TV94 Turboshaft Project

[stoffe64]

nice ash that they fixed that NGV part for you,looks very nice everything,im following this build with great interest!

[jdw]

Hi Ash great project. I was curious if there were any new updates on this build??

[jetfuel]

Hello Ernie,i am a new member,i am verry interesst on this TV 94 engine project. I like to know please its this engine ready to run today.Will be posible to get some more information please.


thanks jetfuel

[ashpowers]

Long time but still at it. Just yesterday I took apart a turbocharger, a power enterprise PE1420 turbo. It is a dual ball bearing turbo and I was itching to see the design. Disassembly was pretty straight forward and once getting all the pieces out I was pretty amazed by the construction I found.

The rotorbearing group consists of two angular contact bearings with a non-metallic cage, an independent bearing tube that utilizes a dual annular oil dampening channel, and a preload spring. Both bearings are pressed onto the shaft and each having a washer that contacts the outer race on the inboard side in which the preload spring is acting upon. The bearings slip freely into the bearing tube with precise fitment but no interference. The rear bearing's preload washer has a tapered face on the side closest to the bearing with a very small hole drilled radially through its mid-section and its outer face has a single-channel labyrinth-esque shape.

The cast iron housing has a single feed port that delivers oil to two points within the bearing tube bore. These feed ports are located so as to deliver oil into two volutes that are formed by two radial rings on the bearing tube's outer face to create an oil dampening effect.

Lube is delivered to the bearings via a galley running axially through the cast iron cartridge and feeding the front bearing from its front side and the rear bearing also through its front side, however, the preload ring for the rear bearing sits center to the oil feed port and the labyrinth-esque face is actually an oil channel to feed lube through the small radial feed hole. From there, the oil will pass through the rear bearing towards the turbine and exit the bearing on the rear side. The housing has a return port to direct this lube into the main oil return galley. In this configuration the lube is moving through both bearings in the same direction, front to rear. The drain port in the bearing tube is solely for oil return coming from the front bearing.

This particular turbo has shown exceptional lifespan as compared to the garrett turbocharger lineup which does not use anything close to this arrangement. In the garrett units there is no preload spring and no oil dampening volute.

Here are some pics of the setup:



rear bearing


bearing tube


housing, turbine end


housing, dampening volute feed ports


housing, oil return ports


I'm really digging this design.......

[Johansson]

Very interesting Ash! Such a simple design should be possible to copy for a DIY gas turbine.

Will the rotor dynamics stay the same with ball bearings located in the same place where the journal bearings used to be, or does everything change?

[ashpowers]

Johansson, rotordynamics are the million dollar question around here it seems. The placement of the ball bearings on the shaft are very similar to that of a journal bearing turbo. Slightly more spacing between them given there is no thrust bearing assembly forward of the front bearing as found in journal bearing setups. There is about half the distance between the bearing and compressor wheel in this ball-bearing turbo as compared to a journal bearing turbo of the same size.

I have played around with many different design layouts in solidworks to achieve similar design goals but as you said, this relatively simple design does it quite ingeniously and obviously effectively based on the past decade of experience I've had with these particular turbos.

I like the fact that the outer races are not pressed into the bearing tube. Allowing them to float will prevent any race alignment issues and the preload spring loading against the outer races to prevent them from free-wheeling as well as maintaining a positive ball/race load for the rear bearing covers all the bases there. Adding in the dampening volutes to the bearing tube is also genius.. And then the oil flow arrangement is like a forehead slap.

By controlling the feed lube into the front of both bearings, it mitigates competition of return flow.

This is by far the best arrangement I've seen yet for a ball bearing setup. Too bad I didn't find it years ago, lol..

[ashpowers]

Also, just to note, I was a bit surprised at how little preload was used. This spring is compressed by about 1/2" and just by the feel of hand when giving it a squeeze I would say there is probably only about 1 to 1.5lbs of force at that length. Granted, this is only a 60mm compressor wheel and bearings of about 18mm od, 8mm id, and 7mm width.

Racket: did your attempts previously with ball bearings (in fat mumma) were you pressing the outer races into the bearing sleeve or did the bearings have a close sliding fit?

IIRC, it was the front bearing in FM that was giving out, suggesting the axial loads were too much. Perhaps doubling up on the front bearing to spread that load would work?

What impresses me is the longevity of these BB turbos - they are used in applications such as 24 hour races where they are experiencing probably 40-60% duty cycles being pushed near their choke limits in pulsating exhaust piston engines, bouncing all over the track, accelerating, decelerating, and hard cornering. They go through hell on a race car but they stay together. They are simple dual bearing arrangements, no fancy interstage pressure control, minimal part count, no ceramics, and just engine oil.....

Im going to make some time in the next few weeks to get back on this project. Been too long and I'm all fired up, lol.

[racket]

Hi Ash

Great to see you're back into the project :-)

With the final configuration of FM-1 and latter 9/94 engine , the ball races were a sliding fit into bronze cups which had a pair of O'rings on their OD between which the pressurised lube was supplied to provide a damper between the loose fitting cup and the shaft tunnel and between the outer raceway OD and cup ID from through holes in the cup , there was a radial hole drilled in the "end float?" side/bottom of the cup which fed oil to an angled "jet" hole that squirted onto the inner raceway , it being the raceway that normally gives problems due to centrifuging of the lube.

A single large diameter spring was positioned between the bronze cups to provide preload to prevent ball skidding under acceleration of the rotor, the outer radial edge of the cups produced the endfloat in the system when the cups sat in their recesses in the shaft tunnel .

In the end I used time expired Allison C20 #8 bearings , 20 mm ID X 47mm OD , combined with an axial labyrinth seal running on the back of the comp wheel to reduce air thrust loads , the Allison bearings seem to work OK , probably helped by their superior metal hardness compared to the "off the shelf" industrial bearings I was having problems with,........ yep , front bearing behind comp would be where the axial thrust loads are if that bearing is fixed , but in my arrangement the rear turb bearing took the loads as the shaft was pulled forward under air loads on the back of the comp, this then bottomed out the turb bearing into its bronze cup which in turn bottomed out in its recess in the shaft tunnel , the forward movement of the shaft reduced comp shroud clearances , depending on the size of the labyrinth seal thrust could be in either direction .

Some of the turbo ball bearing raceways are in M50 material .

Its interesting how low the preload spring force is in that turbo, the JFS turbine starter runs 100 lbs preload in the gas producer and 25 lbs in the freepower , but then its designed for use in minus 65 F conditions with the lube viscosity somewhat increased , whereas a turbo shouldn't be on load until the engine lube is up to temperature and nicely thinned out .

When I first started making FM-1 I thought I'd just use the "standard" RC micro turbine "simple ??" system of bleeding off a few percent of fuel mix flow to lube the bearings and combined with bleed air for cooling them , but it didn't work out so well due to a number of problems both with the cooling air flow and the lube , .......in the end I went for the slightly more complicated "standard" gas turbine practice of oil jet lubrication only to take care of both lubrication and cooling..........it would have saved me a lot of headaches if I'd gone that way in the beginning

With my latest engine build , the SKULD 666 , it'll be brass bushes all the way, with lotsa pressurised lube .

Looking forward to more news on your build Thread :-)

Cheers
John

[finiteparts]

Thanks for posting these pictures of this bearing arrangement. It is interesting how the damper is arranged....Garrett and Borg Warner use a more traditional approach to the squeeze film dampers similar to what is used on many aircraft engines. Typically the squeeze film is relatively thin...around a few thousands of an inch, spread over the length of the bearing cartridge. It's interesting that they have four relatively small "squeeze" films separated by the annular volume. What kind of clearances do you see between the cartridge and the housing? What about the bearing to the cartridge?

Bearing pre-load is always a challenging calculation. Most of the major bearing manufacturers offer free design guidelines for calculating general usable life (typically called L10)...Timken offers a nice engineering manual:
www.timken.com/en-us/products/Documents/Timken-Engineering-Manual.pdf

Also here is a good article on bearing internal clearance: machinedesign.com/mechanical-drives/understanding-bearing-internal-clearance

Figure 2 of that article shows the typical bearing life vs preload curve...the important thing to take away from that curve is that if you get the preload wrong, it would be better to be on the excessive clearance side. Excessive preload will destroy bearings very quickly as can be seen by the steep drop in life after ideal preload is reached. This is further complicated when you try to go to really high rpms, since the centrifugal stresses add to the contact stresses of the balls on the race, effectively increasing the preload. Most of the big manufacturers (Barden, Timken, SKF, etc) usually give suggestions for bearing preload.

As for the lubrication, I have been going through some nice papers on oil-air mist and oil-jet lubrication. Here is a nice NASA paper applicable to our goals...

ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20010090717_2001149983.pdf

From this you can see that bearing temperature is the major issue, doubly so when we stick them on the hot end of a rotor shaft! So the challenge will be to determine the best way to remove the heat from the aft bearing. My Fairchild J-44 engines were built back in the 1940s and incorporate bearing thermocouples and air-oil mist...I will be putting in bearing thermocouples to monitor the bearing health and maybe an air-oil system... but, the use of an oil-jet arrangement seems to lend itself better to using pressurized squeeze film dampers, so there is some more research to do before finalizing the design. SKF has some really good info on their online catalog...here is a example for the bearing that John describes above...

ng-origin.skf.com/group/products/bearings-units-housings/super-precision-bearings/angular_contact_ball_bearings/acbb-skf-high-and-super-precision/index.html?prodid=1270510204&imperial=false

Notice that they give a suggested location for the oil-air jet location and bearing preload values for various classes.

I am working on the rotordynamics for my engine and the bearing selection has become a big issue since it defines the stiffness. One of the other design features that I am playing with is to shorten the distance between the compressor and turbine wheel since this greatly increases the rotor natural frequency.

I am in the process of relocating for work and I haven't been able to finish up the design work....but hopefully soon, I can get moving on this. Good luck to you on your project!

~ Chris

[ashpowers]

Interesting find today in Modified Magazine. The Turbonetics x275 turbo uses a combination of ball bearing and journal bearing for the rotating group.....

[racket]

Hi Ash

LOL....an admission of failure to control the rotor dynamics with all ball bearings .

That big lump of Inco on the hot end has some serious energy when it starts to wander around .

Cheers
John

[ashpowers]

Hi John,

I dont think I'd go as far as to say that. There are a lot of turbines of similar sizes to ours that use all rolling element rotorbearing systems. One difference though is that for the most part all of these engines use a cylindrical roller bearing for the turbine end, which makes a lot more sense. It will allow for thermal dimensional variations and a roller bearing is capable of handling higher radial loads. Ball bearings require preload to sinch up the slop but with sufficient radial loading and/or insufficient preload it will force the inner race to shift away from centerline. Roller bearings do not have this issue and from what I'm seeing they also tend to have higher speed ratings compared to similar size ball bearings.....

If a rear roller bearing were to be used and a duplexed angular contact ball bearing set used on the comp end, all contained within a common bearing tube where the bearings have a non-interference, no slop fit into the tube, the inner race of the roller bearing is pressed onto the turb shaft, the front end of the shaft machined for a non-interference fit into the front bearings' i.d., the shaft having a flange that butts up against the back of the rearward frontal bearing's inner race, the shaft seal sleeve abutting the front bearing's inner race where the seal sleeve is locked down by installation of the compressor wheel thereby constraining the frontal bearing group to the shaft. Vibration dampening would be handled by a clearance between the bearing tube and the bearing housing in which the lube would be forced through before entering the bearing tube. The roller bearing at the rear end will constrain the radial motion of the turbine to keep that heavy end true to the centerline of the bearing tube and as a result, prevent any alignment problems with the duplexed AC bearings at the comp end. Any rotational dynamics will be buffered on the rotorbearing group as a whole since the bearings share a common tube; the tube's motion is dampened rather than the individual bearings being dampened. There's a lot more surface area along that bearing tube than what the sum of the bearings themselves have at their outer races. In effect, this is pretty much the same idea we see in Garrett's GT ball bearing turbos - they share a common outer race "tube" that constrains the outer races to perfect alignment all of the time. That tube also has a clearance between it and the housing of which oil is fed to create a damper. Only difference here really is that we would be using a roller bearing at the back end.



The picture of that Turbonetics turbo with the journal bearing at the rear does raise some questions, and one of my own. At what point will a turbo manufacturer take a leap to gain market advantage through the performance advantage resulting from using a roller bearing at the turbine end vs journal?


Do you know of any high precision cylindrical roller bearing used in a production turbine that would fit our application?

[racket]

Hi Ash

I think the garrett "full sleeve" ball bearing setup is probably patented , so the turbonetics guys are trying to find a way of not paying royalties , but I still don't like their design as theres going to be considerable clearances in the "brass bush" , this has got to produce misalignment issues with the ball race at the other end of the shaft , the Garrett setup is the way to go to prevent this .

One other problem worth mentioning is , compression or tension within the shaft , which reduces/minimises/eliminates the shaft dynamics best, .............would a roller bearing cause more problems with a turbo rotor ??

The setup I used in FM-1 and the 9/94 engines allowed the preloads to work at both ends of the shaft.

Checkout these guys for aircraft grade bearings ....

www.bardenbearings.com/content.barden.us/us/products/products_1.jsp

I used the #8 bearing from the Allison C20 , 20mm bore ball race with massive amounts of axial play ( ~0.014" ) .............clunky :-)

Cheers
John

[ashpowers]

Hi John,

Agreed on the design point re: royalties, as well as the alignment issue you point out. I have some questions about the front bearing layout with that Turbonetics turbo - its just a section picture and really only the housing was sectioned. I can't tell what other bits may be in that collection of parts at the forward end of that turbo but I do agree with you - that combination of bearing support raises an eyebrow.

I really dont know what the differences are in the rotor's stability when comparing the use of any of the rear bearings we are talking about. All of the complex rotordynamics that occur in our homebuilt engines can be taken into account and an appropriate design can be engineered with mathematical precision. Problem is really the amount of time and money it will require to obtain not just the software, but all of the accurate information pertaining to every single component within the rotor and rotorbearing group. There just isn't a shortcut around the requirement to mechanically test how all of the components behave individually. Without that information and process, we are going about it a very hard way of trial and error. But we all like a challenge, yea?

The only difference I Really see between using a roller bearing or ball bearing at the turb end with respect to shaft compression/tension would be dependent on if a preload spring is used, what amount of force does it apply, and what effect does it have on the modulus of the shaft? Unfortunately that last question is where the difficulty comes from; we just dont know. Best we can do is try to over-engineer the parts and cross our fingers.

After seeing these other production turbines' rotorbearing systems I do feel the need to smack my forehead.

[racket]

Hi Ash

Trial and error development gets very expensive in rotor replacements as well as emotional capital, as I found out with FM-1 :-(

After the third or fourth "malfunction" the spirits start to flag and you question if you're on the right track or going nowhere fast, thats why it was a relief to go to "brass" and simply use off the shelf components, Garrett can afford development "costs" , us DIY'ers can't . ................if we're lucky its only replacement rotors and bearings but if other components get collateral damage and need replacing, its stops being an enjoyable hobby .

LOL......I did a bit of head smacking myself :-)

Cheers
John