Big blackout. What happened? 40

[Skogsgurra]

When I got words about the big outage, I immediatley went to my puter to find out what my engineering friends in the US had to say about it. But no Eng-Tips page available. Of course I can understand that. No power - no Internet.

Power was restored piece by piece and I now find Eng-Tips up and running again. My question is still valid: What happened?

Glad to see you again!

Gunnar Englund, Sweden

 

[lukedric]

I toil for the building services sector and I have never worked for any utility. But back in the 80's we were doing studies on the steady stability of power systems and we were able to look at the '65 and '77 outages as voltage collapse in the steady state sense.

Most of the discussions above have so far focused on the dynamic state of the power failure as a result of frequency, voltage relaying etc failures.

If I remember correctly, we were able to show in our study that dynamic collapse occurs after the steady state stability is compromised. The compromise by overburdening of the system can be in the form of loading the system over its network capacity, generation capacity, network component failure among others.

Even FirstEnergy refuses to be blamed for the power failure stating that voltage fluctuations in the western part of the grid were noted on Thursday morning. Large swings in power flow in the 1000MW magnitude were also noted in some transmission lines.

My question to the power utility sector or NERC people is whether steady state stability is being used to measure the reliability of systems. One of the tools that can be used is to assess the singularity of the Jacobian without going through the cumbersome state estimators/ load flow analysis. Could it be that the NE grid is overburdened that there are actually two solutions to the power flow which are 1000s of MW apart? And a little perturbation will cause the system to settle from one solution to the other solution.

Just my guess.

 

[Shortstub]

I can only relay a story I heard from a nephew of mine, who, as a one of a team of firemen, was called upon to rescue people trapped in an elevator. They had to burrow thru a wall to gain access. Upon entry into the darkened elevator, and because of his chivalrous concern for the women inside, he asked, "Any pregnant ladies in here?" A man's voice from inside replied, "Hell, we haven't even become acquainted yet!"

 

[SidiropoulosM]

Lukedric, the Western Electricity Coordinating Council (WECC) mandates reactive margin studies, in addition to power flow and stability. The objective of reactive margin studies is to determine how far the system is from voltage collapse. This is done by computing the Q-V curve for critical busses and for all critical contingencies. The Q-V curve is a plot of reactive power injected into a bus vs. the resulting bus voltage. It has a U-shape and the bottom of the U is typically below the x-axis. The two points of intersection with the x-axis are the two possible solutions of the system. If there is reactive deficiency at a certain bus, the two points will be close together. In such a system, the operating point can move from one point to the other, for small changes of operating voltage. This is voltage instability which may lead to system collapse.

The reactive margin method has been advocated by the WECC following the two 1996 western blackouts and, to my knowledge, it has not been used in the eastern system yet.

 

[lengould]

Way back was mentioned "illegal to disconnect from grid to salvage a local island". Makes sense at the micro level, e.g. individual generators, but ridiculous at the regional level. There surely should have been automatic isolation at the interties from Ontario to Michigan, and New York to Ontario. If Michigan goes black surely Ontario and New York should be capable / allowed to "island" themselves? My quick glance at a line diagram indicated what I guessed to be thyristor load managers at all major points except Niagra. Shouldn't that be the first (immediate) step? Place thyristor connections at Niagra, then REQUIRE systems to AUTOMATICALLY (under a set of NERC-determined conditions) back these down first before exporting one area's problems into another area?

At least might stop NY Mayor from going on CNN and doing another "blame it on Canada" again. Too bad, 50 million CNN viewers get that information, then 50 people follow a thread like this to find out a bit more.

 

[margaret695]

You guys have got me to resort to looking at text books, etc. Which is good, eh.

I also did some searches on the internet and found this. I find tripping in relationship to the rate of change for frequency criteria interesting.

http://www.transmission.bpa.gov/orgs/opi/Power_Stability/EmergencyControlHydroQuebec.pdf

Margaret

 

[pablo02]

The problem with islanding is that the island load must be sufficient and STABLE enough for a generating unit to remain on-line regardless of fuel type -- as for nuclear units, in addition to the load, there are additional (NRC) requirements for operating the units -- which if not met, they must shutdown... I have elected not to go into these requirements, but I suspect they were also a factor..

 

[SidiropoulosM]

Pablo02, if islanding is designed into the system, it must be combined with automatic load shedding. A computer-based special control system (SPS)will determine the load that must be shed to match the available local generation, resulting in a self-sufficient and stable island. Load areas that lack local generation will have to be blacked out until the system recovers. This should motivate utilities to add local generation at load centers, reducing their reliance on incoming transmission lines.

 

[lukedric]

SidiropoulosM, I think it is unfortunate that an authority will only mandate reactive power margins after the collapse of the power system. Let us hope that NERC and N-E council(?) will follow and include reactive power margins to designate their system as stable.

My utility, which also shutdown last week, routinely depresses its system voltage in times of high demand. I think this is one of the optimization techniques used by the deregulation experts. This measure eats up the reactive power margins of the network with less capacitive power generated and more reactive power taken by the transmission lines due to increased current flows.

 

[pablo02]

Sidi: The system is set up with areas to serve as islands (and what areas of load) and then there are sub-islands, and even sub-sub-islands depending upon what the relay scheme is -- selecting what load is out there is general -- and it will change somewhat hour by hour -- and the market now dictates what unit load is or even if a unit is running... there is little monetary motivation to add local generation unless it beats the market... (and the regulators approve)

yes there are load shedding schemes involved, but these, to the best of my knowledge are given areas (usually determined by location of substation breakers)with a semi-known load and not specific MW (although there are areas where the load is somewhat constant) -- and these load areas change all the time (we call it urban sprawl)..

and (I have been in a disturbance that resulted in a black-out) just because you are shedding load and trying to or actually islanding, the significance of the disturbance may or may not let you control the outcome (big islands are easier to handle than the smaller ones) -- in the case of PJM interchange, they were far enough away and the disturbance had't reached the proportions of closer in, they were able to open the appropriate breakers in time to isolate themselves... and help stop the blackout at their borders...

the answers seem easy, but in some cases, you're looking at a moving target and the best scheme is a shotgun approach that will probably not protect in all cases... just how many $$ do we spend? I think there are many short comings to the current schemes that will be addressed; there are a few that may be swept under the rug, and those with political agendas will still have a voice in the outcome..

 

[SidiropoulosM]

Lukedric, any criticism on WECC is unjustified. Voltage instability was not a well understood phenomenon until recently. If anything, the WECC deserves credit for learning from the 1996 blackouts and for pioneering the reactive margin methodology, which was unknown at the time. It will be interesting to see if voltage instability was the cause of the recent blackout anf if the WECC method will be adopted by the eastern system.

Pablo02, I don't know what you mean by "semi-known loads". We have sufficient metering to know what the loads are at any time of the day or night. Load shedding schemes trip distribution or subtransmission lines that feed load substations. The proposed special protection scheme would know what the loads are from metered quantities. It's not as complicated as you think.

 

[alehman]

A brief exerpt from NERC's 2003 Summer Assesment Report (dated May 2003) in regards to Eastern and Central Area Reliability Coordination Agreement (ECAR) which includes MI, IN, OH, WV and KY, on the subject of transmission -

"There is a continuing need for the reliability coordinators, transmission planners, and operators to communicate and coordinate their actions to preserve the continued reliability of the ECAR system. It is anticipated that the ECAR system could become constrained as a result of unit unavailability and/or economic transactions that have historically resulted in large unanticipated power flows within and through ECAR.If these conditions occur again this summer, local operating procedures, as well as the NERC Transmission Loading Relief procedure (TLR) will need to be invoked in order to maintain transmission system security. As long as trnasmission limitations are identified and available operating procedures are implemented when required, no cascading events are anticipated."

It goes on to discuss some improvement projects that are planned or underway in the ECAR region and advises that until the upgrades are completed, reliability of the system will depend on operating procedures.

It would be interesting to know if any TLR's were in effect at the time.

 

[margaret695]

I just did an internet search for information on the reactive margin methodology that SidiropoulosM mentioned. Found this techical paper that relates low inertia compressor motors for air conditioning, heat pumps and refrigeration, electronic power supplies, higher transmission and increasing use of capacitor banks, to problems with voltage stability.

http://www.transmission.bpa.gov/orgs/opi/Power_Stability/PwrElectPanelAmSC.pdf

Me thinks that perhaps fasting tripping on some types of loads should be considered.

Margaret

 

[jstickley]

alehman --

Yes, there were TLR's in effect at the time of the cascading outage. See:

http://www.nerc.com/~filez/Logs/index.html

Keep in mind that there were also several transmission outages that occurred within a short time frame which triggered the event. TLR helps to prevent overloading of facilities, and typically is implemented for known areas of concern (facilities which have been observed to overload either through real-time operation or through studies, or flows / outages which can cause voltage or stability concerns), although it can be implemented for new facilities as needed (through temporary flowgates). I believe this event to be far beyond anything which was studied or anticipated. I don't believe this event could have been prevented easily, certainly not without massive preemptive shedding of load.

All --

A major lesson to learn from this event is that less-probable contingency studies must be performed, and linked to relay setting analysis. Although this event appears to be more related to voltage stability than to overloading, the lesson is still applicable -- how does relaying react to problems on the system, especially those due to unlikely, but plausible, outages (parallel lines or transformers, bus outages, etc.)? It's quite possible that some facilities have relay settings (such as backup 3-phase time overcurrent relays or zone 3 distance relays with settings that approach load levels) that will trip the facility when it's overloaded, especially if the overload is large or if it persists for some time. Such tripping, if not coordinated, can simply transfer the problem to other facilities, which could also then trip.

 

[SidiropoulosM]

Jstickley, you make a distinction between voltage instability and overloading. We should keep in mind that voltage instability is a direct outcome of heavy loading. Transmission lines produce reactive power when loaded below their surge impedance loading (SIL). They absorb reactive power when loaded above their SIL. The heavier the loading is, the more reactive they absorb and the greater the reactive deficiency...and the greater the voltage sag along system busses. There is a loading limit which, if exceeded, will push the system into voltage instability, due to a huge reactive deficiency. For high voltage lines this limit is usually well below the thermal limit of the lines.

 

[jstickley]

SidiropoulousM --

My concern about clarity earlier is coming back to bite me. Should have reread my post.

I certainly did not mean to imply that voltage instability and loading issues are unrelated -- they most definitely are. However, in general, here in the Eastern Interconnection, a good portion of transmission lines are capable of exceeding their thermal limit without much concern about voltage stability (true for lower voltages primarily, up to about 345 kV -- higher voltage facilities are a different matter).

What I intended to convey through my post was that sometimes it's better to overload a facility than to trip it out of service. Tripping heavily loaded lines can possibly cause overloads on other facilities and start a cascading outage (due to relaying or voltage collapse, either could be triggered in this manner). I guess the bottom line is this -- when lines are heavily loaded, that's when they're needed most, and when you least want them to trip.

What I'm getting at (this could be a completely new thread, relaying purpose / philosophy) is that preventing overloads is not the primary purpose of relaying -- that's an operational issue, one that must always be a priority. Protective relaying is intended to prevent damage to equipment due to faults. Of course, with anything, there are exceptions (special protective systems are one example).

 

[SidiropoulosM]

Apparently, PJM has an automatic islanding scheme that was instrumental in keeping PJM alive during the blackout. For an interesting description of control room activity just before the blackout, visit:

http://www.energycentral.com/sections/news/nw_article.cfm?id=4081982

Jstickley, I agree that tripping a heavily loaded line is not a good idea...it will likely overload other facilities and may lead to cascading. We don't do that here in the west. We just cut schedules. Non-firm (interruptible) loads are cut first. Higher priority loads may be also cut, if necessary.

On your last paragraph, I agree that the purpose of protective relaying is to protect equipment against faults. We usually make a distinction between protection systems and the so-called "special protection systems", which here in the west we call remedial action schemes (RAS). We have several RAS schemes in place. Their purpose is usually to take specific action for specific contingencies in order to prevent instability or collapse. I like to think of them as "system protection" as opposed to "equipment protection".

 

[lengould]

So you guys have got me interested. Investigating SIL, I found several. One is at

http://www.ece.nmsu.edu/~sranade/Ee535l5new.PDF

From my "interpretation" of this, it appears that due to a variety of means of developing impedance in AC power lines, a line's capacity and therefore the cost of transmission change significantly above about 250 miles (at certain sufficiently high volatges).

Does this shed any light (for me) on the recent e.g. Enron scandal(s)? Were they trying to avoid a SIL based penalty charge for a long distance transmission (e.g. Texas to Cal) by selling blocks to "theoretical" intermediate customers at 250 mile intervals who then re-sold to the next "theoretical" intermediate customer 250 miles further on, until the block of power arrived at Cal?

 

[jstickley]

lengould --

This is off-topic, but I'll answer. Perhaps forum admins can move this to another thread.

SIL is totally unrelated to the Enron scandal. Enron was basically manipulating the market to "create" congestion (and thus increase the value of their energy trades). In essence, they took advantage of loop-holes in market rules that allowed them to artificially inflate prices to benefit themself. I'm sure there are reports available on the web to further describe their actions (a simple google search such as

http://www.google.com/search?hl=en&ie=UTF-8&oe=UTF-8&q=enron+market+manipulation

brings up a lot of links, although you still have to sort through the junk).

Although I haven't read the paper you posted, the cost that you're reading about is surely in regard to cost of building transmission. Basically, as the length of a transmission line increases, so does the cost of building the line, the cost of the losses absorbed by the line, and the cost of supplying reactive power (via capacitor banks, etc.) to support voltage on the line. At some point (as I recall, about 250-300 miles according to most studies/textbooks), the economics dictate that it's better to build high-voltage DC rather than high-voltage AC (DC lines are cheaper to build as compared to AC lines due to materials required, but require expensive converter stations to interface with the grid that bump up the cost quite a bit).

As far as I know, there is no transmission tariff out there that addresses SIL in any way -- typically tariff rates are dictated by the revenue requirements of the grid (determined by value of facilities, costs of operation and maintenance, etc.).

 

[jstickley]

Interesting story I just ran across on CNN. Looks as if the 345 kV outages identified in NERC's preliminary report caused massive overloads of other transmission lines in the system, leading to cascading outages of the facilities.

http://www.cnn.com/2003/US/Midwest/08/22/blackout.investigation.ap/index.html

Of course, NERC's investigation is just beginning (I understand massive data requests have been made to utilities in and surrounding the area of the blackout). It will be interesting to see what specific event triggered the cascade and what could have been done to isolate the problem.

 

[Modula2]

What about the situation with Hydro Quebec, which is a major energy supplier to North Eastern US areas including New York. They seemed to continue without outage, whereas the adjacent province of Ontario was without power. Major cities such as Montreal were unaffected, whereas everyone in the same region of North America was down. I read a statement that they were deliberately out of phase with the rest of the region. What does that mean?