power system cable shield grounded only one end to reduce surges 5

[electricpete]

IEEE Std C62.21-2003, “IEEE Guide for the Application of Surge Voltage Protective Equipment on AC Rotating Machinery 1000 V and Greater” states:

A.4 Summary of motor starting surge levels and surge rise times
Given a maximum [theoretical] prestrike voltage on third pole closing of 2.82 pu, the steep-fronted motor terminal surge can vary between 2 pu and 5 pu depending on the configuration. The highest surges are associated with small motors fed by shielded cables grounded at both ends with a large number of other loads present on the motor-starting bus. The most effective mitigation measure is to ground only the motor end of the shield.

The calculated surge rise time, using time constants, can range from 50 to 1500 ns depending on the configuration. The fastest rise times are associated with small wye-connected motors, short shielded XLPE or EPR cables, and narrow width switchgear (short bus length at the supply end).

Has anyone heard of this configuration... shielded power system cable with shield grounded at one end only?

I assume you still have stress cone on the other end, but it just has no external ground connection

What concerns would you have for this configuration?


=====================================
(2B)+(2B)' ?

 

[electricpete]

I should have mentioned that I am interested in medium voltage applications.

=====================================
(2B)+(2B)' ?

 

[magoo2]

Ground it at both ends and you get circulating current in the shield. Ground it at one end and you don't.

If you're dealing with concentric neutral cable, the additional circulating current may not bother you. If however you're dealing with tape shields or LC shielded cable, the thermal capability of the shield is quite limited.

Ground it at one end and you get a longitudinal voltage induced along the shield. For high currents and long cable lengths, this could be an issue. For short lengths, it's not a real concern.

 

[electricpete]

Our installation already has shields grounded at both ends and I’m sure shield current was considered in ampacity. I am interested in what hazards or problems could be created if we switched to grounded at one end only with intent of reducing surge strength at motor.

You mentioned induced voltage. I see a table here where current level assumed based on conductor ampacity and lengths are tabulated to prevent going above 25vac induced voltage.

http://www.okonite.com/engineering/shielding.html#table

2

The concern is that 25vac is considered hazardous to personnel?
I can imagine we would be more concenred about voltage induced during fault condition, but the table doesn’t take that into account.


=====================================
(2B)+(2B)' ?

 

[dpc]

The issue with grounding at only one end is safety, as you have mentioned. The induced voltage is the primary issue and just the belt and suspenders approach of having two grounds as opposed to one. For industrial systems, the circulating current in the shield is generally not a major concern.

My experience has always been to ground at both ends in industrial environments. But there may be other factors that we never worried about.

David Castor

http://www.cvoes.com

 

[magoo2]

I agree with dpc that safety is the primary issue. I'm not sure 25 V is the correct value as I've seen reference to 50 V on occasion.

 

[electricpete]

7anoter4 – thanks for your post... which is no longer here... but I did get a chance to read it.

I assume it must have been flagged by someone (not me) as a result of the first link.

I was not aware of IEEE 575, but that’s a great reference... exactly what I was looking for. It gives limits for induced voltages and some means to calculate them.

All the motors I am interested in are 13.2kv with single conductor construction (not three-conductor cable).

It is an interesting idea to transposing the shields. We do have one oddball application where we might be able to do someting like that. In this application, the breaker feeds a a long run of cable, which leads to a second cubicle where it transitions to bus, then back to another long run of cable and on to the motor. (don’t ask me why it’s built that way... long story). That intermediate cubicle has stress cones on each end of the bus and might be an easy place to isolate the shields from ground and transpose them.

I wasn’t able to access the IEEE Power Conversion document since it is outside my subscription, but I have a few documents that discuss ways to estimate the effects of the power system upon the surge seen at motor terminals (IEEE Std C62.21-2003 and EPRI 5826 Volumes 1 / 2).

Thanks again for you post. You deserve a star.

=====================================
(2B)+(2B)' ?

 

[7anoter4]

Thank you, electricpete for your kind appreciation.
The link to IEEE I inserted in my post was illegal so I have to apologize for the inconvenient and in the future I'll endeavor to be more circumspect. Now I'll try to revise the post.
First of all the maximum build-up permissible voltage could be 65 [even 90 V!] as per IEEE-575-1988 Appendix C.
Second, in a three-phase shielded cable, were the shield is exposed under the outer jacket and most of time in contact one phase against other, there is no build-up voltage [theoretically, at least].
20 years ago when vacuum breakers invaded the world a great concern was about the surge [voltage] and short rise time of the surge that could damage the motor. The problem was not the level of the surge but the multiply prestrikes.
The study conduced by EPRI CALIFORNIA and Ontario Hydro Research CANADA in 1988-1992 stated if the shield is grounded only at motor end the surge will drop from 5.1 to 2.3 pu. But this study does not mention if the cross-bonding of the shields- which could reduce the voltage build up [or even canceled it!]-will produce the same effect in the motor end grounding case. My opinion is it could be.

 

[electricpete]

Thanks again 7anoter4.

Each phase is a separate single conductor shielded cable. The shields of one phase will not contact the shield of another phase due to presence of outer jacket. The construction is similar to this:

http://www.okonite.com/Product_Catalog/section2/sheet8.html

In our case, the cross bonding that you mentioned will not apply, correct?

=====================================
(2B)+(2B)' ?

 

[electricpete]

I would think that cable construction (jacket outside of shield) may also ease personnel hazard... if we know the insulation rating value of the jacket.

=====================================
(2B)+(2B)' ?

 

[7anoter4]

I think the cross-bonding is dedicated to single core shielded cable like yours.
What you need is link boxes. See:

http://www.eea.co.nz/images/Events/...ations/Pres 5 Zhou, Zoseph - Presentation.pdf

or:

http://www.elektroskandia.ee/pub/to...ge/Tyco-Electronics-Raychem-132KV-LinkBox.pdf

The entire length of the cable -starting from the breaker end ending at the motor terminals is divided into 3 equal parts and the shield will be cut and cross bonded at each link box. Screen voltage limiters are provided in each link boxes to avoid overvoltage.
It is not absolute necessary to transpose the cables and you don't need to ground both ends- in order to avoid surge. Instead, you need another screen voltage limiter at the breaker end.
A capacitor of 0.1-0.2 microfarad at the motor end may increase the time rise.

 

[cuky2000]

To get the best of both arrangement, there is option to use surge arrester at mid-point grounded in a such as way that in normal operating conditions, the cable system will operate as a single point ground to avoid circulating current and maximize the cable ampacity.

If any voltage is build up to a define threshold, the surge arrester will discharge to ground maintaining the voltage profile in the cable to a defined controlled value.

Another way to optimize the cable voltage profile is using a cross bonding of the cable shield.

Althought it is possible to use those techniques in MV applications, will be fare to mention that is more frequent used i higher voltage UG applications.

 

[electricpete]

7anoter4 – Thanks. Now I see I had misunderstood your post 29 Mar 11 10:33. When you said “this study does not mention if the cross-bonding of the shields” I initially thought you were saying the shields were in contact along the full length. Now I realize by cross-bonding of shields, you meant the same thing that I was calling “transposing of shields”. I see your terminology is more standard than mine since “transposing” seems to be reserved for conductors, and cross-bonding for shields.

Now I have a question. What is the basis for concluding that cross-bonding of shields would reduce surge at the motor? I realize cross-bonding of shields accomplishes a similar result to single-point grounding in terms of minimizing losses, but I don’t know whether or not it accomplishes a similar result in terms of reducing surge.

To take a step further back, a question for everyone…. I really don’t understand exactly why single-point grounding reduces surge. Here is what is stated in the EPRI document:

…For shielded motor cables grounded only at one end, the surge splits into an internal component and an external component. The external component propagates between the shield and an equivalent ground plane with surge impedance Ze.

For shields bonded at the motor end, the surge splits at the supply end and the external component is eventually shorted at the motor end. Thus the motor initially sees the internal component only, followed later by a sloped back external component. The risetime of this external component is typically twice the cable propagation time. Thus for long cables, the slope back is significant and only the internal components stress the motor inter-turn insulation. This may reduce the ratio Vm/Vp to as low as 0.6 (see Table 5-2, Section 5.7).

…For the cable shield bonded only at the breaker end, the fuIl surge travels to the motor where it splits into two components. The internal component enters the motor, while the external component returns to the breaker, is reflected at the breaker end and appears at the motor as a sloped back surge. Again, the risetime of the external component is typically twice the cable propagation time. The surge reduction for bonding at the breaker end is normally modest since Rm is large compared to Zc,.

I read that a few times, but I can’t quite follow the logic of internal and external components.


=====================================
(2B)+(2B)' ?

 

[ausphil]

electricpete,

when you have a wavefront travel along a transmission line, when it reaches a change in impedance, you get some reflection, and some transmission.

my assumption is that the "surge" is something generated in the system, not by the motors, hence may be detrimental to the motor.

if you have a wavefront travelling along from the system to the motors and have the cable shield grounded at the motor and not at the supply end of the cable, when it reaches the start of the cable, there is a large change of impedance (because the impedance changes from a transmission line that has a relatively close ground, to one that then has a ground with added impedance because it is connected only from the motor end of the cable). This will cause a large amount of the surge energy to reflect at that point, and hence a smaller amount of the energy will travel through to the motor.

if you have the ground at the supply end, there is very little of the surge reflected at the supply end of the cable, because it is a nice transmission line with a nice close ground, and the majority of the surge will travel to the motor end. The other problem you can come up with is an increasing of the magnitude of the wavefront where you get the large change in impedance - in an open circuited cable, you will see a doubling effect.

my reading of the quote is that if you have the motor end shield grounded, you are essentially creating 2 transmission lines at the supply end of the cable - one one between the conductor and the shield, and the other between the shield and the ground. The surge comes along and then travels along 3 paths:
1. some reflected as described above - this is sent back tot he supply, so doesn't get to the motor.
2. some travels in the transmission line created by the conductor and the shield - this gets through to the motor
3. some travels in the tranmission line created by the shield and the ground - this gets shorted out by the shield to ground connection at the motor.

so overall, the amount of energy contained in the surge that gets to the motor is reduced.

the question is how much? that's probably hard to quantify without all the impedances of the cable etc

ausphil

 

[electricpete]

Thanks Phil. I think I understand what they're saying now.

We know how to predict behavior or a "Tee" from simple transmission line (TL) theory. If we make a simplifying ASSUMPTION that all TL's have the same characteristic impedance, then the impedance of the parallel combination of lines downstream of the Tee is 1/2f the impedance upstream. This gives a transmission ratio TR = 2*Z2/(Z2+Z1) = 2*(0.5)/(0.5+1.0) = 2/3.

Ordinarily we think of the high-potential conductor Tee-ing... in this case the Tee occurs on the ground/shield side, which shouldn't make any difference in behavior.... it's still a Tee as you mentioned

Attached powerpoint shows analysis using a ltspice model (which perhaps is not required for the simplicity of the problem, but gives a nice graphical picture imo).

Slide 1 is overview of the spice model, discussed below LEFT/MIDDLE/RIGHT
LEFT: On the left we have source and articificial delay TL's associated with the source for reasons explained in the slide... pay them no mind.
MIDDLE: In the middle we have model of the shielded power cable. It is represented as 2 TL's: one representing conductor to shield, and the other representing shield to ground. Since shield shows up in both, the lower conductor of top TL (representing shield) is shorted at both ends to the upper conductor of bottom TL (representing shield) at both ends. (I wasn't 100% sure if this would be correct when I tried it, but seems to give all the expected results verifiable by hand). On each end of the shield is a resistor for connecting shield to ground. Resistor is set either to very high value (1 gOhm) for open circuit or to very low value (1 nOhm) for short circuit.
RIGHT: On the right we have a resistor representing the motor.

Slide 2 is result of analysis of shield grounded both ends. Shield remains at 0 volts as expected as shown in plots V(s1), V(s2), V(s3). The upper TL acts like simple transmission line throughout. Voltage at the right end increases based on higher motor impedance as expected: 2*2/(2+1) = 4/3

Slide 3 is Shield grounded at motor only. There is a "Tee" labeled where the shield is introduced at the source end of shielded cable. When the Tee: is hit, the voltage on each TL reduces since parallel lines have lower impedance (1/2). Each TL on output of Tee transmits 2*1/2/(1/2+1) = 2/3. Voltage from conductor to ground is sum i.e 2*2/3 = 4/3. Upon hitting shorted termination at right side, the lower 2/3 is reflected (inverted) and the upper 2/3 remains... that 2/3 hitting motor impedance rises by 4/3 as before to give 8/9 at motor (node N3). This motor voltage is 2/3 lower than it was when shielded both ends... this is a result of assumption that all TL impedances are 1 (ratio 2/3 would remain the same regardless of Z0_motor). It may be helpful to visualize a very short (small delay) Z0=1 TL between the termination end of shield and the motor so that we have 2 separate interfaces. I did that with simulation and results were the same except for the small delay.

Slide 4 - Shield grounded at supply end only. Results voltage of 1.0 at motor, which is lower than shielded at both ends but higher than shield grounded at motor end only. Note that if we had selected a higher Zmotor, grounding at supply end (only) would not show as much benefit over grounding at both ends

Slide 5 - notes about units etc.

So in summary this particular simulation grounded at motor end only was best, grounded both ends was worst, grounded supply end only was in the middle. (Lots of assumptions)


=====================================
(2B)+(2B)' ?

 

[electricpete]

A followup question. If I ground the stress cone shield at the motor end, and remove the stress cone ground strap at the supply end (leaving stress cone grounded only through cable shield from other end), is the function of the stress cone compromised?.... will we expect any possible cable failure at the supply end stress cone?

=====================================
(2B)+(2B)' ?

 

[ausphil]

electricpete,

Single point bonding of HV cables in outdoor terminations is done by having the stress cone isolated from earth at one end, and the earth connection is made through the sheath back to the next earth point. The whole termination structure sits up on small insulators.

When you think about it, the main function for the stress cone is to provide an "earth reference" to ensure that the voltage is nicely graded through the termination. If the reference voltage is 0 (solidly earthed) or, say 50V (if it sits at the end of a single point bonded system) then compared to the possibly 100s of kV, it is still effectively earthed, hence the grading of the voltage along the stress cone is still valid.

This is the same situation as the system you describe, but the motor is obviously at a lower voltage (maybe a couple of kV).

I can't say whether you will expect any failure at this termination, because there will be many construction inputs that need to be done right to ensure that the termination is made correctly. There shouldn't be any increased risk.

On a frivalent note, just looking at your tagline, you could probably embelish it by adding the symbol for "that is" in there:
(2B)+(2B)'ie ?

ausphil