L64D_P/N - Intermittent Voter Mismatch

Having intermittent voter mismatch for the above alarms on MKV. Started out on just R core but has progressed to S & T now. Diagnostic counters are showing voltages that don't seem possible going to -150 volts to 150 volts randomly. This is happening consistently across R S and T cores but it's never the same on each core. Might see 150 on the positive and 50 on the negative, might be -122 on negative and -18 on the positive. My point is the voltages on DIAG C TCQA power supplys are all over the place. Thought we had a smoking gun with our compressor bleed valves but didn't resolve the issue.

Any help would be appreciated.
Can reply to here or email [email protected].
 
zachary747,

I've seen this a few times, and the best way to measure the DC voltage split is to use a voltmeter and measure it at the input terminals to the <PD> Power Distribution core--with respect to ground (earth). If the voltages with respect to ground are, indeed, fluctuating as they are on the screen, then the likely causes are: bad panel grounding (to the plant ground grid); bad plant ground grid (which does happen to many plants over time--the soils and moisture in the soils can deteriorate the ground grid and make it ineffective); a high AC voltage on the ground grid (from a bad motor or something similar); a bad battery charger (the output filter capacitors do go bad in battery chargers placing a large ripple voltage on the output); or a severe induced voltage on some wiring connected to the Mark V (usually caused by high-current/high voltage AC cabling run in the same cable tray or trench with low voltage signal wiring).

Even if the voltages at the <PD> input terminals are not fluctuating (or don't seem to be fluctuating) the causes above and below should be investigated and eliminated as the potential cause.

Poor grounding practices during construction resulting in high induced voltages are very difficult to find and correct. A couple of times new high voltage and -current cables were mistakenly added to the wrong existing cable tray or -trench which caused the problem (no previous problems, until the new cables were installed and current started being carried by the cables).

One thing that was often overlooked during the installation and commissioning of the Mark V was the grounding (earthing). There is a ground bus bar in the bottom of the Mark V, and no matter where the Mark V is located (even if it's in a compartment provided by GE or the packager of the turbine and auxiliaries) there MUST be a ground (earth) cable from the plant's ground grid directly to that ground bus bar. MANY people believe(d) that if the Mark V was bolted to a metal plinth or foundation and the structure the Mark V was located in was grounded (earthed) properly that the Mark V was also grounded (earthed) properly. NOT SO. The Mark Panel components were painted--usually well painted, and so they didn't always make good metal-to-metal contact. To avoid any possible grounding/earthing problems it is necessary to add a proper ground/earthing cable from the plant's ground grid directly to the ground bus bar in the bottom of the Mark V.

Bad battery charger output filter capacitors have cause lots of similar problems. Including high voltage spikes when taking voltage readings, as well as higher than normal battery charger output voltage readings.

Many ground grids are just not well constructed in the beginning, and that can lead to many problems decades down the road.

Many soils have eroded some ground grids over the years. They can be acidic or highly alkalinic. Combine this with a poorly constructed grounding (earthing) grid, and it will eventually be ineffective over time.

Large motors whose insulation has deteriorated over time have also caused problems with grounding (earthing) at many plants.

Another potential culprit is the <CPF> conditioning power filter used in many Mark V panels. The power coming into the Mark V passes through the <CPF> and then to the input terminals of the <PD>. If the components of the <CPF> have been damaged (say because of lightning strikes and/or voltage spikes) problems like this can occur. The <CPF> is usually located in the absolute worst possible location to access and work on--at the center of the very bottom of the Mark V panel, at the back of the Mark V panel, under the <QD1> core, and where MANY cables usually enter the Mark V from underneath the Mark V.

Another potential problem could be the <DACA> module, if the panel has one (or sometimes two). These modules convert AC into DC and the output is connected to the Mark V (usually through the <CPF> if present). Many sites used high power inverters instead of clean UPSs (Uninterruptible Power Supplies) to feed the AC to the <DACA>. And, if the output of the inverter (or the UPS) is not "clean" (if it has large ripples and spikes and dips) the <DACA> actually magnifies the voltage excursions which is NOT good for the Mark V.

That's a pretty comprehensive list of potential problems. It's a good idea to periodically test ground (earthing) grids; best to have a professional company come use proper instruments and methods to do the testing.

As with all problems which begin after some time of normal operation it's a good idea to think back to when the problem started. Was it after a trip from load? After an electrical storm or a lightning strike or a grid disturbance? Was it after a maintenance outage? Was it after some new equipment was added to the plant? Something that wasn't mentioned? Many times, problems like this are eventually traced back to such events. Good troubleshooting always takes things like this into account when developing a list of things to check--and eliminate--until the root cause is found. And, something to remember--it might be a combination of things that are contributing to this (or any) problem.

People always cringe when I say this, but, ... It's very rare that the DIAGC configuration file used by the DIAGC Display matches the exact cards and PROMsets in the Mark V. Which means, the data can be misleading. If PROM upgrades have been done at the site, it's even more likely that the DIAGC configuration file doesn't match the card/PROMsets in the panel. The only way to every ensure the DIAGC configuration file exactly matched what was in the panel (at that instant in time) was to capture the output of CARD_ID, and write down all the cards and their revisions and send that along with the DIAGC configuration file to GE in Salem, VA, and they (actually there were only a couple of people who could do this) would review the information and make the necessary changes and send the proper DIAGC configuration file back to the site.

Hope this helps! Please write back to let us know what you find.
 
Top