Steel rolling mill chatter marks

[tensor47]

I have been asked to investigate "roll chatter" on a "skin pass" steel rolling mill used for final cold rolling of previously hot rolled, 1 mm thick material for car bodies.

The mill has 2 work roll sizes (450 mm dia for 150 tons and 650 mm for 400 tons) and 2 backup rolls both 1100 mm dia.
Each work roll has a gearbox and 750 KW motor driving it.

3rd and 5th octave type chatter is documented and is speed sensitive.

This chatter problem occurs at all speeds and was not noticed at commissioning but after 2 years it has grown to a quality control problem easily visible but difficult to measure.

The backup rolls show chatter marks with a pitch of 15 to 25 mm at the end of a 30,000 ton run but there are no marks on the work rolls at this time.
The severity of the marks increases with speed.

Roll grinding is said to be good and the chatter problem is there from the start of a new roll set.

The work roll drive shafts are universal joints but the marks occur vastly more often than 2x work roll speed.

Any thoughts, anyone.

Cheers

John

 

[PeteDB]

Do the chatter marks coincide with tooth mesh frequency of any of the gears in the gearbox? It may be you have a dynamics problem which the tooth mesh frequency is exciting.

 

[tensor47]

Hi Peter,

Thanks for your reply.
There is a single reduction stage, 35/62 and then a dual output stage with 54/54 teeth.
For work rolls 450 mm dia these 2 stages would produce marks at 26.5 and 23.1 mm.
For work rolls 650 mm dia these 2 stages would produce marks at 37.81 and 32.94 mm.
Both these combinations (and half values) are in the range 15 to 25 mm we see on the sheet and backup rolls, but there are no marks on the work rolls.
Work and backup rolls are ground in pairs and the surface speed ratio changes, so one would expect that ant gear marks on the backup rolls would average out to nothing.
The work roll max speed is 15.12 Hz and with 650 mm dia the backup roll max speed is 8.52 Hz so they are quite fast for such large masses and intertia's.

The simplest fault would be that the backup rolls have the fault on them after grinding but I'm told this is not the case, and that the problem is there from the start of new rolls. So it's all a bit confusing.

 

[ivymike]

I'm told this is not the case
It's hard to find a solution if you're narrowing your search based on misinformation. I would insist on reviewing measurements (data) for the rolls before simply accepting that the simple answer is incorrect.

 

[Hatch]

Its been years since I did Chatter investigations on a 4 stand Cold Mill. There are many conditions which cause the mill to go into Chatter but it takes time and measurements to sort it out.

Some observations and comments I have are:

1) Its interesting that the work rolls do not show chatter marks but they appear on the backup rolls and the steel strip. You mentioned averaging out but what I've seen in the past is overlaid chatter marks on the work rolls.

2) Check the backlash/free play on the universal joints. It can cause much higher excitation frequencies than 1X or 2X.

3) If you spare work roll drive shafts (spindles), install them and see if it improves/solves the problem.

4) Check your gearbox for worn/damaged gears or excessive backlash. These cause torsional oscillations which can end up as chatter.

5) Are your work/backup rolls within tolerance? They do wear out after repeated grinding.

6) Listen to the drive motors during chatter. Can you hear any unusual sounds. On our mills we used to hear the mill in chatter. Torsional problems can cause chatter.

7) Change the roll grinding profile but this requries expertise to maintain product quality.

8) Try a different lubricant or increase the lubricant spray rate and flow.

9) There used to be Companies that provided Chatter detection devices for Cold Mills. They might be able to help.

 

[tensor47]

Thanks very much for your replies.

IvyMike "I'm told this is not the case"
The potential job is in India and I'm relying at this point on what they tell me about the rolls, but like you, I'll leave nothing to chance if I get over there.
To date they have sent end to end profiles to show they are within tolerance but no circumferential data.

Hatch
1
Overlaid marks, interesting !
There are 2 visual repeat lengths on the backup rolls.
I once saw a Calcine mill (large cylinder with 2 metal circumferential tires supported by 4 wheels) driven by a huge chain, and also with foundation natural frequencies activity.
There were oddly spaced scallops on the tires and measurements at the wheel axles only showed up the natural frequency of the foundation.
Holding an accelerometer lightly against the tire as it rotated showed frequencies proportional to both the chain pitch length and the length resulting from the natural frequency.
I still think that if there is only 1 frequency and it’s not synchronous with the roll it will average out.
I’m proposing a tacho on each roll to do some time sync averaging and phase measurements.
2
Yes I agree that there can be a string of harmonics increasing with any wear.
But it seems near impossible to get a long term integer relationship with rolls that are regularly ground and I suggest again averaging will never let them get started.
3
They do have 2 different diameter sets of work rolls.
If the rolling requirements allow it I’ll certainly get them changed as this effects the compression Vs distance into the mill.
I once worked with a guy who knew a lot about continuous rolling of white metal bearings with steel backings, and to get the “curl” of the bearing to suit the journal diameter, he used different work roll diameters.
He spoke about chatter and said there was a lower limit of roll diameter at which point the surface friction was not great enough to pull the material into the nip, given the rate of compression that the smaller roll radius caused.
Do you know anything about this ?
4
Yes, I will do this.
5
Are you saying there is a minimum diameter that you can grind down to ?
Is this to do with hardness Vs depth in casting etc ?
The lengthwise grinding profile they showed was within tolerance.
6
Very interesting !
Torsional vibrations usually go unnoticed unless there are gears to produce tangential forces that cause the vibration we measure.
So from surface speed and diameter of the backup rolls I may be able to see that frequency and harmonics in the gearbox as non synchronous ( referring to 2), as well as the tooth meshing frequencies which are synchronous.
These frequencies, depending on work roll diameter used and allowing for 15 and 25 mm spacing fall between 1x and 2x gearbox tooth meshing frequency of 817 Hz.
So they should be audible at the mill as well.
And with no slip the relative ratio should be constant with mill speed changes.
I was thinking of strain gauging the drive shafts with radio off the shafts, but maybe this is not needed.

New point.
Also, there is a possibility that the work rolls can translate or flex in the horizontal plane as the backup rolls only resist vertical movement.
I was thinking of mounting 3 prox probes on a stiff beam to measure the work roll motion at each end and in the middle to check for this possibility. What do you think ?

7
Beyond the present scope of work I think.
But I do intend to use a prox probe to measure the circumferential shape of the backup roll while still in the grinder with the grinding wheel stopped. I can also do a sync average to look for fluting in the grinding process.
If there are marks here a the right spacing I may have to do some modal work on the grinding wheel pedestal.
Do you know anything about this ?

8
Do you mean a lubricant that is sprayed onto the plate as it goes into the mill ?
If so seems similar to the roll diameter issue (3), surface friction !
To stop the chatter do I reduce the surface spray volume and/or viscosity ?

9
I’ve checked out their papers and they seem to provide band pass filtering for 3rd and 5th octave resonance faults.
I also notice an active system that varies the hydraulic pressure in real time right on the cylinders to neutralize the chatter.

However, according to the client, this problem has been slowly growing with time, and increases severity with speed which does not sound like a resonance problem.
More a geometrical problem which increases it’s force with speed.

Cheers
John

 

[Hatch]

As I said before, its a complex situation and not normally solved in a few days.

Items 3 and 5 are related. In our mill we found that older work rolls were consistently involved when we had chatter. One solution was to use all newer work rolls when we had chatter. For some critical products, only new work rolls were used. We also found that chatter occured more frequently at certain times of the year (temperature?). Certain products were also more prone to chatter. In the end, a combination of maintenance techniques and rolling practice kept incidences of chatter to a tolearable minimum. Gearboxes were rebuilt every 4 years (1 stand per year), older work rolls were not used on critical products, roll grinding qualtiy was monitored carefully, rolling speed was reduced a bit, experienced rollers (foremen) were in charge during critical product rolling, roll profiles were changed to a chatter resistant profile. I don't know much about the chatter resistant profile. A fellow specialist did the roll profile modelling and set it up.

Item 9, Resonances are always present. They may never be noticable until a certain set of conditions occur. Therefore I would not rule out a resonance.

I guess you could look into horizontal deflection (movement in the plane of the strip). However, when you consider the stiffness of the work roll and the much lower forces involved in that direction, then its a low probability issue. I now recall that work roll chocks (bearings) were changed and it showed improvement. It could be that bearing wear is another trigger.

Hope that helps, good luck on your trip.