Rotating shaft specifications 1

[Patrick02]

Hi all,

Does anyone have information on manufacturing methods and specifications for rotating shafts operating in a process industry in order to minimize shaft run-out and vibration problems? The industry allowable vibration grade is G6.3 I believe.

What I am interested in is the experience of anybody involved in this type of project.

I am a recently graduated mechanical engineer who now works for a company that manufactures slurry pumps for process industries.

On one of our vertical spindle pump products we have been experiencing synchronous shaft vibration, which is consistent with a rotating unbalance. One likely cause for the rotating unbalance is that the shaft, which is long and slender, has a run out during operation. The run out is possibly caused by residual stresses formed in the shaft during manufacture, which then causes the shaft to bend during handling and assembly into the pump.

I have been tasked to look at the shaft specifically as part of a wider program to analyze the rotordynamics of the pump. My duties include looking at the way in which it is manufactured and to suggest a method which will reduce potential run out of the shafts.

The shafts are typically 2100mm long with a mean diameter of 70mm. They are supported by two bearings 700 mm apart and have a large overhang (1300 mm or so) from the non drive end bearing. They are currently manufactured by our supplier, who purchases steel bar from a forging company and then performs milling operations on the shaft to create the bearing locations etc. No heat treatment is performed on the shaft.

I was just wondering if anyone has come up against a similar situation where they had to look into manufacturing methods and specifications for rotating shafts.

Kind regards,

Patrick

 

[Tmoose]

I'd measure what I'm making, and dissect a few problematic pumps to better understand the relationship of runout to vibration. Do you have a runout or straightness tolerance for the finished shaft? If so, what are the datums referenced? A starting point for a shaft runout might be the mass eccentricity equivalent of G6.3 .

The repair and assembly procedures for Devices like turbines and multistage pumps often include runout checks before and after each phase of assembly. I suspect these evolved soon after struggling with problems of good parts inducing runout at assembly.

Cutting keys, assembling impellers, clamped impellers and spacers, and even commercial shaft tolerances.

Reliability Magazine and others have bulletin boards that have many discussions about the ways in which vertical pumps can be messed up. 1X vibration can be caused by many things.

http://www.corrosion-products.com/images/Vertical Pump3.jpg

 

[Patrick02]

Thank you for the advice Tmoose. I am currently developing a runout specification for our shafts based on the G6.3 specification. To do this I will select three points on the shaft and a datum value. I need to figure out at each point what the maximum allowable runout can be in order to give the specification value me=6.3 kgm by modeling the shaft as a discrete system. I am working on this at the moment.

Thanks for the picture and the reference to Reliability magazine. Once we have ensured that the shaft itself is straight we still need to tackle the fact that the impeller at the end of the shaft will wear and change its CG over time. As part of the larger program we are also considering the rotordynamics of our pump under various operating condistions. I have produced a rudimentary model on the effect of shaft runout on vibration amplitude at any point along the shaft. I still need to include the effects of eccentric bearings though.

I will check out the forum in Reliability magazine

 

[electricpete]

For slow-roll runout conducted with dial indicator, the requirements should be given by the manufacturer. If not 0.0005” per foot of shaft length is limit suggested by one EPRI pump document.

Runout at selected locations can also be measured at rotating speed using proximity probes.

I agree with the comment above that in general there are a wide variety of possible causes of vibration at 1x:
Operation near critical speed or resonance (do a coast—down test to check)
misalignment
unbalance – possibly due to eccentric assembly or excess clearance between assembled parts (may also show as harmonics).
etc


Now G6.3 is a balance spec (and a fairly loose one for critical applications). I have never heard of converting a balance spec into a runout spec and at first glimpse I don’t think I like it. Do you plan to allow for dynamic bending of the shaft?


=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

Also if other components are mounted onto the shaft, there should be an allowance for cuumulative unbalance.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

Back to the subject of dynamic bending... that was based on your description of long/slender shaft... but I guess it still may not be necessary if this were a slow-speed applicaiton. What is the speed?

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[Patrick02]

Hi electricpete

The application is high speed (between 1000 and 2400 rpm). The 1st lateral critical of the rotordynamic system is below 1000 rpm and the second critical is above 2200 rpm. So you are correct in that I must take into account dynamic bending. In this case the first mode shape of the shaft is a basic cantilever “wagging tail” shape. Also the ratio of the mass of the shaft relative to the impeller is something like 8:1. I thought I would use the G6.3 as a starting point (although as you pointed out it is not suited for my application, only rigid single degree of freedom bodies). I have produced a simple excel model which models the shaft as a continuous uniform beam and calculates the total run out of a shaft at any point and at a certain speed based on an initial shaft run out value at the impeller. It is essentially a static analysis using Euler beam theory and simple mrw^2 to generate deflections which then get added to the initial run out value. I hope to find the initial value that would make the total value of the run out less than G6.3 at the impeller and at the maximum speed of the pump. If successful my plan is to specify that value as the maximum allowable run out value for the shaft tip. I also plan to include bearing misalignment into the equation.

If that is not sufficient I may model the pump as an n-degree of freedom system and repeat the analysis, this time solving for the dynamic mass, stiffness and damping matrices with the forcing function being mw^2rsin(phi). The aim would be to get deflection values along the shaft so I could work back in order to specify maximum initial run out values for the system. This may be slightly more than what is needed which is why it is my secondary option.

Once I have value of maximum permissible shaft run out the plan to develop an inspection rig to test just the run out of he shaft alone and then later the run out of the integrated shaft, bearing assembly when assembled. I will compare measured values with those obtained in my model. We have not decided what measuring instruments to use but we may settle on lasers. I have not done my research in this area yet and so the jury is still out on that one.

Does that sound reasonable for developing a shaft run out specification chart and corresponding inspection procedure or am I complicating matters?

As for the impeller, although it is statically balanced it does not undergo dynamic balancing because it wears within a relatively short space of time in the field and the wear pattern typically is not axisymetric. Since its mass is typically 1/8th of the shafts I cannot see any eccentricity in the impeller causing significant vibration, even though it is located at the tip of the shaft. I performed calculations to check this assumption and I found I needed a very high eccentricity value on the impeller to make it cause significant vibration in the shaft (something like 60% the mass of the shaft located on a point on the shaft outer diameter).

I believe that the 1x vibration of the shaft is caused by a combination of factors including shaft run out, bearing eccentricities and hydrodynamic forces in the pump which push it laterally out of its axis of rotation. All this is compounded by alignment issues during assembly and transportation of the pump. It is hoped to isolate and try define the effect of each individual force over the broad operating spectrum of the pump envelope. Our starting point has been to focus on shaft run out.
I know a common procure to check the balance of a rotordynamic system is to have the whole rotordynamic assembly dynamically balanced. I will also explore that option as a further quality control check built into the process; however the relatively low capital cost of the pump compared to our other products may not entice management to purchase a dynamic balancing system or sub contract the function. Once I have completed my analysis of the whole procedure I may have a better understanding of what is needed.

 

[electricpete]

Those type of calculations sound like something an OEM would do. As a user facility we don’t go that far into analysis but mostly use experience-based thumbrules such as 0.5 mils per foot.

I think the approach of calculating a deflection based on centrifugal force would be correct far below resonance in the “spring-controlled” region. Above first resonance I think you need your second option = modal analysis if you want to predict the change in deflection from static to operating above resonance.


=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[Patrick02]

As you suggested I think I better be safe rather than sorry and have a good look at modelling the super critical region. I will also take your field experience of 0.5 mils per foot into account.

The drawings for the shaft dimensions and specifications were developed by the company when designing this pump (which initially was designed before I was born) and we simply subcontracted them out to one of our suppliers. As I understand it after the years went by the company started re-evaluating suppliers and component costs and then simply outsourced the shaft fabrication to the cheapest supplier, who in this case makes a variety of base components for our company but does not specialize in rotating components. With the recent implementation of our Quality management program we are now re-evaluating this issue and considering better suited suppliers, manufacturing techniques and inspection methods since failures of the pump are costing the company both money in trying to make in field repairs in remote locations and consumer confidence in our product.

I will let you know how my analysis and the development of an inspection procedure work out.

Thanks again for your advice.

Regards,

Patrick

 

[dgallagher]

API670, 612 and 617 specs for the maximum allowed "glitch", or combined electrical plus mechanical runout is not to exceed 25% of allowed peak to peak vibration amplitude or 6 micrometers (0.25 mils), whichever is greater.

Bently recently updated their old runout appnote:

http://www.bently.com/articles/articlepdf/3Q05_Runout.pdf

 

[electricpete]

Runout will depend on how close you measure to the bearing (or Vee-blocks or rollers in the shop). Assuming for the moment you have no out-of-round condition, perfect shaft finish and the shaft is supported at two points, then the runout would approach 0 near the bearing/support and will be maximum far away from the bearing.

The 0.25 mils I assume applies to the accessible location for prox probe monitoring typically close to the bearing. In order to achieve that requires a good surface finish and absence of out-of-round, but measurement at that location doesn’t tell us much about shaft bow I don’t think (someone correct me if I’m wrong)

I’m pretty sure for a 10’ shaft you would not likely achieve 0.25 mils along the entire length.

I just talked to my pump engineer and he reminded me that in addition to 0.5 mils per inch, we also apply a maximum of 5-10 mils (depends on appliation) regardless of shaft length (becomes limiting for shafts longer than 10-20 feet).


=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[Patrick02]

Hi electricpete

Sorry for the delay and thanks for speaking to your pump engineer for me.

I have spoken to few people who deal with alignment and so on and they also recommend 0.5 mils per inch. One of them is coming around to our factory within the next week or so to give an opinion, which I will take into consideration along with other data I have gathered. Like you, I agree that there should be a limiting factor in the total shaft run-out.

In my opinion these long vertical spindle pumps should be classed as a different type of equipment to the run of the mill slurry pump we sell. Although I have been involved in the field only recently I am aware of that our pumps are designed for a rough and ready environment where they take a lot of abuse over their operational life. Most of our horizontal workhorse pumps are over-designed in that the safety factors on their casings, bearings and shafts are large. It is my understanding that they can withstand situations such as shunting the shaft during start-up to clear the pump of any settled sediment, although I by no means condone this situation as it represents bad operational practice and will lead to structural impeller failure after a time. They can also withstand abuse during maintenance. The impression of our products by our customers is that they are robust in the common understanding of the word. The design emphasis has been on the hydraulics of the pump and material specifications of the wet end components to prolong wear life. Thick steel base supports are typically installed as frames for these pumps to act as a vibration dampener. In the way these pumps have been designed and manufactured this has been the driving methodology behind them and I believe for our typical horizontal pumps it is a winning methodology. The vertical spindle pumps are far more delicate in that, as I explained earlier, they have very unforgiving rotordynamics. The way in which my company manufactures and assembles components for vertical spindle pumps needs to be reviewed if we are to gain a foothold on the problem. It is an interesting design challenge for me and the team I work with because the simple solution of adding a bearing support near the impeller of the pump is not currently feasible because of the corrosive and abrasive environments we operate in.

The shaft and the impeller are our main points of focus. We may stiffen the shaft or lighten the mass of the impeller or both. As I stated earlier, we are also looking at the way in which the shaft is manufactured. If we stiffen the shaft too much we run the risk of moving the first critical in the prime operating region of the pump. We are also looking at how effective heat treatment will be on out shafts, as some of our shafts are only machined near the ends and not across the entire length. We will only be able to make informed judgment on this once we have an effective measurement process in place.

dgallagher thank you for your link to the Bently Nevada article. I have studied it carefully. I have also been looking for API670 and similar documents relating to stuff as well as guidelines local manufacturers in other industries (such as power generation) use.

Regards,

Patrick