Load sharing on 3ph motors 5

[Grimo]

We have 2 geared motors that are driving into a common load; the actual application is a CCD rake drive but it's easier to describe if you imagine it as a cement kiln with 2 geared motors driving the girth gear instead of one. Both motors are the same type and make (Weg), same power (5.5kW), same speed (1440 rpm) and same current rating (10.8A) on DOL start (no VSD). The drive train starts with vee belts from the motor to a flat helical box which is shaft mounted onto the input of a worm box, the worm box output goes into a set of spur gears, and the output shaft from the spur gears holds the pinion that meshes with our equivalent of the girth gear. Overall reduction from motor to the pinion is around 800:1. Each drive has a strain gauge fitted to the rear of each worm shaft to measure torque through movement of this shaft as the load increases; this signal protects the system from overload by monitoring total torque but the individual signals are also compared for torque imbalance between drives and trips the system if the imbalance exceeds 30%. My question is, how can an increase in load cause one motor to see this increase before the other when they're connected by a girth gear that has no "give" in it like a conveyor might? Surely they would see it at the same time, yet we are plagued with constant imbalance trips. I suspect these trips might be more related to problems in the mechanical drive train or the strain gauges needing calibration (this takes about 10 hours to do and it's difficult to get a production release for this amount of time). We presently have no means of accurately measuring motor work current as an indication of torque (which could be compared with the strain gauges to see if the trend is the same) but I intend to fit units to do this in the next few weeks. Am I on the right track here? Any comments would be gratefully received.

 

[waross]

Motor current is an indication of torque rather than a measure. However, a 30% torque imbalance between identical motors should result in a noticeable difference in motor current.
Is it possible to interchange the change gages without calibrating them? If the problem follows the strain gage, suspect a faulty strain gage. If the problem remains with the drive, suspect mechanical problems or motor problems. Note, if a motor is failing it may show up as an increased torque load on the opposite drive.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[Grimo]

Thanks for the response, Waross. Firstly, yes, I have tried swapping each strain gauge to the opposite controller and the problem does move, which is why I strongly suspect calibration as the main culprit. To do a cal we have to strip the front end of the drive train to gain access to the worm input shaft and fit an hydraulic ram to simulate worm shaft movement at a known force - this takes time that the production team freak out over, and they pressure our maintenance people into simply "tweaking" the gauge controllers to bring them back into balance. The current monitoring I'm looking at differentiates motor "work" current from magnetising current by the different power factor between the two and uses an algorithm to sensibly linearize the work current component to indicate torque via a 4-20mA output - I don't believe this will be accurate enough to act as our primary form of over torque protection but it's probably more accurate that the uncalibrated, "tweaked" system we have now! I wanted to use motor current as indication only that an imbalance did exist as, like you, I'd expect a 30% differential to show a measurable difference in motor currents. We have 6 of these CCDs with twin motor drives and they all suffer from the same torque differential problem so I don't think it can be put down to a faulty motor and/or strain gauge on all of them. I'm still struggling to get my head around how a load induced imbalance could occur anyway; we basically have a revolving drum with a girth gear around it being driven by 2 identical motors and if more product is heaved into the drum, thus increasing the load, how could this affect only one motor instead of the two simultaneously?

 

[itsmoked]

How can a 30% difference in torque only be qualified by the current? That represents a 30% difference in hp! That means a large change in real power.

I can not believe you can't do a more than adequate job using more reliable power measurements than the present problematic mechanical headache. I would do it with either current or power and totally ditch those strain gauges.

Having real power monitoring of those valuable motors would also dish up other benefits lacking from strain gauge readings.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[waross]

If you are going to develop a 4-20 ma signal, have you considered using a watt transducer. Watts will give you a more accurate indication of the power and torque that the motor is developing.
Another option that may work is to run "A" phase from each motor in opposite directions through a CT and monitor the difference.
I suspect that your instrumentation is mis-applied.
If you are having intermittent trips on several machines with no indication of imminent failure of either motors or gear trains then perhaps you only have an instrumentation problem.
I would look at the belt drive in relation to the strain gage to see if the strain gage is affected by the belt tension.
I would also inspect the pinion gears and girth gear. If the set up clearances are not the same, may this affect the force on the pinion gear? Also, is there any possibility of dirt or some other substance building up on the pinions unequally and affecting either the torque requirement or the force on the pinion gear? Is the mechanical set up of the strain gages such that a slight movement or misalignment of the the hardware will cause a gross error in the measured strain?
I have just re-read your post. I suspect that the instrumentation is badly applied and is not holding its calibration.
Single phase watt meters on the same phase on each motor should give you a good indication, possibly good enough for primary indication. If one motor is pulling substantially more watts it is a good indication that there is a mechanical problem and the calibration of a good watt meter transducer should be stable. Again look for spacing, alignment and crud build up for a possible cause of unequal torques.
What is the normal range of motor currents on these motors?
What's a CCD rake?

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[CJCPE]

Here is a link to a good electronic load monitoring device that has current transformers built in.

http://www.loadcontrols.com

I am not connected with that product and I don't really like posting links to them without also posting a competitor, but they are the only one that I am familiar with.

 

[sreid]

Do you do any filtering of the load imbalance signal and how long do you delay before declaring an imbalance? From you posts it would appear that the imbalance appears when there is a load transcient. I can visualize a "Ping-Ponging" of load between the motors due to "Step" load changes; inertias and drive compliances (or dead bands) forming a spring-mass system that takes some time to damp out to steady state.

 

[zekeman]

I have some comments and questions on the description
1) what kind of motors are in use? Are they matched induction or synchronous? Synchronous could be a problem.

2) you state that the 2 motors are solidly coupled but in fact they are NOT. You have (correctly designed) belt drives which inherently slip and cause differental slip in the motors (assuming synchronous motors)and spell your major difficuly; torque is proportional to slip for near synchronous speeds and belts clearly invite differential slip and will immediately result in torque mismatch.Now for a sudden increase in load, the less compliant motor-to-load coupling will momentarily take on a greater share of the increased load, possibly increasing the torque disparity.
Startup adds differential backlash as a third problem for torque disparity.

Acordingly, instead of asking why, you should be thinking of a redesign, since no amount of improved accuracy of your torque measurements will correct the underlying problem.

 

[Grimo]

Grateful thanks to all for their assistance here, the picture is becoming more clear in my mind now. To address a couple of specifics, CCD means Counter Current Decantation - a device that's also called a thickener. There is no filtering or damping of the torque signal mainly because the system is mechanically very sluggish - the total speed reduction leaves the rake turning at 1 revolution every 5 minutes in a slurry that has a smooth and even consistency; there are no lumps or other obstructions to cause sudden transients and in hours of recording the signal I've never seen one. Lowering the rake further into the slurry will certainly increase the load but the rake lowers at 25mm per minute so the load increase is gradual, not rapid. Even the torque differential trip doesn't happen as a transient, it can take a day or more to slowly climb from negligible torque differential to the 30% trip point.
The motors are standard 3ph squirrel cage induction motors but supplied as a matched pair from the manufacturer. I haven't been able to dig up any detailed performance information for them but feel it is safe to assume the load/slip charactaristics would be sensibly identical for them both. Maybe we do need to look more closely at the triple "B" section drive belts - we've watched the belts as differential trips have occured but neither seen nor heard any indication of belt slip, although that doesn't guarantee it's not happening.

 

[sreid]

Could one of the motors be going "Single Phase?" A slowly rising resistance in one feed line phase?

 

[itsmoked]

It certainly sounds pathological if just this one system has this issue. You really need to monitor both or at least each motor's power consumption during one of these cycles to see if this is a motor issue verse sensing issue.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[zekeman]

It does possibly point to relative slip in the belting.
To test this,I would try to accurately monitor the output speeds of the belt driven shafts for speed differntials.If this theory is correct, then:
The percent slip differential would be the per cent differential in torque vs the nominal torque, since torque is almost linearly related to slip.

 

[zekeman]

Correction:
Monitor the motor shaft speeds, not the output speeds which must be equal.

 

[waross]

Slipping belts may be an issue if a heavy load is suddenly applied. If you are getting 30% imbalances under steady state condition due to slipping belts you may often locate the problem by the sound of the squealing belt. There may also be smoke.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[zekeman]

A general purpose motor at full load slips 3-5%; we're talking 30% of 3-5% or a differential slip of 1-2% between the motors, about 15RPM, not enough "see" or to cause smoke.