Ground Mat potentials

[buzzp]

What kind of potentials would you consider normal between two points taken directly from two extreme ground rods in a plant-generation (hydro) ground mat?
My background is in product design and I know, ideally, you would expect no potential difference. But there has to be some and I would like to know what some have seen for potential differences or noisy grounds in continuos operation (no faults or storms).

 

[cuky2000]

The follow are reference values for order of magnitude for expected allowable touch and step potentials.

Surface Step Potential (V) Touch Potential (V)
Concrete (50 Ohm-m): 250 - 375 225-325
Crushed rock (2500 Ohm-m): 2500- 3500 850-1200.

NOTES:
1- Allowable step and touch potentials are mainly a function of the body weight, shock duration and resistivity of the soil layers.
2- Postulated grid voltage rise is mainly depending of the injected current into the earth, soil resistivity, topology of the ground grid, and length of buried conductor.
3- Allowable voltage shall be always higher than actual voltage to meet safety criteria.
4- Allowable step potential are close to touch potential for low surface resistivity. However, for high resistivity values (>2500 Ohm-m) the step to touch potential ratio is above 3.

SUGGESTION: To determine the suitable values, considere use the guide lines of IEEE std 80, similar applicable standard or recognize softwares.

 

[peebee]

Wow, those numbers are higher than I would have guessed. What's the source on those, cuky?

Anyway, those numbers would only show up during faults, and would only be measured at distances of 3 to 5 feet. Buzzp is asking about voltages under normal conditions, and over distances that I suspect could reach tens or hundreds of feet. I don't know for sure, but my guess is that you could achieve tens of volts.

Lets say you had 100 amps of ground current, which is maybe high but doesn't seem unreasonable, and tried to shove that through a ground mat equivalent to 500kcmil of copper over 200 feet: per NEC Table 9 that'd mean an impedance of .057 ohms x (200 ft/1000 ft) = .011 ohms, that'd give you 1.1 volts over the 200 feet. Hmmm, lower than I would have guessed. Obviously, this is not a very rigorous analysis, but I bet it gets you in the ballpark.

Key things that haven't been addresed here are operating equipment voltages and current levels. Induced voltages could be much worse, too -- as an extreme example, if you have a single phase conductor routed through a metallic conduit over a long distance, the end-to-end voltage measured along the conduit could rise to roughly the same voltage as the conductor, it acts like a poor 1:1 ratio transformer.

So my final answer (guess): 50 volts.

How come you're asking? Just curious, or do you have a practical interest? Most people seem to be more concerned about the level of ground currents, but your question is certainly a valid one.

 

[rbulsara]

Am I missing something?

Draw an equivalent ciruit of the two ground rods with earth resistance and the ground mat! It will be two resistors in parallel. The top of the resistors (top of ground rods) are obviously at equipotential as they are connected through the ground mat. Unless the resistance of the ground mat is high, regardless of the current throuh the individual rods, the potential between the two points will be zero (they are shorted by the grounding mat!). If you measure any thing more than a few volts the ground mat needs to be redsigned

The earth resistance does not even come into the play!

Again any of the ground current is divided in as many parallel path as the grid of the mat so I would be worried if you measure more than a few volts (singel digits, if any).

The current is flowing from 'somewhere' through the earth and through the rods (resistors in the diagram) to a common point. But there is no current 'between the two rods through earth. Assuming these two rods are of similar in nature and more or less equidistance from the common point (even though it does not matter compared to the earth ressitnace), the current in two rods will be more or less equal. And so by any math again the voltage between top of two groudnd rods of a common system should be zero.

What peebee described is the potencial between the top of a ground rod a remote unconnected ground rod.

Feel free to shoot holes in this concept.

 

[jghrist]

The voltage between two distant points on the ground mat would be equal to the current flowing between the points times the resistance of the ground mat. Both quantities are small, so the voltage would be small.

Let's say you have a 500'x1000' grid with 50' grid spacing. From one end to the other, the lengthwise grid wires would average 1250' long, including the distance to get to them from the middle of each end. The parallel combination of 9 lengthwise wires, neglecting current flowing in cross conductors, would be around 7 milliohms if I figure correctly. You would need 140A flowing to produce one volt. If you had a high current bus or line running along the length of the station, you could conceivably induce this much current in the ground grid.

How would you measure the voltage? Voltmeter leads would probably pick up more induced voltage than the actual voltage difference in the grid.

 

[peebee]

rbulsara: You're definitely right that the potential should be very very low. But the assumption that "there is no current between the two rods through earth" definitely simplifies things a bit too far for the purposes of determining ground potentials.

There's definitely ground currents flowing in medium-sized industrial plants or generating plants. Actually, there's plenty of utility-induced ground currents flowing in open farmlands due to the common practice of many utilities to utilize the earth as a ground return path -- this is bad enough that cows can detect their step potential when they drink water and will in fact refuse to drink their water rather than receive a shock.

I realize the cows are somewhat off topic, but not all that far off. It's common for even small residential or commercial systems to have measurable ground current flows due to poor installation practices, aging motors, etc. In large industrial facilities, the problem is even worse due to induced currents, zero-sequence harmonics, etc.

Instrumentation and communications engineers have long been aware of the problems ground currents can cause on their systems, and those problems are the reason that isolated-ground receptacles are in widespread use (I have serious doubts about the effectiveness of IG receptacles, but that's definitely a separate topic). This is also the reason that the shields on shielded communications wiring is often terminated at only one end -- terminating both ends could lead to such high currents flowing on the shields that the cable overheats -- on a long cable run, that definitely implies a substantial voltage drop, too.

I certainly agree that if a ground mat is well installed and cared for, and relatively new, that the voltage measured should be fairly low, probably less than a volt. But it would not surprise me if larger older installations that have received average to mediocre care develop substantially higher voltages, say in the 5 to 20 volt range.

50 volts is kind of high, I admit. But my guess is that this would not be an abnormal condition at some facilities.

jghrist makes the very valid point that measuring any voltage would be extremely difficult. Similarly, the effects of any such potential would typically be negligable, at least to safety and the operation of power equipment, except under unusual circumstances. That's part of the reason that ground currents are typically discussed more often than ground voltages.

 

[buzzp]

I am asking because we are experiencing a strange problem with some instrumentation (related to another post). We have a simple meter hooked to a 4-20mA loop. Everything works fine until one end of the shield is hooked to ground. This is attentuating the signal somehow and producing erroneous readings. The same instrument is hooked to another panel (30' away) with the same set-up and everything is fine. (cable was ran on the floor to confirm the cable was not the problem). I plan to go there and take some measurements myself.
Obviously, the ground is suspect here and I inquired about the ground mat and they told me it is rock solid and no matter where you measure in the plant, they will all be at the same potential. I did not buy this statement based on my field experience and related design work. I was mostly looking for some expert input into ground grids. This hydro plant was built in the 60's (or maybe the 40's, cant remember) so it would qualify as an old installation.

 

[peebee]

buzzp:

I'd suggest the following:

1. Provide a solid ground bond between the 4-20 tx & rx units (maybe a #4/0). Also provide a solid bond to the building ground system. Terminate one end of the shield.

2. If #1 doesn't work, try isolating the rx unit from ground (ONLY if permissible by code). Disconnect both the shield and any other ground connection.

3. If neither #1 or #2 work, try an optoisolator or isolation transformer in the 4-20 circuit.

 

[cuky2000]

Buzzp,

Perhaps you should help us clarifying your questions.

- What is do you mean by “normal voltage” .
- Are you interested in known the expected voltage under normal operation at steady-state, fault conditions or both?

 

[buzzp]

The device works fine with the shield totally disconnected. This is how it is hooked up now. This is not required to be hooked up but it has gotten my curiosity up and I am a little concerned that this may be more of a problem as we update the plant with digital equipment.
My question was in regard to no-fault conditions, steady state. By normal voltage, I mean what have you seen in the field as far as the voltage between two ground points that was or was not causing problems. You can assume that the wires tying to the ground rod are negligible for the sake of this discussion. I am interested in two points (I guess two ground rods) and not necessarily grounds anywhere within a panel that houses electronics unless this is typically where the physical connection to the ground mat takes place.
I am not real familiar with the power language but I was wanting some data on what more experienced power persons have seen between any two grounds within a generating facility that are tied (as I understand it) directly to a ground mat via a ground rod. In my background, ground loops and ground bounce are a real problem with measuring equipment (background in this area). This post was more for information solicitation than troubleshooting a display device.
We were able to drive the disply through the existing cables with a loop calibrator with no problems at all. I think there is an issue with the cable (capacitance and inductance) that is introduced when one end of the shield is hooked up. I am pretty sure I can figure it out once I get there and am able to see some waveforms for myself as well as do some troubleshooting. I hate second hand information.
Peebee; your #1 above is real close to how it is supposed to be hooked up. Not sure about cable size though.
#2 has not been tried yet
#3 supposedly the transducer output is isolated but I do not know this for sure and will be one of the first things I check

I walked into this problem with very little info. My apologies. I will post more info as it becomes available or post the solution.

 

[peebee]

By the way, I assumed that both the tx and rx are powered from the same source electrical panel. If not, and this really is a grounding problem, try putting them on the same circuit from the same panel and see if that helps.

My gut instinct is that bonding the shield should have no impact on cable inductance or capacitance except in common mode -- and there, it should only help, not make things worse.

Have you actually measured ground potentials between the tx and rx units?

 

[jghrist]

The problem with ground loops is current flowing in the shield, not the voltage difference between parts of the ground grid. Grounding one end of the shield only is supposed to solve this problem, but obviously it didn't in your case. Are you grounding at the rx end or the tx end? Is the control equipment at the rx end connected to a ground with a single point of attachment to the ground grid?

 

[peebee]

30' is a really short distance. Grounding seems like a really unlikely source of problems unless there's a broken/bad ground connection somewhere or bad wiring practices were employed.

To double check for bad grounding connections between the tx & rx, put a common ohmmeter between the ground terminal/pad/conductor on each device (tx & rx), use any available wire to jump the 30'. You should see a very low resistance, less than one or two ohms tops. This assumes both the tx and rx are grounded devices -- obviously, if either of them are ungrounded, then the resistance will be very very high, but in that case you can also generally eliminate bad grounding practices as a source of problems.

Common bad wiring practices which might affect your installation include missing neutral-ground bond or not routing all circuit conductors through a common raceway.

It's really strange that single-point termination of your shield is killing the signal. Single-point termination should, worst case, impose a small common-mode voltage on the 4-20mA twisted pair -- but as 4-20 is a balanced signal, common-mode noise should have little effect. 4-20 was designed to go thousands of feet through electrically noisy areas.

Impedance and capacitance should have very little effect at 30', too. You should be able to get this thing running using coat hangars for conductors.

Are you certain that the shield is not shorting to one of the cable conductors? That could certainly cause the problem you're experiencing. A quick check could be made with an ohmmeter, check DC resistance from each conductor to shield and from conductor to conductor. They should all register very high, essentially looking like an open circuit.

 

[buzzp]

The electricians told me they disconnected both ends and verified there was no shorts or anything resemebling a short between the shield and the twisted pair and between the twisted pairs themselves. Has anyone ever done this test and find out later that with voltage present (albeit it will be low voltage) that the cable was bad? Or has anyone done a hi-pot test on this type of cable (twisted pair shielded)?
We have done all the other checks you guys mentioned (at least the electricians told me they did, got to see it for myself).
They did not measure ground potentials between the two units because they do not see the importance (all grounds are at zero volts they say, hence my questions on ground grids). I know there has to be a potential present no matter how good it is. It may be small but it is still there. This is something I will do on my own.
They said they tried grounding at both ends at different times with the same result. Also, both are supposed to be powered from the same panel. In any case, all equipment IO's are supposed to be isolated. If for some reason they are not, we could have the source of the problem.
I still need to verify the cable they used. Make sure it is not something totally ridiculous. I have a trip scheduled for next week to check things out for myself.

 

[peebee]

Hipot testing is OK. Most such cables are rated to 300 volts (verify this includes yours), some are 600 volts. Hipot accordingly.

A Time Domain Reflectometer (TDR) test might be helpful. Usually, I'd say TDR testing is way overkill for you application, but if you're really pulling your hair out you might consider that as a last resort.

I'm tempted to say trying to measure a ground potential between the two devices could be a waste of time as mentioned by your electricians -- try it anyway if you have the time. Checking the resistance to ground at each location could be useful, I'd strongly recommend they try that, and you could do the two tests (potential & resistance) at the same time with a flick of the switch on the multimeter.

"They said they tried grounding at both ends at different times with the same result. " It's not clear what you mean by this.

"I still need to verify the cable they used. Make sure it is not something totally ridiculous." Can't they give you a catalog number? Either way, even ridiculous cable should work on a 4-20 running 30 feet.

Any chance you can get your instrumentation vendor to stop out at the site to take care of this for you?

Good luck on your trip. Let us know what you figure out.

 

[buzzp]

"They said they tried grounding at both ends at different times with the same result. " It's not clear what you mean by this.
This was in reference to the shield. One end was grounded and they tried different ends, the transducer side and the meter side.
This particular situation is a contractors problem and not really mine as it is a new installation. They sent someone out and they are scratching their head too. They have consulted with their engineers several times. This is a large well known contractor.
This is one of them have to know things more than a serious problem. The origination of the problem will be stored for future reference. Although, the problem may include several different smaller problems and unique.

 

[ScottyUK]

buzzp,

Is the sensor in an area of high varying magnetic or electric field? I seem to recall this was a power plant application.

I had some wierd results from sensors producing a mA output - actually Hall-effect current transducers - where the output of the sensor had huge levels of distortion and very abrupt changes in magnitude and polarity. The current being measured was in a highly inductive circuit, so the observed results were unlikely to be a transducer problem, and this was confirmed using a different sensor from a different manufacturer.

I spent a long time checking grounding for shorts, open-circuits, ground loops, and so on. The problem was traced to capacitive coupling into the transducer electronics from high dv/dt conductors in the immediate area. Shielding the transducer sorted this problem.

Just my 2 cents worth,


Scotty.

 

[buzzp]

That certainly is a possibility and is definately not excluded. Thanks to everyone for your help. Will try to post some results when I figure this out.

 

[peebee]

A gaussmeter might be helpful in checking ScottyUK's magnetic field theory -- they can also be really helpful for identifying grounding problems.