100mm Inducer jet engine won't start

[racket]

Hi Patty

Thanks for the drawing :-)

May I suggest you do a check with a dial indicator to find out just how your shaft moves back and forth under the influence of the wave washer/s .

Setup the shaft/comp in the bearings with everything in place and tightened up , then with the dial indicator mounted on the shaft tunnel flange so that the indicator shaft rests against the end of the shaft , force the shaft forward and back to determine what is happening and to what distances the shaft can be moved .

With my FM-1 engine I only had ~0.006" endfloat , a nice little "click" back and forth .

The fact that you had a comp bearing fail sorta indicates a problem , I never had a comp bearing fail , they always came out looking as new even when the turb end bearing was badly worn/failed.

Maybe theres binding of the sliding components resulting in the bearing preload "unloading" .

Theres a lot of potential reasons why your bearings are failing , we just need to work through them , could you please elaborate more on the modifications you did to the bearings to change them over to full compliment with SiN balls , grade of ball , sizes , were they exactly the same diameter as the steel balls they replaced , etc etc etc ................the more info you supply the more chance we have of solving this problem

Cheers
John

[jetjeff]

Hi Patty,

I've never seen a rotor setup with preload on both bearings. I'm wondering if machining the inner race on one and outer on the other causing an imbalance of the bearing itself? Whats the surface finish on the machined areas of the bearings?

Your wave washer may not be providing enough "travel" when the rotor gets up to temp.

I might forego the compressor end preload altogether and substitute one of the flat washers, on the turbine end to use two wave washers in this sequence SlS.

Regards

Jeff

[finiteparts]

Oct 23, 2020 8:19:29 GMT -5 madpatty said:

Hi Chris and John.

Thanks for your valuable inputs.
If you think a normal unbalance cannot produce such a catastrophic failure every-time, that too at a low rpm of 24k then definitely we should be looking somewhere else.

I always thought these shafts are ground between centers and that's why those machined centers on either side of the turbine shaft and the inconel wheel are there for. I also read somewhere that these centers in the nose of the turbine wheel are located at the centre of mass of the wheel and not the geometric centre.

If I may ask, what do you think should be the best way to check for any misalignment or eccentricity in the shaft? I have some basic tools only like a lathe and a dial gauge which may or may not be able to measure this angular misalignment or centreline offset.

I did a small checkup last time when this thought that this shaft maybe bent crossed my mind.
I held the shaft between the 4 jaw chuck and the tailstock dead centre(turbine wheel's outer 'hex nut like' portion held between the jaws and the machined centre on the compressor side of the shaft in the tailstock dead centre). I then Dialed the shaft to near zero on the turbine end (jaw side of lathe) and then checked for any "bend" in the shaft at various points across the length of the shaft.

All the points were within 0.01-0.02mm.

what should be the other way of doing it for checking all angular alignments and collinearity.
Should I be holding the turbine shaft between it's factory machined centers on both sides.(By machining a dead centre in the lath first and a dead centre on the tailstock length).
I can also grind the shaft between its factory machined centre so that everything is true and straight between the factory machined centers.

What you guys think? I need your suggestions on how this system should be checked for trueness and how further improvements can be made in case we need to be sure there isn't any misalignment.

Regards.

Patty,

You are correct.  I have been focused on resizing the shafts on my turbos to meet the fit requirements for the bearings, which lends itself better to centerless grinding and apparently just blurted out centerless grinding without thinking.  In general, yes they probably do use the drilled centers to locate for od grinding (also referred to as cylindrical grinding).  When they are grinding the od of the shaft, they are targeting tolerances in the tenths (+/- 0.000x inch) or +/- 2.54 um. 

I highly recommend downloading SKF's bearing catalogue and reading through the engineering section.  It will give you a strong appreciation for the many factors that go into the bearings application.  You can find it here:

www.nodeshk.com/skf/art/skf-bearing-catalogue-pdf.html

It might be interesting to you to look up your recommended fits for the bearing inner race to the shaft.  If you assume a normal load on the shaft and use the value shown for the stainless steel shafts (allows for a bit more thermal expansion), then you will see that the shaft itself should have a deviation of -4 to +4 um (max to min local radius variation from the centerline).  The theoretical interference that is recommended should be between -12 to + 4um.  You need to have a very controlled fit for rolling element bearings. The IRC on these bearings is very small, especially for high speed bearings and a portion of that is taken up by the fits on the shaft and housing (this is why Ron was asking if you had any press fits).  So you have to be very accurate on these fits.  In fact, if your shaft was poorly machined and had a slight egged shaped profile at the bearing seat, instead of completely cylindrical, that shape gets translated through the inner bearing race and you can get tight spots corresponding to the high spot.  This would cause local pinching as the balls passed through and more heat generation there.  This may be what you are seeing.

I struggle to understand why you would only get one hot spot on the rear bearing inner race.  The above is one idea, also if the shaft was slightly bent so that the inner bearing was stationary to the tight spot, I think that might do it.  Ron's idea of instability is also a possible idea, but the loads seem to be too low at 24k rpm. 

I ran the calculations for your bearing (as a 6003, not as a modified angular contact) just to get a raw order of magnitude (ROM) idea of what sort of loads would be required to exceed the L10 life in less than a minute.  This is still conservative, but should let us get a feeling about how this might play out.  So to exceed the L10 life in 1 minute, it requires 22.08 kN (4964.6 lbf).  If this imbalance was located at 1.25 inches from the rotational axis (seemed like a good guess based on the balance spot onmy HT80 turbine), then you would have to have 0.243 lbm of imbalance to generate the 4964 lbf.  This seems like a bit of a stretch, since I think that just rolling the rotor on a table would easily highlight a quarter of a pound imbalance.

As stated previously by Jeff, I would use v-blocks, on a granite plate (or other VERY accurate surface plate) to establish the primary and secondary datums. Use a straight edge to align the v-blocks very accurately.  I have used soft plastic covered clamps to clamp a thick machinist bar to the granite plate, then small clamps to lightly clamp the v-blocks to the bar.  The v-blocks have to be very accurately aligned to establish the primary and secondary datums over the length needed to carry the shaft.  The blocks have to be super clean...wipe them down, wipe the table before sitting them on it, wipe everything...you are checking for accuracies in the microns. For the tertiary datum, you need something like a square block to set the axial location.  Here is a quick sketch of the set-up.



Check several planes and record the dimesions.  What I like to do is to mark 12, 3, 6 and 9 o'clock so that I can know that I am not getting confused about wthe shape of the shaft.  What concerns me here is that if the shaft was turned on a lathe, it is very hard to keep the cutting force and heat low enough so that the shaft doesn't get bent.  Even when od grinding, it is super important to have the cooling keeping the part cool or the thermal distortions (grinder side hot and other side cooler) to keep the tight dimensions that you need.  My second major concern is that you have sleeves on the shaft, which also have their own manufacturing tolerances. You now have three tolerances stacked up.  The shaft tolerance, the collar id tolerance and the collar od tolerance. This could easily cause a centerline issue.  If possible, you might also want to check the runout with the collars pressed onto the shaft...maybe even rotate the collar 90 degrees relative to the first postion that you install them to see if they have some eccentricity that is being put into your shaft.

You have something seriously out of alignment to cause that damage.  Like I said before, the issue must be stationary relative to the shaft , because if it was stationary relative to the housing, you would have a hot spot all the way around the inner race.

I hope your shaft measurements show some issue, because beyond what I have said here, I am not sure what else you can do.

Good luck!

Chris

[finiteparts]

Patty,

An additional concern that I have is the home-brewed angular contact bearings. I know that these have worked for smaller engines, but when bearings are properly designed for the use of ceramic balls in steel races, there is usually a change required in the race groove curvature. When you are using steel on steel, the balls and the race actually elastically deform slightly at the contact surface, which adds a tiny amount of area to spread the load that is being carried through the bearing. This means that the Hertzian contact stress is slightly reduced as opposed to the "ideal" contact based on the bearing and race geometry only. As you might have guessed, the ceramic balls are much more stiff than the steel and thus they cause the contact ellipse to be smaller for the same load. Thus the contact stresses are higher and the subsurface stresses are more peaked, leading to much higher fatigue stresses. Manufacturers generally decrease the raceway groove curvature to "spread" the load more evenly, which is being missed in these home-brewed bearings.

I also saw what James said on the GTBA site and I also agree that skidding due to lubrication may be a potential cause. Since I have never seen such a sever case of smearing, it didn't even click in my mind. Smearing is usually much smaller and more local, but the metal is smeared quite substantially. Again,I would have thought that the 24k rpms would be too low to start thinking about skidding, but depending on the amount of lube, it is a definite possibility.

I also had a thought about your collar providing a contact resistance from the heat conduction. But if there is a better fit at one spot between the shaft and the collar, you might have a local heat conduction that is higher than other locations on the collar and thus you might get a thermal closure.

Lots to think about. Good luck!

Chris

[racket]

Hi Jeff

With both my 9/94 and FM-1 engines there was a single large coil spring between the two bronze bearing cups , the whole shaft with bearings and cups were free to move back and forth within the 0.006" endfloat, the comp shroud clearance would have accomodated the comp axial movement without contact, there being plenty of clearance at the turb end for any eventuality , but as the axial thrust is forward with our turbo rotative the engine was setup with that in mind .

Having only a single preload spring meant both wheels always had preload

Cheers
John

[racket]

Hi Patty

Heres some "parameters" from a 18/4/2008 test of FM-1 that might be of interest for comparison.

With some diesel poured through homemade hybrid 6204 bearings (20 X 47 X 14mm) loaded with used SiN balls from a previous failed test , and with 25 lbs preload in the engine , the leafblower produced ~3,500 rpm , on removal of blower it took 40 seconds for the rotor to come to a halt .

After a 5 minute engine run things were warmed up and the leafblower produced 4,000 rpm and it took 65 seconds to come to a rest.

A finger flick of the warmed compressor took 18 seconds to come to a stop .

Spooldown from idling/shutdown 25,000 to 0 rpm took 93 seconds, ......them there SiN balls are super smooth compared to steel balls , steel balls only produce ~50% of these times .

It took me nearly 18 months to get my bearings to survive , it was a heartbreaking experience everytime a rotor was wrecked , hopefully we can get yours sorted a lot quicker.

Have you tried using standard bearings , unmodified , I initially used standard 6204 "rattly" C3 deep groove bearings and ran the engine at 35,000 rpm before making up the hybrid bearings, the SKF catalogue gives a Reference speed of 32,000 and a limiting speed of 20,000 for the 6204, your smaller bearings have another 4,000 on top of my speeds, so something approaching 40,000 rpm for you , the standard bearing will indicate whether or not your modified bearings are at fault .

Cheers
John

[turboron]

Patty, I support John's suggestion on switching to an SKF bearing. Your failures maybe caused by more than one factor. Going to a standard SKF high speed bearing could perhaps eliminate one of the contributing factors.

Thanks, Ron

[madpatty]

Hi Guys.

I already have tested once Standard SKF 6003 bearings on this engine with same catastrophic results. Both bearings destroyed. I thought the cage came apart and then started the chain reaction.

Pics below-










I did some recent investigational study on my rotor and other engine components and found some conflicting results.

I went to a local shop who repair and grind crankshafts etc. I also took a brand new turbine shaft for comparison purposes.

Trial 1 (Hold both shafts between centers, chuck side center is dialed zero using a dial gauge)

Old Shaft

Area closest to Turbine wheel(where oil control piston ring goes)- runout of 0.22mm
Compressor side(where bearing goes) - 0.06mm(or less)

New Shaft

Area closest to Turbine wheel(where oil control piston ring goes)- runout of 0.15mm (This is what appears conflicting to me, I expected it to be quite less)
Compressor side(where bearing goes) - 0.06mm(or slightly less)

Trial 2 (Hold both shafts between centers, Turbine wheel side oil ring groove region of new shaft is dialed zero using a dial gauge, chuck is adjusted)

New Shaft

Area closest to Turbine wheel(where oil control piston ring goes)- dialed 0
Compressor side(where bearing goes) - Almost zero or below least count


Old Shaft

Area closest to Turbine wheel(where oil control piston ring goes)- runout of 0.20mm
Compressor side(where bearing goes) - 0.06mm


Not convinced by the readings of the new shaft I went to a friend's garage who has a lathe installed. We don't have dial gauge now but we somehow set a visual stationary reference point to see which shaft appears visually bent.

Trial 3

Method- Machine a dead centre held in the chucks so that you know it's running true to lathe axis and then hold both the shaft between the machined dead centre and tail stock dead centre

New Shaft-



Old Shaft-



Now the new shaft doesn't visually look that bent(0.16mm runout) atleast as compared to Old shaft(0.22mm run out)

Regards.
Patty

[racket]

Hi Patty

This pic ibb.co/kqHMW1S indicates a shaft that has been extremely overheated , the fact that colouration has extended all the way back along the shaft is worrying , no wonder the bearing failed .

At what speed did the standard ball race fail at , after what time frame of running ??

We need bleed air across the NGV to provide "cool" air at the turb boss to prevent this sort of severe overheating ..................LOL, I've abused a few turb wheels but have never seen this amount of colouring :-(

Even your front bearing failed ibb.co/Hx97sRV .................what are all the "brass" bits ??

It looks suspiciously like a lack of lubrication/bleed air to allow such shaft colouration and turb bearing inner raceway colouring

Cheers
John

[madpatty]

Next step.

I went ahead and got that old shaft machined so that at-least now shaft is running true between the centers (like the new shaft was).
I also machined new spacers and bearing ID journals so that everything matches the new reduced diameter of the shaft.



I was expecting at-least some improvement after all this.

BUT again same failure.

Disappointed I haven't opened the engine up for checking after this last failure. I don't know which bearing failed this time but I am pretty sure it's the rear bearing.

Failed-



Regards.
Patty

[madpatty]

Oct 25, 2020 21:46:18 GMT -5 racket said:

Hi Patty

This pic ibb.co/kqHMW1S indicates a shaft that has been extremely overheated , the fact that colouration has extended all the way back along the shaft is worrying , no wonder the bearing failed .

At what speed did the standard ball race fail at , after what time frame of running ??

We need bleed air across the NGV to provide "cool" air at the turb boss to prevent this sort of severe overheating ..................LOL, I've abused a few turb wheels but have never seen this amount of colouring :-(

Even your front bearing failed ibb.co/Hx97sRV .................what are all the "brass" bits ??

It looks suspiciously like a lack of lubrication/bleed air to allow such shaft colouration and turb bearing inner raceway colouring

Cheers
John

Hi Racket.

The bearing failed at about 30k rpm. Within a few seconds. I don't recall exact time frame BUT i never crossed that point. Almost failed instantly.

I have 12 x 2mm holes pointing directly at the turbine boss area.



The brass bits are from front compressor spacer which when everything went into gyration contacted the front seal plate.

Regards.

[racket]

Hi Patty

What grade of standard bearing did you use ??

With the degree of colouration present one could assume the shaft reached maybe 400 deg C , now 16 mm dia X 0.00001 ( coeff of expansion) X 500 = 0.08 mm , the outer ring being fitted to an alloy heat sink would be running a lot cooler , maybe at ~100 deg C say , so theres considerable reduction in the radial clearance between rings .

The SKF catalogue gives clearance for a 17mm ID bearing in microns , the sudden expansion of the inner ring would jam your balls :-(

I'd suggest you greatly increase the bearing cooling and fit the "rattliest" grade of bearing possible, the Allison C20 bearings I ended up using are very rattly .

Could you please post a pic of your NGV like in the above representation so that we can determine how hot the area around the securing bolts gets , my 12/118 stainless is still shiny "chrome"

Cheers
John

[madpatty]

Oct 25, 2020 22:51:16 GMT -5 racket said:

Hi Patty

What grade of standard bearing did you use ??

With the degree of colouration present one could assume the shaft reached maybe 400 deg C , now 16 mm dia X 0.00001 ( coeff of expansion) X 500 = 0.08 mm , the outer ring being fitted to an alloy heat sink would be running a lot cooler , maybe at ~100 deg C say , so theres considerable reduction in the radial clearance between rings .

The SKF catalogue gives clearance for a 17mm ID bearing in microns , the sudden expansion of the inner ring would jam your balls :-(

I'd suggest you greatly increase the bearing cooling and fit the "rattliest" grade of bearing possible, the Allison C20 bearings I ended up using are very rattly .

Could you please post a pic of your NGV like in the above representation so that we can determine how hot the area around the securing bolts gets , my 12/118 stainless is still shiny "chrome"

Cheers
John

Hi Racket.

The Bearing was this one-
www.skf.com/in/products/rolling-bearings/ball-bearings/deep-groove-ball-bearings/productid-6003

I think it had P5 clearance.

The shaft tunnel is made of stainless steel as well.

I don’t think shaft temperature reached 400degC.
my thoughts are since I don’t run any cooling cycling on a failed engine all the turbine heat is soaked by the shaft and turbine end bearing. MAYBE.


Here are some NGV pics that I have after a bearing failed.










Thanks.

[jetjeff]

Hi Patty,

What is the shaft made from?

Regards

Jeff

[racket]

Hi Patty

This pic imgbb.com/sV1pCZF indicates flames/hot gases at the shaft ............not good ............your bleed air across the NGV isn't working :-(

Could you post a pic of the flametube please , your engine is giving me the impression its suffering from the same problem I had with my 10/98 before fittting the air deflectors at the diffuser outlet

Cheers
John