Dear all,

I would like to know when I found the natural frequency of pump support is 24 Hz and the machine running frequence is 25 Hz. Does it will cause the resonance problem by the 1 Hz difference ?

thanks

kk

The general rule is 10-20 % of the freq, so 25hz *.1 = 2.5hz, so you are close even on the conservative 10% rule.

It really dependes on the how "wide" the resonant freq is. If it is very broad, the system may be very sensitive over quite a range of frequencies in the area. But if it is narrow, it may drop off rather quickly once you get off the peak.

The broad or narrow of the drop will depend on the damping of the rotor/bearing system.
Sleeve bearing has higher damping than roller element bearing.
When the damping is high the drop is slow and when the damping is low the drop is fast.
In fact all the system has some of damping. The lowest problem for 1 Hz difference above critical speed is for shaft between roller element bearing.

Regards and have a nice day.

I agree with Martin - 1hz is too close.

Is the machine running? If it is, you can perhaps let vibration be your guide. Low vibration means you are probably not causing any damage yet, but even then in the future you may be in for trouble if the exciting force increases or the resonance decreases.

If this is a machine which hasn't yet been run, I'd say the odds of problems will be high.

Dear all,

After serveral tests, we confirm the pump support have the natural frequency about 24.5 Hz and closely to the equipment running speed(1500 rpm and 25Hz).

Does anyone know which ISO talking about the resonance ? becasue we try to use black and which to reject the pump support from the supporter.
(they think resonance is not a problem !!)

regards,

kk

How is the pump support, can you stiffen the structure, by welding extra braces to it?
The natural frequency of the support is
sqrt( k / m) in rad/sec, thus increasing the stiffness k will bring the natural frequency higher.

It is a vertical pump, a natural freq 24.5 Hz is found on the pump support from hammer test. Since the pump is running at 25Hz , high vibration at about 250 micron is found at the motor outboard bearing and directionally.
The supplier put about 60 kg additional weight on the motor top cooling fan cover to reduce the 1X excitation force and the a significant vibration improve to about 44 micron.

Although the vibraiton problem seem to be solved, my superivisors do not acceptble this method. They want the supplier to modifiy the support to increase the stiffness....

so they want me to write a letter to reject the method which put 60kg at the motor top.

Actually, I don't know how to reject....

kk

API 670 (Pumps) requires rotor critical speed separated 20% from operating speed.

Rotor critical speed is not in general the same thing as resonance measured by bump test performed external to the machine, but the same principle applies. (maybe someone else can comment further on the relationship between rotor critical speed and resonance by external bump test?).

At the following link http://www.update-intl.com/VibrationBook3m.htm it is suggested that resonances be detuned 20-25% from operating speed.

When we talk about critical speed of shafts/rotors we treath the shaft as a flexible element. If a flexible shaft carries an unbalanced rotating mass (rotor), the unbalance produces bending in the shaft (also called whirling). If the shaft is supported by two bearings, a rotating bow will be seen in the shaft. At a certain speed, known as critical speed the deflection becomes very large. This condition is critical for high speeds (jet engines, turbo machinery etc..) at many 1000 x rpm, were a small deflection can destroy the fan blades.

In the case of your pump the stiffness
k = 3E/L * pi*d^4 * 1/64

and the mass m would be the mass of the impeller


With the bump test, the structure is considered a flexible element..

By the way, if they piled 60 kg on the top of the fan cover, I assume that that this extra mass is not rotating, we are not dealing with critical speed.

They put 60 kg on the rotor cover, it is surely not a 1/2 hp electromotor

Dear Jazzchan, you did put a specific question: How can you use the support of an ISO standard to reject a resonance. The answer is - like we use the stadndard over here. There are two conditions discussed in the standard. A stiff foundation with resonances above the running speed and a soft standing machine with the resonances below the running speed. This is indeed a risky simplification, but if you use it right off, you can claim that a standard for the resonant condition in between does not exist, hence is not allowed .....

Your project people do not want to accept a soft machine, i.e. a machine with the 60 Kg weight on top. Based on that, you are asked to write a rejection claim. You can do that politically, but not technically. Look only to a car, does that mean that all foundations of car engines must be stiff to the ground? :-)

Joke aside, the pump will survive perfectly for 30+ years if you just avoid resonance. You should not accept the mass as a principal 60 Kgs as such. Instead look to the position of the resonance now. Is the speed 1500 RPM (Asynch motor, probably approx. 1460 RPM), you should not allow the resonance above 1275 RPM (some 21 Hz). Maybe more is needed.

I have made many many cases in both softer (more mass) and stiffer (supports to the floor and so on), and I have never observed a real difference in operating time to next revision. Resonance free runs well. Using a mass on top can be a good way to avoid extra handling work during revision work. Nothing to align and setup. Just add the mass as a thick fan cover on top of the motor.

Pete cited the API spec and it is good. Many good comments were made.

I think it is standard practice to shoot for 25% away in the design stage w/API spec as the rule. However, that's not always what you encounter.

API 670 is on Machinery Protection Systems.

API 610 Centrifugal Pumps for Petroleum, Petrochemical and Natural Gas Industries, 10th edition is the same as ISO 13709.

Without copying the entire text one section reads

quote:

5.6.15 The rotor of one- and two-stage pumps shall be designed so its first dry-bending critical speed is at least 20 % above the pump’s maximum continuous operating speed.

The dry critical would be the pump without fluid. Thus, the actual critical should be much higher.

There are others in the document. Other seperation margins in API take into account the damping.

You're right - I meant API 610.

Why would fluid increase the critical speed?

It's not so much the fluid as it is the effects of flow (pressure gradient) across the seals. The seals act as bearings.

Pete, just an example from real world: A 10 stage charging pump (also used as safety injection in nuke places) has critical speed in balancing machine at 1780 RPM. When the two wear rings on each stage (inlet nozzle and back side ring) and the balancing drum all run in water the critical is at 5460 RPM. I used old critspd.exe from 1987 to check and it verified it perfectly. This ends up in a question: Should the vendor balance the complete rotor in air running operation speed at 4420 RPM and hard balancing machine (A Schenk Trebel) using just impeller 1 and 10 to adjust dynamically? Or should he say this rotor will be taking on a banana shape at full operation some 25 percent below the critical speed so I should balance it in three planes (impeller 1,5 and 10) running 25 percent below the air critical speed, taking dynamic error in impeller 1 and 10 and static in impeller 5?

It wouldn’t be unusual to assembly balance a machine like this (if I picture it correctly in my head). That is balance the components and balance the assembly as the rotor is stacked only on the added components.

However, to a assemble the entire rotor and only balance on two planes introduces some risk if you are only 20% below a critical. But 20% separation is a good margin, helping to mitigate the risk.

High speed balancing should produce the best running balance. Again one should use balanced components to start. One of the downsides of using a high speed balance is that if one replaced a rotating component, one should perform another high speed balance. There are fewer places to get a high speed balance compared to a low speed balance. If the rotor is assembly balanced, it may be possible to repair it at location without the need for a high speed balance facility.

If one takes all the static component out of the middle in a low speed balance of a flexible rotor, it is possible to over balance the banana mode. Stack balancing should work better unless this is not possible.

Arne - That's an interesting example.

The speed sounds like what we call our Centrifugal Charging Pump

That is a pretty dramatic example of an increase in critical speed 1780 on balance machine to 5460 installed.

In your critspd model, did you add a bearing (to model the wear rings) between every stage? I wonder what the first critical mode shape would look like.

I can see that situation poses a lot of challenges. High speed balance at operating speed 4420rpm woudln't create the same mode shape as during operation at 4420rpm. It seems the only options are stack balancing as suggested by Bill or some kind of creative approach based on knowledge of the actual mode shape. It seems to me like your idea to balance 75% of air-critical speed would work if the expected first critical mode shape is bananna shaped. But then again it's seems hard to predict how the machine would act with all those wear-rings acting like bearings (it wouldn't make sense to try to balance a 12 bearing machine in 2 or 3 planes). What was the actual balance approach taken?

Bill mentioned overbalancing... what did you mean by that.

This message has been edited. Last edited by: electricpete,

Originaly posted by Electricpete

quote:

Why would fluid increase the critical speed?

John Vance show an experimental research related to, with a same rotor running in air and in kerosen (I am not sure because I have not the book front of me). The thin is related to the damping added by the fluid when the viscosity is higher.

Let me explain how balancing these charging pumps is done here since some 25 years. The shaft is burred with a pneumatic hammer behind a plastic head to get straight inside 1.2 Mils TIR. The shaft has been ground at each impeller seat (10 pcs) to be stepwise 3 Mils thicker from the nut stage 1 to 10. Each impeller has been numbered and the hole ground to be a slight hand push fit at around 60 degree Celsius at its own final position. Each impeller is balanced using a precision expansion arbor shaft (same as when rigging for grinding in a grinding machine). Balanced to near zero and including indexing 180 degrees to remove any arbor error.

The rotor is now built with half keyes when needed with two impellers at a time, each step trimmed in two planes adding clay, not grinding weights in this stage. When rotor is complete, essential dynamic (couple) unbalance is trimmed in impeller 1 and 2 and 9 and 10. Any static unbalance is trimmed over impeller 4-5-6. Smooth, only outside, well spread, no edges of course. Rotor is stored hanging from coupling end. During revision, when coupling is there to install, it is added and only what it has caused as unbalance is trimmed. Now the pump can be built, old pump is taken apart and the balanced pump rotor is built with stator parts to a pack that is installed. This procedure will minimize the errors from taking a balanced rotor apart and assemble again without being able to trim balance it when it is inside the pump barrel, thanks to excellent repeatability of rotors radial positions. It happens that the balance will be trimmed a slight bit in final operation using tiny Allen screws in a locking ring next to the mech seal.

The procedure may sound tediuous, but the result pays its own way. A drastic reduction of revision hours for all sorts of reasons. Only remaining problem is a bit of erosion in the mech seals that nobody saw before since the seal did not survive enough to get it.

The fact that the critical speed will raise with the addition of liquid is well explained. I think the first time in a papar by Mr Lomakin, Russia, before 1964. Dr Dernedde at KSB, Germany published a paper on the subject 1964. Maybe somebody out there has a better documentation on this. Yes, even the old critspd.exe shows the effect nicely when adding "water bearings" at each wear rings. The effect has been shown to be real in many different rotors over the years. Water is one thing, attraction forces in a motor/gen airgap is another interesting application.