the bottom of this page otsturbo.com/products/ceramic-ball-bearing-cartridge/ , now the turbine wheel on the GT55 is kinda small compared to our wheels at only ~111/102 mm , so undoubtedly considerably lighter and less able to generate the loads you are encountering ............there is a very good chance your bearings are a tad marginal , especially at the turb end Cheers John Hi Racket. I am also inclined towards the bearings being marginal. BUT the compressor bearings is what I seems is critical than the turbine bearings. I also intuitively used to think heavier the turbine, more the load on the turbine side bearing which is true up to an extent, but it doesn't explain the entire picture, if you don't take unbalance response and critical speeds into the account. That's what I have been seeing in my testing and my analysis. Here are a few important points that I observed- - Adding a sleeve to increase the effective shaft thickness does decrease the forces acting on turbine bearing BUT it doesn't have much significant effect on the compressor bearing. (Adding a 26mm OD sleeve reduced the turbine bearing force from 9500N to 7900N BUT compressor side bearing forces from 4900N to 4500N)
- Also reducing the unbalance on the compressor side doesn't have any significant effect on the compressor side bearing forces
- Changing the compressor bearing overhang also doesn't have very significant effect on the compressor side bearing forces.
Take into account the unbalance for this is 90g-mm on turbine side and 30g-mm on compressor side so compressor already has 3 times less unbalance than turbine.
I think what's happening is, since in our rotors turbine is much heavier than the compressor, the COM is located very close to the turbine bearing and in conical mode where my bearing failures were concentrated the rotor wants to rotate about that COM(center of mass) thus turbine bearing intrinsically has less motion than the compressor motion (Much like a see-saw with heavier one side and fulcrum also shifted to that heavier side)It's nicely explained in this paper as well-dyrobes.com/wp-content/uploads/2019/02/failure_analysis_of_2_l_engine_turbochargers_gunter-ver2_linked.pdfI am still collecting more data from testing and analysis. BUT in my last multiple tests turbine bearing has survived perfectly and compressor bearings has seen quite some damage. Regards.
|
|
|
Post by madpatty on Jan 9, 2021 13:04:28 GMT -5
Hi Patty,
You need to determine what the brg system stiffness is and from that you cna find the optimal stiffness.
It is neither the KE or PE (elasticity) particularly that kills the bearings...it is the total energy of the system and how that energy is dissipated to ground or it's surroundings. As I said before, when the energy is stored in the shaft as elasticity, there is nothing that we can do to help dissipate it out of the system. But, if we can soften up the bearings, then we get more movement (KE) and since dampers are just devices that oppose this movement, now we can do something to dissipate this energy as heat out of the system. So you can see the trade...if we have a stiff system, we can control the rotor eccentricity near the bearings, but this comes at the cost of increased bearing loads (due to high amplification factors). If we soften up the system, we can lower amplification factors and thus lower bearing forces, but this may come at the cost of larger centerline eccentricity.
Two additional points....first, imbalance loads go up with the square of the rotor speed, so getting the modes down to lower rpms, means that when we are moving through the modes, we are doing so at regions of lower energy. Secondly, at lower speeds, the required amount of damping is also lower, so it can help if you don't have a lot of length to stuff a damper in there.
Since you have access to Dyrobes, I would encourage you to play around with the bearing stiffness and see where you can "tune" your systems frequency response.
Good luck!
Chris
Hi Chris. How do you determine the bearing system stiffness? One idea is to use the experimental data of at what rpm my bearings are failing. A critical map analysis can be done and rpm vs stiffness can be plotted with critical speed mode lines and then by looking at which stiffness aligns with my experimental critical speed gives me the stiffness. Other question is what control do we have on bearing stiffness. I have assumed 4E6 N/m for my bearings. How can I make them more stiff (at-least 1 order of magnitude more so that critical speeds are above my running speed) when they already are located in a stainless shaft tunnel. Regards.
|
|
|
Post by finiteparts on Jan 9, 2021 18:38:04 GMT -5
Hi Patty,
You need to determine what the brg system stiffness is and from that you cna find the optimal stiffness.
It is neither the KE or PE (elasticity) particularly that kills the bearings...it is the total energy of the system and how that energy is dissipated to ground or it's surroundings. As I said before, when the energy is stored in the shaft as elasticity, there is nothing that we can do to help dissipate it out of the system. But, if we can soften up the bearings, then we get more movement (KE) and since dampers are just devices that oppose this movement, now we can do something to dissipate this energy as heat out of the system. So you can see the trade...if we have a stiff system, we can control the rotor eccentricity near the bearings, but this comes at the cost of increased bearing loads (due to high amplification factors). If we soften up the system, we can lower amplification factors and thus lower bearing forces, but this may come at the cost of larger centerline eccentricity.
Two additional points....first, imbalance loads go up with the square of the rotor speed, so getting the modes down to lower rpms, means that when we are moving through the modes, we are doing so at regions of lower energy. Secondly, at lower speeds, the required amount of damping is also lower, so it can help if you don't have a lot of length to stuff a damper in there.
Since you have access to Dyrobes, I would encourage you to play around with the bearing stiffness and see where you can "tune" your systems frequency response.
Good luck!
Chris
Hi Chris. How do you determine the bearing system stiffness? One idea is to use the experimental data of at what rpm my bearings are failing. A critical map analysis can be done and rpm vs stiffness can be plotted with critical speed mode lines and then by looking at which stiffness aligns with my experimental critical speed gives me the stiffness. Other question is what control do we have on bearing stiffness. I have assumed 4E6 N/m for my bearings. How can I make them more stiff (at-least 1 order of magnitude more so that critical speeds are above my running speed) when they already are located in a stainless shaft tunnel. Regards.
Hi Patty,
Ah, you have hit the big question! How do we get the bearing's radial stiffness capabilities if the manufacturers do not provide them???
Generally, I have been just assuming that the angular contact (or in your case, deep groove) bearings are just significantly stiffer than the squirrel cage that I am designing...the flexible squirrel cage will dominate the brg system stiffness value. I did some digging and I found a nifty formula to approximate the bearing stiffness given a few basic parameters. Now of course, with such limited data, this will be an order of magnitude sort of calculation and will not accurately represent the actual bearing stiffness...but, for our purposes, it should be a good enough value to point us in the right direction.
This was found in my copy of Hguyen-Schafer's "Rotordynamics of Automotive Turbochargers, 2nd Edition", which I do highly recommend for anyone serious about learning applied rotordynamics specific to turbos. So here is the formula (pg 192);
k = (1.3 X 10^3)*(z^(2/3))*(d^(1/3))*(F^(1/3)) [ N/mm ]
where, z = number of balls in the bearing d = ball diameter (mm) F = force applied to the entire bearing assembly (N) (not the force applied to just one ball)
I did a quick check of the SKF 6003, which I assumed had a quantity of 10 balls in the bearing, each being 10 mm diameter and a force of 10 lbf (44.5 N). It should be noticed that as the bearing is loaded more and more, it's radial stiffness increases. So what I came up with was 263053 lbf/in (46067 N/mm). Then I jumped it up to 200 lbf and the bearing stiffness jumps up to 713941 lbf/in (125030 N/mm). So you can see that the radial stiffness of a rolling element bearing is very stiff. Your bearing stiffness number is way off, unless you are lumping the o-ring mounts into that value.
You can't make your bearings stiffer...well...you can go to larger bearings and get the stiffness up but you will quickly run out of DN capability, but rolling element bearings are really stiff and increasing their stiffness would usually hurt you due to increasing amplification values. You can stiffen up the shaft to get your n^th critical frequency increased...but I think you are going the wrong way with trying to do that.
Here is my sweep of bearing stiffness verses critical speed rpm....
As you can see, for me I do not get much benefit in operating space when I take the bearing stiffness to really high values, similar to bearing stiffnesses (500 klbf/in to 1 Mlbf/in). My current design space is trying to keep the bearing system stiffness around 50 klbf/in, which is a good compromise in that I still have a wide space between the 2nd and 3rd modes, but I am also no so soft that my squirrel cage is a flimsy contraption that would be tough to make so soft (very thin or small to allow such flexibility).
Typically, you can't get the first critical above your running speed and if you do, your first mode is going to be a very energetic one. I forgot to mention in the last post that another primary benefit of the soft mounts is that the rotors 1 and 2nd mode can be very solid body like, so when you balance the rotor with the ISO-1940 type method (which assumes a solid body rotation) can work very well. Once your rotor starts to behave flexibly, the balancing becomes more challenging. Because the shaft crosses the centerline differently when it is flexible, the mass distribution changes drastically and you need more balance planes to account for this behavior.
There is a great paper I found the other day when I was doing some research for the squirrel cages.
Hamberg and Parkinson, "Gas Turbine Shaft Dynamics", Continental Aviation and Engineering Corp., SAE 620563
It is an excellent summary of why you want to soften the bearings and increase the shaft stiffness till it behaves rigidly. The paper is specifically referencing their J69 and Model 217-5 engines. I highly suggest it if you have access to SAE papers.
Good luck!
Chris
|
|
|
Post by finiteparts on Jan 9, 2021 22:35:02 GMT -5
Your balance numbers seem way too large for high speed turbo rotors. With your turbine being 90 g*mm and around 1400 g weight, that puts your residual specific unbalance at 64.3 g*mm/kg (as per ISO 1940-1), meaning that you're worse than a G250 balance quality for 70 krpm. If that is the best balance that you can get,I think you might be on a loosing battle. Even for a relatively generic G6.3 quality, you should be down to a permissible residual specific unbalance below 1 g*mm/kg. This means you should be balance to less than 1.4 g*mm for 70 krpm or, in other words, your imbalance is 90X where is should be.
- Chris
|
|
|
Post by madpatty on Feb 19, 2021 12:26:11 GMT -5
Hi Guys.
So after months of research, modelling, analyses and number of wrecked bearings finally I got this beast running.
Current rpm is 42-43000. Startup to running for 7 minutes straight.
The fluid leaking just before the startup in the video is some diesel that got pooled inside the engine when I mistakenly switched on the diesel pump when I connected it’s power.
Waiting on the SiN balls to get some homemade bearings done to get it up to full power.
Regards.
|
|
|
Post by wannabebuilderuk on Feb 19, 2021 13:42:42 GMT -5
Congrats! That leaking diesel made me panic and think it was gonna be a hot start and runaway lol
|
|
|
Post by turboron on Feb 19, 2021 14:34:14 GMT -5
Patty, what is the current squeeze film damper configuration? Inquiring minds want to know.
Thanks, Ron
|
|
|
Post by racket on Feb 19, 2021 18:46:36 GMT -5
Hi Patty Looking good :-) Could I make one suggestion .................your blower starter appears to have an outlet tube diameter smaller than your comp's inducer , if this is the case it would pay to fit a reverse "funnel" section so that the blower can be "sealed" against the comp housing to maximise the available "pressure" from the blower to minimise your spoolup times Like the "funnel" on the end of my blower www.youtube.com/watch?v=02kK7DKk-1Q&t=3s , just a plastic funnel trimmed to size and taped to the end . LOL.............your offsider has a very worried look on his face at the spoolup ............he was feeling more "comfortable" at the spooldown :-) Cheers John
|
|