Medium voltage soft starter - commissioning

[ters]

The SSM series 6kV ABB soft starter is used to drive a 3.5MW asynchronous motor, which drives a large fan. The vendor seems to be unable to provide commissioning assistance in a timely manner, so the Client may have to commission this using its own resources.

The primary soft starter configuration starts with a disconnect switch, fuses, contactor and thyristor modules. There is also a bypass contactor.

On the secondary side, the equipment uses an ABB controller and an optional Multilin 469 motor protection. Optional – since the soft starter electronics seems to have a number of protective functions already built in.

We have an idea how to set the protections, but the controller also needs all kind of settings to be applied. The process is relatively steady and the maximum number of starts is probably less than three per day.

Could anyone provide some general guidance as how to set things like start mode (there is several to choose from – dual ramp, custom curve, jog voltage, etc), starting torque, acceleration (ramp time), current limits (min and max), initial voltage, etc. Thank you.

 

[panter]

Hello, Ters!!
I have attached picture of oscilloscope recording one of many unsuccessful start and it looks very strange. It looks like that tiristor starts in the middle of the cycle and I think that we have problem with bad synchronization between control voltage and 6kV network.
Greetings from Elektrika USSSerbia.

 

[Zogzog]

OK, how do you do the quote thingy?

 

[rbulsara]

zogzog:

Look up "Progress TGML" below the reply window.

 

[Marke]

Hello panter

Those waveforms look perfect to me. That is exactly what I would expect a soft start waveform to look like.

ters

Your description is certainly one of insuffient torque, but that does not preclude another problem as well.

When you are attempting a start, are all three current balanced and stable. There should not be a big difference between the currents and they should not be flickering.

To start a motor and load, you must develop sufficient torque to overcome the work output and to provide sufficient surplus torque to accelerate the load inertia to full speed in a reasonable time.

The minimum torque requirement is a function of the load characteristics and the current to develop that torque is a function of the motor characteristics. For more information have a look at my web site at

http://www.LMPhotonics.com/m_control.htm

I would expect that for a motor of this size, you are going to need a start current of 400 - 450% to start a fan.
I expect that if you do the calculations, you will find that the torque available from this motor at 300% is very low and will not start many types of loads.

I would try for a 400% current limit and a 3 second ramp time. I would expect that at this setting, the acceleration time could be in the order of 30 seconds, but that is very dependent on the motor characteristics.
You may need to increase the current limit if the motor does not reach full speed.

Best regards,

Mark Empson
L M Photonics Ltd

 

[Zogzog]

Look up "Progress TGML" below the reply window.

Like this? Sorry not trying to hijack or anything, you can all ignore me.

 

[rbulsara]

Yup! You can see by yourself! Now you can also use the "preview post" feature to see what you will get, rather than this method.

 

[ters]

Mark, thank you very much for a detailed reply. Mr. Panter is on the same project and equally desperate to find a solution, except he is the Owner and I’m a consultant involved in this since recent times.

It seems to me that the issue is exactly what you and some other folks above said, insufficient torque, due to insufficient voltage.

The motor is fed from a high impedance 110/6kV transformer, which is fairly big, 13/20MVA ONAN/ONAF but the impendence may go as high as 20%, depending on the tap changer position, and the best we can hope for is 11%.

In addition to the TR high impedance, there is also a short circuit current limiting reactor on the 6kV side + some 300m cable to the motor (which is oversized).

So today we started the motor at 470% FLA, and it accelerated much better (reached 1000 rpm), but the soft starter start time elapsed after 40s and it tripped off.

During the start, the voltage dipped initially for about 15%, then after a few sec it recovered back for some 5%, dipped again where it was before, and so on, recovering after each swing slightly higher.

It seemed that the voltage dips/recoveries were initially following SCRs’ voltage ramp steps, although swings continued after the ramp time elapsed. After each voltage upward swing (partial recovery) the motor would seem to catch some more speed, etc.

It also seemed that during the dips, the transformer voltage was starting to collapse, perhaps not only delivering insufficient and unstable energy to the motor, what made the motor to slow accelerating, but possibly having somewhat distorted voltage forms as well, what contributed to increased motor growling sound.

We used longer ramp time (20s) as the voltage seems to collapse less severely with a prolonged ramp.

We then repeated start with a shorter ramp time (10 sec), but even though the current was theoretically limited to 470%, what is about 1800Amp, we tripped off the upstream protection, which Mr. Panter set to 2000Amp. Not sure why it tripped, unless what the SSM measures as 1800Amp is not exactly what the relay sees.

I also have to note that Mr. Panter is not particularly generous when it comes to setting upstream protections higher . He agreed to change the current setting higher, but he then shortened the time, so I2T is still about the same...

So, if the issue is the system impendence, we are unsure what the solution is going to be, but we are still hoping that some of you, experts, may post that magic resolution here!

 

[Marke]

Hello ters

Ok, the time has come.
I think that to progress, there should be some engineering done. I prefer to do this before the equipment is purchased.
It is possible that the motor is just not suitable for the application. I have found plenty of those!!
Try to get the speed / torque curve of the fan with the dampers closed, and the speed torque curve of the motor, and the speed current curve of the motor.
Determine what the full speed of the fan is and what the full speed of the motor is. The last thing we need is the inertia of the fan and motor.

With this data, it is possible to calculate what the starting current and time should be.

Best regards,
Mark.

Mark Empson
L M Photonics Ltd

 

[rbulsara]

Better Marke than me!! Beat me to the punch.

Also a motor transient analysis using a software can be done, provided necessary data is available.

One thing worth doing, if there is a means to do it is to try a full voltage (DOL)start (or close to it) and just see what happens. If still motor does not come to full speed in reasonable time, it is time to do some re-engineering.

The last data still does not say what was the maximum voltage reached at the motor terminal at 1000rpm and what is the rated rpm?

 

[ters]

Thank you very much gents. Mark, I will post all curves we have, hoping that you can provide some comments. We did some calculations during last few days, and it shows that the acceleration time at 500%FLA should be 36s. We did not achieve that, possibly because the calculation didn't factor in limitations related to system impendences and voltage instability (swings) during the start, or we possibly made some other errors.

Rbulsara, at the moment we cannot do a DOL start. It would require about 2300Amp and for now we are not is position to try it yet due to upstream limitations, but eventually we will probably try that too.

I posted nominal RPMs as 1485 somewhere far above. I'm not sure what was the voltage at the motor terminals when it reached 1000rmp since at that moment I was monitoring what was going on with voltage before SCRs, but it if I remember correctly, it was about 5.2kV, according to others.

 

[ters]

Here is some more info on the motor and fan. Hope that Mark or anyone esle could provide some comments.

The fan impeller weight is 15,000 lbs, and the Impeller moment of inertia (WR^2) is 66,000 lb-ft^2. The motor rotor weight is about 7000kg and the diameter is about 0.9m.

 

[Marke]

I have worked off the curves and it would appear that provided that everything matches the curves, then the minimum start current is in the order of 280%, but at 300% current, the start time will be 180 seconds.
At 400% start current, the start time will be 60 seconds, at 450% start current, the start time will be 44 seconds and at 500% start current the start time will be 34 seconds.

So from the information, it would appear that provided the equipment matches the data, it is possible to start this fan with the dampers closed at 500% current in around 34 seconds.
Note : This is right on the specified limit for that motor.

From your descriptions, it sounds as though the torque required is higher than is shown on the curve supplied and if this is the case, you will probably need to be around 450% to start the load and the time could be several times longer than the calculations.

Best regards,

Mark Empson
L M Photonics Ltd

 

[Marke]

450% and 500% start current curves can be seen here :

http://www.lmphotonics.com/eng-tips/3500kw.htm

Best regards,

Mark Empson
L M Photonics Ltd

 

[panter]

Mark,

Thank you very very much. Looks like results of your calculations are about the same as ours, we also came up with similar numbers, such as close to 3 minutes for 300%FLA and 36sec for 500%.

However, tests show that we are not getting there, i.e, today we repeated three tests @ 470%FLA with different ramp times, all resulted with the same outcome, it took about 50 sec to accelerate to 1200rmp, which is close to the “hump” and then the soft starter times out.

This (50 sec) is probably a severe abuse for the motor (rotor), since we reach 100% thermal capacity (according to Multilin settings) after one start.

One of the reasons why there is a discrepancy is between calculations and the reality may be that that calculations don’t factor in for random voltage swings and dips. Every 3-4 seconds voltage dips and recovers partly and so on, what I believe reduces the torque and extends the acceleration time. The calculations also may not factor in for the distorted wave form.

Also, people running the test insist on a long ramp time (they started with 20 sec but shrinking it down to 10), in order to avoid the initial larger voltage drop (if ramp is only 2-3 sec) which creates subsequent swings with larger amplitude and may trip on O/C, but the consequence of this is that for about 15-20 sec almost nothing happens, the motor starts moving but hardly reaches 100rmp before 20sec.

They also insist on current ramp rather than voltage although the manufacturer did not recommend it nor there seems to be a big difference in essence, you control the current controlling the voltage or another way around.

Finally, we cannot be sure about accuracy of the supplied curves, especially for the fan, nor the dampers can be considered fully closed, there is some leak.

Speaking of the motor, I'm not an expert on motors, but it seems to me that this unit is of very modest starting efficiency – relatively low torque @ high current. Do you concur with this?.

I’m not sure how it affects the starting torque (does it make it worse?), but the motor is of an oversized frame, it is 800Amp frame, while the actual current rated is below 400Amp.

Just to make the first successful start, we may try to jog the motor before the start or kick the bypass in after about 30-40 sec, but for normal operations it doesn’t seem we will get out of the puzzle without supporting the voltage in some way, which would be possible paralleling two transformers during the start (provisions exist)or supporting the voltage from a local plant generator, which would also be possible with some system reconfiguration. Plant people for now are not agreeing with any of this, but eventually there may not be much other options left.

 

[ters]

I wrote the above, but since Panter was logged on this machine once, since then, for some reason, his name keeps remaining logged in by default any time I visit this page (even though he was logged out), unless I log him out again and log myself in, what I forgot to do this time...

 

[Marke]

Hello ters

I would recommend that for this application, you use a fast ramp, not a slow ramp. Usually around 2 seconds is fine for allowing monitoring equipment to track the voltage dip. If you use a slower ramp, you are prolonging the start time and dissipating more power in the rotor.

The sagging voltage should not make any difference to the performance during start as the voltage applied to the motor is being regulated by the soft starter. A drop in the supply voltage just means that the soft starter causes the SCRs to conduct for longer.

A start current of 470% is going to lengthen the start time. If the start time at 450% is 44 seconds and the start time at 500% is 34 seconds, then the start time at 470% will be between the two, say around 40 seconds. If you have a ten second ramp, then it should take around 50 seconds provided all things are as quoted.
Looks to me like things are very marginal all round!!

I would suggest that you have a good look at the mechanicals and see if there is any way that you can further reduce the start torque. Belt tensions, motor alignment are all factors that can have an influence.

The start characteristics of the motor are determined by the rotor design. It is possible to change the rotor to get very different results, but in a motor of this size, that is not easy.

kick the bypass in after about 30-40 sec

Make sure that the bypass contactor is rated to DOL start the motor. Manufactures often use AC1 rated bypass contactors and these can not carry the DOL current.
If you bring the bypass contactor in early, the voltage will go to full voltage and the current will go to full voltage current. This will increase the voltage drop and at that point you may as well just start the motor DOL unless you bring the bypass contactor in around 90% speed.

this unit is of very modest starting efficiency – relatively low torque @ high current.

This is not uncommon on large motors. I would not suggest that this motor is bad, probably more average.

Best regards,

Mark Empson
L M Photonics Ltd

 

[ters]

Mark,

Thank you again for clearing some of my confusion. We will keep trying. However, as I said, I'm not particularly optimistic. At the present, we have a 13/20MVA transformer reserved just to do this testing. In normal operations, however, that transformer will have also other things to do (feed other loads) and given its high impedance, things don't look particularly promising - voltage sags will be higher on a transformer already carrying some load. But we have a new larger transformer coming, which should make a difference, but that one will not be commissioned that soon, will take several months.

 

[Marke]

voltage sags will be higher on a transformer already carrying some load

Don't think so, the change in voltage during start is a function of the transformer impedance and the change in current. If you have a start current of 2000 amps and it drops the voltage by 7% during start with the transformer unloaded, it will drop by the same amount when the transformer is partially loaded. The end point voltage will be lower due to the additional load, but so will the start point voltage.

Best regards,

Mark Empson
L M Photonics Ltd

 

[ters]

My thinking was about voltage collapse in case of the overload. Say if the transformer is already working at 30% capacity and then we start the motor and pull additional 100% of its capacity. Since this is an overload condition, I imagine that the voltage collapse may not be linear any more? Even if it is still linear, I'm not sure that other loads on the same bus will be able to easily cope with substantial voltage swings which would last up to one minute.