Tony Sleva, SEA Consulting Services
Before modern computers became available, it was pretty difficult to develop load flow studies. Therefore, “base load” studies were developed and seat of the pants methods were used to operate a “real time” system using “base load” studies.
These old time methods recognized that there are only four variables in electrical systems—voltage, current, phase angle and frequency. And, when all possible abnormal power system conditions are evaluated, only three conditions are relieved by instantaneous protective relay action (see table)—namely, short circuits, out of step, and low frequency.
Short circuits are cleared instantaneously in order to minimize equipment damage (refer to Figure 1). More recently, generators are tripped instantaneously if they get out of step with the system in order to preserve the windings of the out of step generator (refer to Figure 2). Distribution feeders are tripped instantaneously if frequency decays in order to halt system collapse.
All other abnormal power system conditions are allowed to run their course because sufficient margins are assumed to be in place to allow the system to recover. Spinning reserve is scheduled to cover loss of the largest generator. Off-cost generation is scheduled if least-cost generation will result in equipment overloads upon loss of any single component. Generators are equipped with automatic voltage regulators and automatic speed controls. If the system doesn’t recover within a few seconds, something happens—sometimes by design, sometimes by happenstance.
When a short circuit occurs, equipment damage is evident and corrective action is immediate. When voltage collapses, however, equipment is undamaged and corrective action is hard to identify as the sequence of events may hide the root cause in the blizzard of information that becomes available.
Consider August 14th. Several short circuits occurred in Ohio. Power flow to Ohio was compromised. Generators east of Ohio continued to produce power that flowed through Canada on its way to Ohio. This unintended power flow caused voltage collapse on the power system in Michigan. This caused several generators in the Northeast to trip and eventually resulted in collapse of a large portion of the power grid. (Initial news reports indicate that generators tripped on overspeed. Out of step tripping or loss of excitation seems more likely as power system frequency doesn’t appear to have to have spiked much above 60.25 Hertz. Subsequent reports point to reactor coolant pump trips at pressurized water nuclear reactors.)
Why do generators trip via out of step relaying when voltage collapses? Because mechanical input power remains essentially constant while electrical output power drops substantially. The result is that a generator starts turning a little faster than the power system and it goes out of step with other generators.
If all generators are spinning at 1,800 RPM (mechanical equivalent of 60 Hertz) and all generators speed up to 1,836 RPM (mechanical equivalent of 61.2 Hertz), all generators remain in synchronism and all generators can continue operating indefinitely. If, however, all generators are spinning at 1,800 RPM and one generator speeds up above 1,800 RPM, the one generator that is out of step will self destruct unless it is isolated from the power grid very quickly.
Why do generators trip via loss of excitation protection when severe power swings occur on power systems? Because loss of excitation protection measures generator impedance and, during severe power system swings, generator impedance can be the same as generator impedance when excitation is lost. The hope is that a severe power swing will subside in less than 400 milliseconds and loss of excitation relays will not operate during severe swings.
Why did reactor coolant pumps trip? Probably because motors slow down when voltage drops. If motor slip increased to 5 percent, reactor coolant pump controllers may interpret this as a reduction in power system frequency.
Recognizing that voltage collapse caused the August 14th event, what can be done to prevent recurrence?
“- Short-term solutions: Regional control of real power and reactive power is an absolute necessity. Setting spinning reserve requirements equal to loss of the largest generator should continue. Setting reserve reactive power requirements is much more difficult because, at a given load level, as voltage decreases reactive power requirements increase exponentially. Multiple large generators operating near var output limits must be recognized as an indicator of impending voltage problems.
“- Near-term solutions: Generator ratings should be changed to include operational power factor limits. Most large generators are rated at 90 percent power factor. Up-rated nuclear plants have generators rated at 95 percent power factor. A better idea may be to require that a pool of generators be able to produce power at an equivalent 80 percent power factor.
“- Long-term solutions: Additional low power factor (80 percent), local generation should be developed and dynamic var compensation should be installed near all major load centers.
Islandized regional grids are concepts that need to be explored. PJM (Pennsylvania–New Jersey–Maryland interconnection) appears to have escaped the blackout because high impedance, phase shifting transformers are installed between the Consolidated Edison system and PJM. These phase shifting transformers acted as a buffer that allowed PJM equipment to ride through the problem.
Power system collapse protection that includes development of high speed data highways to centrally located protection centers must be developed. Each protection center must have the ability to initiate high speed trips of remote circuit breakers based on regional, rather than local, conditions. For this to work, data must be received at protection centers within 8 milliseconds of an occurrence. Data processing must be completed and automatic corrective action must be initiated in less than 200 milliseconds of an initiating event while rapid voltage collapse is occurring.
A major concern is that power system collapse protection centers must trip circuit breakers at load centers when problems occur at generation stations or on the lines between generators and load centers. In other words, the best solution to a problem in one are may be tripping loads in another area if the solution is to reduce power transfers.
Instantaneous: No intentional delay, 16 milliseconds or less.
Short time delay: More than 100 milliseconds, less than 400 milliseconds.
Long time delay: More than 400 milliseconds.
Out of step: Not in synchronism.
Low frequency: Less than 60 Hertz.
Very low frequency: Less than 57 Hertz.
Spinning reserve: Instantaneously available reserve generation.
Real power: Watts, kilowatts, or megawatts.
Reactive power: Vars, excitation energy for motors, transformers, transmission lines and distribution lines
Power factor: Ratio of real power and reactive power.
Will more transmission lines prevent future blackouts? Maybe. Transmission lines are power conduits. More transmission lines will allow power to flow across the system if sending end and receiving end voltage are stable.
Will additional generators prevent future blackouts? Maybe, if new generators are located close to load centers and are designed to provide reactive power (vars) as well as real power (watts).
Will distributed generation help? Yes, but first electric utilities must embrace distributed generation. Presently, many electric utilities require distributed generators to trip immediately if voltage drops to less than 95 percent or frequency drops to less than 59 Hertz. This “quick trip” philosophy exacerbates grid disturbances.
Can protective relays be developed that will prevent recurrence of the August 14th event? Yes. When will they be ready for service? 10 or 20 years from now as these new relays will require a completely different paradigm.
Could the August 14th blackout have been avoided if selected loads were automatically tripped in Cleveland, Columbus, Toledo, Detroit, Toronto, and New York City in the initial moments of the voltage collapse? Why certainly, but only if the loads were shed in accordance with commands issued from a centrally located protection center that specifically monitors power transfers and grid voltage collapse.