Induction Motor Auto-restart/Ride through 2

[gordonl]

I would appreciate thoughts on the following. I have four 350HP 2400V motors fed from a single bus. I have two seperate supplies which are usually, but not necessarily in phase. All of the motors are normally fed from one source, with the the second as a standby (breaker open). When the breaker feeding the normal supply trips, the second supply automatically closes to re-energize the bus. The breaker contacts are chosen to gurantee open of one supply before close of second. There are no capacitors or synchronous machines on this system, only the induction motors. How long should I wait before re-energizing the bus to avoid a double inrush situation?

Thank You,
Gord

 

[jbartos]

Suggestions:
1. There was one posting in this Forum that suggests that the motor terminal voltage shall decrease to approximately 30% of its rated terminal voltage. This would mean that you would have to look at the slowest terminal voltage decaying motor and then close the other supply circuit breaker.
2. Another approach would be to install a fast solid state transfer switch (2milliseconds to 4milliseconds or so). Motors would not feel this fast bus transfer.

 

[electricpete]

Gord
I put in a lot of words on the subject in the following thread:

http://www.eng-tips.com/gviewthread.cfm/lev2/11/lev3/47/pid/237/qid/9710

Briefly - the approach ou describe (time delay before reclose) is called "Slow Transfer Reclosing" in NEMA MG-1-1998R1 - section 20.34.1 . In order to allow residaul flux/voltage to decay sufficiently, the power should be interrupted for >1.5 times the open circuit ac time constant for the motor. That time constant is defined in Section 1.60.1 as:
"Open Circuit AC Time constant"
= (Xm+X2)/(2pi*f*r2) where Xm, X2, r2 are equivalent circuit parameters (per-phase). f power frequency.

Some EPRI documents suggest a typical value might be 10 cycles... but it varies with the personality of the motor.

 

[gordonl]

Electricpete,

Thank you, that thread was excellent.

Gord

 

[jbartos]

Suggestion to the previous posting: Although excellent theoretically, practically when it comes to many motors and a variety of motors, some lacking parameter documentation; this approach may be unfeasible.

 

[gordonl]

I liked the comment in the other thread of using motor terminal or bus under volatge at 30% as an interlock for re-energizing. This is practical and makes a lot of sense.

 

[electricpete]

I agree with the 30% rule. Even if you happen to be 180 degrees out of phase when you reclose, the total voltage difference between internal and external voltages is then limited to approx 130%.

The above implementation uses a simple undervoltage device.

It is also possible to use a time delay in place of the undervoltage device, with basis identified above. The question "how long do I wait...." steered me towards this second implementation.

Regarding availability of the parameters:
#1 - You can ask the motor manufacturer... these are not obscure parameters... they are defined in MG-1.
#2 - You can estimate these parametsr from other available info as follows:
2A - r2 can be estimated from the slope of the Torque vs speed curve in linear region from zero slip to rated slip.
Slope = V^2/R2/wsync. I provided derivation in the thread on this forum that I initiated entitled "Estimating Power from slip -> temperature correction"
2B - Xm can be estimated from No-load current. Xm = Vll/(sqrt3*Ino_load).
2C - X2 can be estimated from locked rotor current.
use I_LRC ~ (VLL/sqrt3)/(X1+X2). You can solve for X2 if you make an assumption about the ratio of X1/X2.

The premise for the model is very simple. Draw the equivalent circuit. Open circuit the primary voltage terminal. Current circulates around the loop that includes the three elements XM, X2, R2.
Equivalent L = (XM+X2)/(2pif). L/R is (XM+X2)/(2pifR).

I can see the now complication comes in more than one motor. My first GUESS would be that you could select the longest time constant among the attached motors. After all if they are in parallel then the slowest-decaying motor will be trying to slow-down the decay of the faster-decaying motors and the faster-decaying motors are trying to speed up the decay of the slowest-decaying motor. By that model, using the slowest-decaying motor is conservative. But it is somewhat more complex than that... really both voltage and speed are decaying (we haven't even considered inertia and load-torque remaining after power interruption to compute speed... but apparently it's not necessary for the single motor case). I'm remembering that EPRI does provide a means for combining multiple motors into a single "equivalent" motor which is reminiscent of the approach used during fault calculations. I'll see if I can find that info.

Although it sounds complicated so far, I'm guessing that if you studied it (and I'm sure someone has), you could develop some relatively simple rules for selecting time delay based on available circuit parameters.

 

[gordonl]

I read an old Westinghouse article that suggested that the capacitance of the system be considered, beyond the installation of just capacitors. In the article they suggested that the capacitance of the cables be included because this will stretch out the decay of the voltage.

They're recommendation in the publication was for systems with long cables or capacitors to wait 5 seconds after the motor reached 75% of rated speed. Otherwise they suggested 3 to 5 seconds.

This was a Westinghouse manual from mid 70's. Any comments on the 3 - 5 seconds in particular.

 

[electricpete]

Gord
I looked in Nailen (Managing Motors 2nd ed) and saw different numbers than 10 cycles mentioned by EPRI.
He suggests a minimum time of about 1 second before reclosing. There's some discussion of the effect of capacitances slowing down the voltage decay... but it's not clear to me what types of capacitances would affect his 1-second number.

EPRI 5036V6 includes the following:
"Reclosing a deenergized motor circuit with connected capacitance is troublesome because the capacitor will maintain the generated voltage - even if low enough so insulation is safe - for a far longer time thatn without a capacitor... resulting in damaging torque transients.... The self-excitation problem largely disappears if the capacitor is small enough - 20 to 30% of motor KVA".
Once again, they don't tie that "safe" capacitance value to any specific closing time.

Some more general thoughts:

The considerations which might prevent you from selecting a "conservatively high" closing time might include:
- The longer delay time may result in drop-out of contactors, so that the motors never restart until manual action is taken.
- The longer delay provides more deceleration time. More likely the motor will have to accelerate it's load from zero resulting in full-duty start. If more than one transfer occurs within a few minutes, you could restart your motor from zero speed several times resulting in rotor damage.

But by and large these considerations are likely not on par with the need to prevent damaging shaft torques. Probably better to err on the safer side with higher time delay if it's not a problem for plant operation.

 

[jbartos]

Suggestion: As discussed above the motor inertia is the main criterion for reclosing a group of motors. E.g. there are motors that have acceleration time over 15 second, e.g. large axial fans. I just do not how the suggested one second in the provious posting:
\\\He suggests a minimum time of about 1 second before reclosing. There's some discussion of the effect of capacitances slowing down the voltage decay... but it's not clear to me what types of capacitances would affect his 1-second number.///
would suffice.
Also, as EPRI indicates with respect to capacitors, the any capacitor added to motor is working in favor of individual motor assessment, including its accessories, and then one would select the longest time to protect each and every motor.

 

[electricpete]

jbartos:
Here's the way I look at it:
Inertia is related to the speed coastdown but unrelated to the voltage decay (at least in the simple model used by NEMA).

So if the objective is to prevent damaging torques from out-of-phase reenergization with a residual voltage which has not decayed below 33% for a SINGLE motor, then the inertia is irrelevant (again based on NEMA model).

Inertia does play some role even for single motor in determining whether the subsequent restart will be as severe as a normal start from zero speed and operating temperature, or less severe (if motor has not yet coasted down for high inertia load).

My parenthetical comments regarding MULTIPLE motors mentioned inertia in a context where it didn't belong (based on above comments in this thread). I think in the MULTIPLE motor case that there will be a dynamics BETWEEN the attached motors which does depend on load inertia since motors will interchange energy differently if they are out of phase. This complication is not present for single motor and hence inertia irrelevant for single motor case. All of this points out that some caution must be excercized in blind application of the NEMA model as you and Gord expressed. Gord also brought up one other point not mentioned by the NEMA model which does have an effect... namely attached capacitances.

 

[electricpete]

jbartos

My last message rambled a little so I'd like to try it again with more direct response to your comment.

I don't think that inertia is the main criterion for voltage decay

It is NOT required for the motors to come to stop before restart (in fact it is preferred that they don't).

The important criteria for reclose is decay of the voltage which is primarily dependent on electrical parameters, not mechanical energy.

Mechanical energy only affects the frequency of the internal voltage... and therefore affects the way energy flows between multiple motors.

At least that's the way I'm looking at it based on the limited info I have read. (If I'm wrong then please let me know). I have never studied induction generators. It might be interesting to know how they behave to gain a little more insight to the behavior of motors coasting down.

 

[jbartos]

Suggestion to electricpete (Electrical) Sep 29, 2001 marked by ///\\\:
I don't think that inertia is the main criterion for voltage decay
///I would not downplay energy since it is generating voltage. As long as RPM are there so is the voltage.
E = B x l x v
E = voltage
B = Magnetic induction
l = length
v = speed
Clearly, as long as there is v high enough there is E high enough.\\It is NOT required for the motors to come to stop before restart (in fact it is preferred that they don't).
///Agreed. I do think that I have questioned it.\\The important criteria for reclose is decay of the voltage which is primarily dependent on electrical parameters, not mechanical energy.
///The motor internal electrical parameters are given by its design. The mechanical energy is variable dependent on the external parameters of the motor. The lower the external impedance, the quicker dissipation of the mechanical energy will be.\\Mechanical energy only affects the frequency of the internal voltage... and therefore affects the way energy flows between multiple motors.
///I would not downplay the mechanical energy since it is generating voltage. As long as RPM are there so is the voltage.
E = B x l x v
E = voltage
B = Magnetic induction
l = length
v = speed
Clearly, as long as there is v high enough there is E high enough.\\At least that's the way I'm looking at it based on the limited info I have read. (If I'm wrong then please let me know). I have never studied induction generators. It might be interesting to know how they behave to gain a little more insight to the behavior of motors coasting down.
///Please, notice that motors are more readily available for any demonstrations and measurements than generators. Also, there are more motors than generators.\\

 

[electricpete]

jbartos - those are good comments regarding the possible role of speed in voltage decay.

Your model (generator model) requires a flux B. It is my belief that the B is decaying in accordance with the electrical time constant. Since your model predicts voltage proportional to B, then your model also predicts that voltage decays at least as fast as the electrical time constant. Your model predicts an additional decay based on decrease in velocity v. I believe that NEMA'a approach was to ignore this effect for several reasons:
#1 - The data needed to calculate decay in v is unavailable... particularly when considering the fluid system parameters can vary depending on the fluid system lineup.
#2 - In many cases (high inertia loads with little friction and retarding fluid torque), the electrical decay is much faster than the speed decay, so ignoring the mechanical decay is fairly accurate.
#3 - It is conservative to ignore additional decay beyond the electrical time constant, since we are trying to establish a MINIMUM time before reclosing.

That's the way I look at it. I don't ask you to take my word for it. Look at NEMA MG-1 which allows "slow-transfer reclosing" based on an "open-circuit ac time constant" computed directly from electrical parameters (X2, R2, Xm), without any reference to inertia, load torque.

 

[electricpete]

Once again, I'd like to thank jbartos for his insightful comment. I wasn't trying to argue with it... only to incorporate it into my proposed justification for NEMA's approach.

I posted another message but I'm not sure where it has gone. In that message I explained that it is logical to conclude that the field B decreases in direct proportion to the decaying current within the equivalent circuit loop composed of XM, X2, R2. It seems very reasonable to me that flux is directly proportional to current thru XM from physical considerations.

Now it's worth a step back to look at underlying physical process. Assume a wye-connected motor. Open-circuit the power supply. At that moment the rotor has a current flowing in it which cannot instantly go to zero due to the inductance created by surrounding the rotor current with stator and rotor iron. Stator current must be zero since there is no loop for current flow in the stator. Slip disappears from the equivalent circuit since there is no stator field and relative motion between rotor and stator therefore seems unimportant. Slip was only introduced into the equivalent circuit based on slip-dependent voltage created in the rotor based on stator excitation... no longer the case... now stick with R2 vs R2/s. There will be a voltage induced into the stator which is speed-dependent as jbartos points out. But since no current flows in the stator which is open-circuited (and which I have for the moment assumed wye connection), there is no power transfer to the stator. The circulating current is decaying DC in the rotor reference frame. In the stator reference frame the induced stator voltage is decaying ac with frequency dependent upon rotor speed.

The rotor DC current will decay with time constant given by the L/R over the circuit which I have previously shown is equivalent to the expression that NEMA calls open circuit ac time constant.

We have already said that stator voltage depends on rotor speed... but what about rotor current. Is the rotor current affected by the relative motion of the stator (stator iron), even though there is no current in the stator. I think an equivalent question would be: If I have a very long rod forming the core of a short coil with DC current in the coil.... is there a voltage induced in the coil or force on the rod when I move the rod (but at all times the center of the coil remains filled with rod... which is much longer). I think the answer is no, but I'm not sure.

Another complication would be wye connection. In that case a large current would circulate in the stator windings and some energy dissipated by stator I^2*R. But would that energy come from the stator field or from the mechanical load? ... again I'm not sure.

 

[electricpete]

Last paragraph of my previous post should begin:
"Another complication would be the DELTA connection. In that case..."

 

[electricpete]

I stumbled across an "Induction Motor Data" sheet from Westinghouse for a 7000hp 3600rpm horizontal motor they built in late 1980's. It lists the "Open Circuit Time Constant, T'do' as 2.13 seconds. There is no indication of how they computed it (and no equivalent circuit parameters listed), but I assume it is intended to convey the same parameter as defined in NEMA MG-1.

I am GUESSING that this value (2.13 seconds) would represent an upper limit for most motors. High speed motors generally have low exciting current, hence high
Lm~ Xm. High horspower motors generally have double-cage rotor, hence lower running Rrotor. Both factors give higher time constant Lm/R.

 

[electricpete]

Stumbling some more....

I stumbled across some words in ANSIc50.41 section 14:
"Slow Transfer Reclosing". That document seems to agree with NEMA the NEMA definition (>1.5*open circuit time constant) with the following additional info

- "If several motors are involved, the time delay should be based on the longest time constant of any motor on the system being transferred or reclosed". (this apparently resolves one question from above)

- "Each bus transfer or reclosing reduces the life of the motor by a finite amount... minimize number of transfers during life of motor".