Over Voltage Condition on Grid

Sounds like some major load shedding event is happening.....Maybe you need some grid stabilizing strategy...and control system rethink...

https://en.wikipedia.org/wiki/Grid_energy_storage

For instance, if portions of the power system can be considered essentially ungrounded, high ground impedance maybe in the middle of a line section, a ground fault can cause an overvoltage on the ungrounded phases, line to line voltage appears to ground. Without a heavy current to ground, an arcing ground fault can charge & discharge the distributed line capacitance, creating even higher transient voltages on the lines, which could then flash over an insulator string with maybe only 150 kV BIL instead of perhaps 200kV typically applied on 35kV systems outdoors.

Lots of missing facts to analyze.

Yes I agree with you especially in the 'hint' .

This has nothing to do with the mini 6.5MW hydro plant.

I spent the whole day analyzing the system events as printed by the Scada computers and physically down loading fault data from protection relays.

I have arrived at some conclusions:

1. The high Voltage manifested only after loss of loads during the incident. Running up to the incident, Voltages were normal.

2. At the start of the tripping sequence, distribution lines, some power transformers, and some transmission lines tripped on 'load imbalance'

3. I have identified only one transmission line with severe imbalanced load.

Still investigating the this imbalance on the transmission line.

1. A sudden voltage rise after dropping a large load is not unusual, especially if it is large enough to be a significant percentage of the line flows and is also highly inductive.

2. What was the nature of the initiating event that led to the tripping sequence, how long did the event last, and was there any system separation or islanding. Also, what type of relaying was used to determine the "load imbalance", what was the degree of imbalance, and how was it determined?

3. Is this transmission line a relatively long line, a tie line, a radial line, or a line feeding a large inductive load and/or significant generation?

I can understand your reluctance to supply an SLD and any other details that could identify your client, but without them, only general answers are possible.

1.The nature the initiating event was a single phase to ground in phase A

2. The event lasted 2 seconds

3. There was system separation when tie lines to a neighbouring country opened but due to over current after massive loss of load

4. Type of relaying used to determine load imbalance is Siemens Over current relay 7SJ802

5. The degree of imbalance was as such: A phase: 50 amps, B phase: 203 amps C phase 200 amps just an example on one line. The rest of the lines had similar imbalances.

6. The imbalanced currents were read as part of fault data from respective relays

7. The Tx line is short, about 5 km only and it is a tie line between 2 substations

Thanks for further information - there are many questions that could be asked, but your original question highlighted overvoltage. It now seems the problem begins with an earth fault.

You now throw down another "joker card" - there is a line to another country - which may have auto-open/reclose for line earth fault, but probably no signal to tell your remote system it has operated. Is this same line "5 km" described in your item 7??

You have not stated if the earth fault was in your system or the cross-border line. Without appropriate relays, it can be difficult to know where the earth fault is located. Loss of voltage on one phase will affect all your feeder currents. Cross border line might see your earth fault "within its area of protection".

It would be expected that a high earth [phase] current would" stand out" in the fault data from one relay - if not the earth fault would be outside your system.

You indicate trip of your relays is on "unbalanced load". Such a trip is usually slow, leading to problems for your generators. Roughly, loss of synchronisation will occur in tenths of seconds, 2 seconds to clear a fault is too long, except as a back-up to primary protection (with loss of supply to comsumers expected).

If the earth was in your system, it seems odd that you do not appear to have fast-acting earth fault protection on your feeders.

Can the Siemens relays you reference be programmed for earth fault protection as well as overcurrent?

Yes the earth fault was in 'my system'. It was in the 132kV grid. The imbalance protection affected only medium Voltage distribution feeders. What I forgot to mention is that protection on the faulting line operated quite fast, but one circuit breaker did not open. The breaker failure scheme took 2 seconds to act! Why this long? I am still investigating

Two seconds is not unreasonable. Since this is a backup function, it is necessary to coordinate its operation with the initiating protection and any intervening reclosing operation; therefore time has to be allowed for the reclosure to not be interpreted as a stuck breaker.

1. In reply to this #20 post and your #19...
   You now tell us you have a 132 kV system - up to now we knew a bout a hydro generator and 33 kV system.
2. As RAMConsult wrote in his first post, CR4 has too little information on your system to attempt a good diagnosis - however, you seem to be doing a good job of hacking through the information jungle to build a picture of what happened!
3. You write "when load was lost, ...power... swung.....to ... neighboring .... which eventually tripped". To clarify, do you mean local generators lost local load [how?] ,but continued at same or increased output by exporting to next country? And how long was an "eventual trip" delayed after first fault?
4. Not knowing your system, it may be that in the absence of twin circuit/ alternative connection or breakers to clear the circuit with the failed 132 kV breaker without fatally splitting the system, the system designers did not include a fast intertripping scheme to isolate the failed breaker and its circuit.
5. Generators have limited capacity to supply unbalanced loads (negative phase sequence). Since load feeder imbalance tripped before the generator protection, it has operated correctly. Two seconds is reasonable since, roughly speaking, generators and transformers can stand a 3 second short-circuit and basic protection should allow as much time as possible for one (or more) selective relayings to work first.
6. If you lost local load, but then exported with maximum excitation; then tripped off that load, leaving negligible load, it may be unreasonable to expect AVRs to stop overvoltage and generator tripping with little load (note that generator transformers have higher overvoltage requirements than normal - but your figure of 180% exceeds it).
7. The usual brushless exciter/AVR can throw off rated load with an acceptable voltage jump, but with maximum field forcing, the AVR system has no means to reverse field voltage/introduce extra rotor circuit resistance to reduce field current quickly - the rotating rectifiers can only energise the field with one polarity. I am assuming you do not have generators with rotating SCRs or brush-connected rotor field when faster response is possible.
8. Usually, AVRs have single phase sensing and a standard circuit probably means all your AVRs were on same phase, which maybe happened to be affected by faulted phase. It may be that 3 phase sensing causes less extreme response, but I feel it makes little difference here.

The SJ802 is a sophisticated progammable multifunction relay that may have had other functions besides current imbalance enabled. The loss of a tie-line has many implications for the post-fault reaction(s) of the separated systems, and depends upon the magnitude and direction of the line-flows, the amount of spinning reserve, and the strength of the remaining interconnections. Without knowing the prefault system status, the relay settings, and/or the weather at the time, it sounds like everything operated as it was programmed.

It does seem odd however, that a single relay could trip an interconnection tie-line, usually there are multiple indications such as ROCOF (Rate Of Change Of Frequency), Over/Under Frequency/Voltage, etc. involved in such a critical trip sequence. More data yields better answers.

When load was lost, a lot of power swung to the neighboring country via the tie lines which eventually tripped on over current

I had the opposite problem once; an under-voltage condition. The problem/solution could be related somehow. First, let me point out that some manufactures try to make things "fool-proof" (In other words, so that a fool can't mess it up). That manifests itself in things like making carburetors nonadjustable and removing "tweakable" controls. One day they might even make cars with the hoods welded shut. But anyway, the point is that everything seemed to be working properly, and finally I disconnected the regulator and supplied the field with an adjustable auto-transformer just to see what kind of excitation was required to produce the proper voltage. In the end, I wound up replacing a resistor in the control circuit of the regulator to bring its output back into range. The age of components do funny things in there that are hard figure out. Without tweakable rheos/pots to compensate for those changes means that tracing/replacement of individual components might be necessary if you want to avoid replacing the whole thing. But, I'm assuming that there was no major drop of a load that caused a sudden voltage over-shoot due from a slow regulator response time. Age can cause slower responses too. Just be sure you cover all bases before you decide to do anything.

True! There was a major drop of load. This I discovered later.

Actually, it appears at some point in time just before the tripping sequence started, there was very low current in one phase compared to the other two phases, and throughout the system!

If you single-phased a large group of 3 phase induction motors, the motor protection would work relatively quickly (seconds, as opposed to milliseconds) to disconnect each and every one, to prevent melting out the rotor bars. That would be an event that could result in a coordinated load rejection that your generators might not be equipped or adjusted to deal with.

This might be a case where saturable current transformer (SCT) excitation on your largest generators might give you the edge to respond to load rejection or fault currents quickly enough to keep the power system stable. Probably more than you can expect of brushless shaft mounted exciters.

Control systems including field forcing and reversal can act as a big reactive sink to control voltage excursion, but an accurate model of the load elements and real time data is needed to predict the response of the transmission elements, and balance the reactive demands of the load and transmission equipment to the generation, certainly doable, but does require some engineering.

An SCT excitation system has excellent response to sudden overloads, but equally poor response to load rejection, because, depending on its exact type, it relies very heavily on the magnitude of the line current out of the generator. Unfortunately during sudden loss of load the line current drops as well, causing the the current (and consequently the terminal voltage) to drop to unacceptable levels before the AVR (if in operation) can recover the terminal voltage to the correct levels. Such misoperation can lead to violent system swings as all the other control systems (governors, fuel management, AVRs, etc.) hunt for the correct operating point.

yes, typically SCT excitation systems also come with voltage based excitation (PPT), to cover startup, and provide the power for field reversal if required to support system stability.

I find it interesting to read that currently, units put on line over a certain size in the US are now required to be fitted with power system stabilization equipment. The industrial systems I am familiar with often provided such equipment based only on local concerns, with the serving utility only tangentially involved with the equipment specifications and controls.

You are referring to the NERC (North American Reliability Council) requirements imposed on the WECC (Western Electricity Coordinating Council) that went into effect on 7/1/17. The reason for this requirement has to do with very large geographical area of the WECC and the long distances between generating stations that results in a form of instability known as SSR (SubSynchronous Resonance).

However these requirement are not imposed on other utilities in the US unless they have also experienced frequent SSR, which is usually not a problem in tightly integrated power pools such as the PJM Interconnection, and most others, where such use would be optional if at all.

Typically what are time delay settings for hydro gen exciter systems? Normally exciters experience imbalance in Voltage or current during single phase faults in the grid. So if an exciter has protection against such imbalances due to system faults, what time delay s are typical?

The excitation system really can't see a problem with phase unbalance, except locally as field winding heating, measured by locally set RTDs or by the voltage required to push a certain current through the windings, higher resistance of the winding than expected or design(rise by resistance).

So the excitation protection would be based on the rotor /slip ring thermal design, and trip setting set by the generator manufacturer to hold a certain insulation design life.

Unbalanced operation of the generator stator windings can be measured with line current transformers, again, this falls back to the thermal capacity of the generator stator windings. The time delays could be significantly different based on the particular machine design, but I would expect in the order of seconds, rather than milliseconds, in general.

"...The excitation system really can't see a problem with phase unbalance..."; the only way to say that with any certainty is to know the type of excitation system, it's one of the arguments against having a separately powered excitation system vs a generator shaft mounted system.

"... locally as field winding heating..."; the field winding overheats because too much current is being passed through it, the cooling system has failed, or large amounts of Negative Sequence Currents are flowing across the surface of the rotor. btw- how do you get a reading off of a rotating RTD? Modern multifunction generator protective relays calculate the rotor temperature by monitoring the terminal PTs and CTs and putting those values into thermal models of the generator, especially since there is no direct access to the main field winding on brushless excitation systems.

A generator cannot operate without an excitation system, that's why main field overcurrent is always an "alarm, wait, then trip" function that disconnects the unit from the load.

The generator protective relay determines the degree of imbalance based the amount of negative sequence current flowing in the stator, then starts an inverse timer based upon the heating (I2²t) effect curves as provided by the manufacturer. The worst case is single phasing which can blue the iron in seconds, while "normal" imbalance can be tolerated for much longer periods of time.

At one of our Hydro plants, exciters trip when imbalanced current or Voltage is sustained in the field circuit for at least one second. A faulting phase may experience a Voltage drop which may prompt the exciter to respond by attempting to 'boost' the Voltage in this particular phase; hence leading to the imbalance. This has caused inconviniencies as at times a one second delay fails to discriminate with back up protection system in the grid.

It wouldn't be unusual to find that the upstream relay coordination was set carelessly, not realizing the import of the generator field protection setting. However, if the regulator is field forcing, perhaps that value is set to a higher value than prudent, and you should lower the boost amount, or change the ramp speed.

The field winding response to the external voltage change may not be included in the regulator control settings, or, perhaps in a more sophisticated regulator, a model of the field parameters, if adjusted properly, would not allow the field forcing to overshoot, it would anticipate the excitation delay, and reduce the ramp speed.

These machines usually have many poles, lots of iron and copper, so it should take some time to add enough heat to raise the temperature to a dangerous level, not like 2 pole steam turbine generators of recent design. The generator manufacturer should be able to help with the capability of your particular design.

I'm surprised to see that no one seems to have has implemented direct temperature measurement of generator rotor windings, even on 200MVA rotors. We've had the technology in mass production for a decade (most tire pressure monitors), the temperatures around the brush rigging are reasonable, and either inductively coupled or RF Link methods have been explored. We must not tear up enough of these machines to justify the development of this equipment

It is quite common for systems to open only one phase on earth fault for long enough for an arc to earth to stop; then auto reclose. Keeping two phases connected can avoid generators losing sync and tripping.

But it only works if the break is short enough.

If your system is still "in construction", available generators may exceed the assumed phase swing - or maybe the break - reclose time is too long because auto -reclose has not been set-up yet.

Further to my post #9.....

You may have a SCADA system, but it does not necessarily catch every event, or put events in correct order at same logged time.

Often, they only poll each point once per second, missing points which change & change back within one second (e.g. auto-reclose).

And maybe the SCADA system is only part-finished.

"Sequence of Event Recorders" are designed to catch all events and time them to the millisecond but do not run on "windows" PCs already doing multiple tasks.