Protective Device Coordination Study

NOTE This standard governs the protective device coordination study as an engineering deliverable: the engineer who performs it and when it is delivered, the degree of coordination the system is held to, the criteria by which two devices in series are judged to coordinate, the equipment curves every setting is bounded by, the setting recommended for every adjustable device and the fuse and overload class selected where the study decides them, the deficiencies the system cannot avoid and who accepts them, the record of settings handed to the field, the report, and the events after which the study is repeated. (1.1)

NOTE A coordination study plots the time-current characteristic of every overcurrent device in series between a source and a load on one logarithmic sheet, together with the withstand curves of the transformers, conductors, and motors between them and the fault currents the short-circuit study calculated at each bus, and then chooses each device's settings so that the device nearest a fault opens first and the devices above it stay closed. NFPA 70 240.12 names the orderly shutdown that coordination buys, 700.32, 701.32, 708.54, 620.62, 645.27, and 517.31(G) require full selectivity on the circuits they govern, and on every other circuit the degree of coordination is a design decision this standard asks the project to make in writing rather than discover at the first fault. (1.2)

NOTE The coordination study is the second of three studies performed on one system model. The short-circuit study under Short Circuit StudyShort-Circuit StudyResolves to the current adopted revision.sync/short-circuit-study builds the model and calculates the maximum and minimum fault current at every bus; this study takes that model and sets the devices; the arc-flash study under Arc Flash StudyArc-Flash StudyResolves to the current adopted revision.sync/arc-flash-study takes the settings this study finalizes and turns the clearing time of each device into incident energy. A setting changed here moves the arc-flash labels, which is why the final settings are issued to the arc-flash engineer as a deliverable of this study and not read from a drawing. (1.3)

NOTE The following are governed elsewhere and are outside this standard: (1.4)

NOTE The criteria statement is a page, and it settles the arguments the report would otherwise start: whether an overlap in the instantaneous region is a deficiency or an accepted condition, whether 0.3 s or 0.4 s separates two relays, and which circuits must coordinate at every current. A report reviewed against criteria the reviewer never agreed to is reviewed twice. (3.1.3)

NOTE The settings a system runs on are the ones in the trip units, not the ones in the report, and the two drift apart on the day a technician cannot reach a recommended value and picks the nearest one. A closeout report revised to the as-left values is the only document that describes the system as it stands, and the model that carries those values is what the next engineer opens when a breaker is replaced. (3.2.3)

NOTE The three studies are built on one model, and the engineer who built it for the short-circuit study already holds every impedance, every device, and every assumption this study needs, which is why one engineer performs all three on nearly every project and the short-circuit study standard is where that engineer is chosen; the Engineer of Record performs the coordination study separately where the settings decide the design, as on a system whose selective coordination governs the choice between power breakers and molded-case breakers; an independent engineer engaged by the Contractor is the arrangement where the coordination study is priced as a construction deliverable and built from the devices actually furnished; the equipment manufacturer's engineering services know their own trip units and their own selectivity tables best and are the only party with the published data for some pairs, and their settings for other makers' devices and for the relays rest on the data handed to them; the testing organization's engineering group puts the settings and the test set in one firm's hands, so that the party injecting each trip unit computed the value it is set to. (4.1.2)

NOTE A coordination study drawn by hand or in general plotting software cannot be checked against the published characteristic of the device, and a study run on a model that differs from the short-circuit study's gives a clearing time at a fault current that was never calculated. One model, one library, and one chain of custody from the first study to the arc-flash label is the whole reason the three studies are specified as a sequence. (4.2.6)

NOTE Stopping at the main of each panelboard sets every feeder and every main and leaves the branch devices to the manufacturer's curves, which is the extent of a study whose purpose is the settings of the adjustable devices; carrying the study through the largest device in each panelboard and through the branch device of each motor large enough to be modeled on its own checks the pair that is most likely to overlap, the largest branch breaker against the feeder above it, and costs one more curve per panelboard, which is why most studies draw the line there; plotting every branch device, down to the lighting and receptacle circuits, is the extent that the selective coordination articles of NFPA 70 require on the circuits they govern, and on normal distribution it multiplies the sheets without changing a setting. (4.3.2)

NOTE Selectivity for the full range of overcurrents, as NFPA 70 Article 100 defines selective coordination, means that at every current from the lowest overload to the maximum available fault current the downstream device clears before the upstream device begins to open, which on low-voltage systems is reached with power circuit breakers carrying short-time delay, with fuses selected by ratio, or with molded-case breakers whose tested selectivity tables cover the pair, and which is the degree the code requires on the circuits it names; selectivity beyond 0.1 s ignores the overlap in the instantaneous region of two molded-case breakers for faults that clear in less than 0.1 s, on the ground that a fault at the available current clears so fast that the loss of the upstream device is accepted, and it is the degree NFPA 70 517.31(G) states for the health care essential electrical system; selectivity beyond 0.01 s ignores only the overlap below one half cycle, where the instantaneous elements of two breakers both unlatch, and is the degree a project chooses where it wants selectivity through the short-time region without buying the selectivity tables; coordination to the extent the devices permit sets every adjustable device as well as the system allows, documents each overlap with its current range, and is the degree a normal distribution system is held to where an upstream trip on a bolted fault is an inconvenience and not a hazard. No one degree is the norm for normal distribution, and the choice turns on what a fault on a feeder is allowed to take down. (5.1.2)

NOTE A device that does not operate for the minimum fault at the end of its circuit leaves an arcing fault to burn until the next device upstream sees it, and the study checks the low end of every curve for that reason as carefully as the high end. The minimum case of the short-circuit study, on the generator alone and with the motors removed, is where the far-end fault is smallest. (5.1.4)

NOTE Which circuits must be selectively coordinated is not a choice the study makes; it follows from the systems the one-line diagram shows and the articles that classify them, which the emergency power standard and the health care standard settle. The study takes the classification and demonstrates the coordination. (5.2.2)

NOTE The exception for two devices in series with no load connected between them is the common case: a feeder breaker and the main of the panelboard it feeds alone remove the same loads whichever one opens, so the code does not ask them to coordinate with each other, and a study that treats that pair as a deficiency finds one where the code sees none. (5.2.5)

NOTE The interval between two relays is built from parts: the opening time of the upstream breaker, about five cycles on a medium-voltage breaker; the overtravel of an induction disc that keeps turning after the fault is cleared, about 0.1 s; the tolerance of the relay and its current transformer; and a safety margin. IEEE 242 arrives at about 0.4 s for an electromechanical relay and about 0.3 s for a microprocessor relay, which has no overtravel, and those are the figures most studies use without discussion; an interval of 0.2 to 0.25 s is used where the relays are field-calibrated and the breaker opening time is known, and every 0.1 s saved at one level is 0.1 s saved at every level above it; a longer interval buys margin against an uncalibrated relay at the cost of a slower main and a higher incident energy at every bus the main protects. (5.3.2)

NOTE A fuse's total clearing curve already includes its arcing time, and the upstream relay need only not have operated by the time the fuse clears, so the opening time of the upstream breaker drops out of the interval; what remains is the relay's overtravel and tolerance, about 0.2 s for an electromechanical relay and 0.1 to 0.15 s for a microprocessor relay. (5.3.4)

NOTE The published band of a low-voltage trip unit already contains the tolerance of the sensor, the trip unit, and the mechanism, so two bands that do not touch are coordinated with no further allowance, and a separation of zero is the practice on nearly every study; a project adds a separation where the trip units are of different makes whose bands were measured under different test standards, or where it wants a visible gap on the sheet for the reviewer. (5.3.6)

NOTE A fuse that has been carrying load is part way to melting before the fault arrives, and a fuse in a hot enclosure melts sooner still. The published minimum melting curve is drawn for a cold fuse at room temperature, so a study that compares it unadjusted shows a margin the fuse does not have. (5.3.9)

NOTE The published curve of a molded-case breaker stops at the instantaneous pickup, and two such curves overlap at every current above the lower pickup, yet the pair is often selective in a test because the downstream breaker is current-limiting and clears before the upstream element unlatches. Manufacturers publish selectivity tables, built from tests of each pair, that state the current up to which the pair is selective, and nearly every study accepts them because a system of molded-case breakers cannot be shown selective any other way; a project that accepts the curves alone treats every instantaneous overlap as a deficiency, which is the posture of an Owner who does not want selectivity to depend on both breakers staying the make and type that were tested. (5.4.2)

NOTE Every second of delay a main is given so that a feeder can clear first is a second an arc on the main bus burns, and the two studies that share this model pull the same setting in opposite directions. Holding selectivity and reducing the energy another way, with zone-selective interlocking that drops the delay when no feeder sees the fault, a maintenance switch during work, or differential relaying, keeps the system up through a feeder fault and is the posture of a facility that cannot lose a bus, at the cost of the arc energy reduction equipment the equipment standards carry; holding the incident energy lets the main trip on a feeder fault in exchange for a lower category at every piece of equipment on that bus, and is the posture of a facility whose work is done live and whose feeders can be lost; resolving each pair with the Engineer of Record is the arrangement where the buses differ, a data hall on one feeder and a loading dock on another. The incident energy itself is calculated under the arc-flash study, which is why this field decides the priority and not the number. (5.5.2)

NOTE A maintenance switch lowers the pickup only while a worker remembers to turn it on, and the study coordinates the system for the other 99 percent of its life; a zone-selective interlock pays the short-time delay only when a downstream trip unit reports that it sees the fault, so it is the one method that gives up nothing in either study, and it needs the trip units above and below to be wired to each other, which the equipment standards carry. (5.5.7)

NOTE A utility that furnishes the curve of its fuse or recloser usually asks the customer's service device to clear first, so that a fault inside the building does not open the utility's device and take the neighbors with it, and the service main is then set below the utility curve as the first constraint of the whole study; a utility that furnishes nothing, or one whose device is a substation breaker set far above the service, is plotted for information and the service device is set from the equipment it protects. The utility's curve is the one input this study owns outright, because the short-circuit study needs only the utility's fault current and X/R and never asks for the curve. (6.1.3)

NOTE The guides publish one curve for a transformer whose secondary rarely sees a through fault, which is the transformer inside a building feeding switchgear and cable, and a lower curve for a transformer whose secondary feeds overhead lines or a process where faults are a matter of routine, on the ground that the winding's mechanical life is spent one through fault at a time. The infrequent curve is the one nearly every building transformer is protected to, and the frequent curve is applied to a unit whose secondary serves exposed conductors or a load with a history of faults. (6.2.3)

NOTE A transformer energized at the wrong point on the voltage wave draws a magnetizing current that a primary fuse or relay cannot tell from a fault, and the inrush point is where the study makes sure the primary device rides through it. Twelve times full-load current for 0.1 s is the figure IEEE C37.91 gives for most distribution transformers and the one nearly every study plots; 8 times is the figure for a small unit or one the manufacturer has measured lower; 25 times is the first half-cycle peak that a fast fuse or an instantaneous element on a large unit is checked against at 0.01 s in addition to the 0.1 s point. Where the manufacturer publishes the inrush of the unit furnished, that value governs. (6.2.8)

NOTE A motor's overload relay is set from the motor's nameplate, and the trip class is selected from how long the motor takes to start; the study does not choose either from the code limits alone, because a relay that meets 430.32 and trips on every start protects nothing. The equipment standards select the class the controller is furnished with, and this study is where the selection is checked against the motor that is actually driven. (6.3.7)

NOTE A conductor's damage curve is the limit the device above it must clear beneath, and the conductor that is at risk is the one whose protecting device waits: a feeder from a main with a short-time delay, a service conductor under a relay with a time dial, or a motor branch circuit whose instantaneous-trip breaker is set high for the inrush. Plotting every service conductor and feeder covers the conductors large enough to matter and the devices most likely to be delayed, and is the extent most studies plot; adding the motor branch circuits covers the conductor under a high-set instantaneous-trip breaker; limiting the plot to conductors under a delayed device plots only the runs that can fail the check and is the extent of a study with hundreds of feeders under instantaneous devices. (6.4.2)

NOTE A generator supplies a fault current that falls within seconds to a value near its full-load rating, and a feeder device set on utility power to trip at ten times the generator's rating may never see that current on generator power. Every device the generator feeds alone is checked against the decrement curve for that reason, and the check is where a shunt-excited alternator or a thermal-magnetic generator breaker is found out. (6.5.6)

NOTE The fire pump device is sized to carry the locked-rotor current of the pump motor indefinitely, which produces a curve that protects nothing downstream and is not meant to; the code chooses a burned pump motor over a dark sprinkler riser. The study's job on that circuit is to keep every device above it from opening first. (6.6.3)

NOTE NFPA 70 517.17(C) requires a six-cycle minimum separation between the operating time bands of the two levels of ground-fault protection on a health care essential electrical system, and six cycles is the separation nearly every two-level scheme is set to, because the service device is already held to one second at 3000 A and every cycle added below it is taken from the feeder; a longer separation is used where the two devices are of different makes whose bands were measured differently, or where a third level below the feeder needs room of its own. (7.1.4)

NOTE A ground-fault relay at the service with no second level below it opens the whole building for a ground fault on any branch circuit that the branch breaker does not clear first, and the second level is what keeps a fault in one feeder inside that feeder. The service relay is set as slow and as high as the code allows for that reason, and the feeder relays are set inside the room it leaves. (7.1.7)

NOTE The moderately, very, and extremely inverse curves of IEEE C37.112 are the shapes the North American fuses and electromechanical relays were drawn to, so a microprocessor relay set on them coordinates with the fuse below it and the utility relay above it on the same geometry, and they are the family nearly every study in this market uses; the IEC 60255 curves are the family of a system whose other relays, or whose Owner's engineering practice, are written to IEC, and their standard inverse is flatter than any IEEE curve at high current; a relay set to emulate the electromechanical relays elsewhere on a campus, where the relay offers that characteristic, lets one coordination sheet hold old and new relays on one shape, at the cost of a curve the next relay may not offer. (8.2.3)

NOTE A fuse has no setting, so the only thing the study can do with one is choose it, and the choice is made before the equipment is ordered or not at all. A fuse clipped in later because it was the one in stock is the most common way a coordinated system loses its coordination without a drawing changing. (8.3.4)

NOTE Most real systems contain a pair that does not coordinate at some current, usually a main and a feeder whose instantaneous elements both see a bolted fault at the feeder's terminals, and the study is not a declaration that none exist. It is the list of where they are, how much current it takes to find one, and what it costs to remove it, so that the Engineer of Record decides with the numbers in hand rather than the Owner finding out at the first fault. (8.4.7)

NOTE The consolidated table is what the Engineer of Record reads to see the system coordinate, and the per-device sheet is what the technician carries to the breaker and signs, so nearly every study delivers both; the native settings files are added where the relays and trip units are programmed by file rather than by keypad and the Owner wants the file that was loaded to be the file the engineer wrote, at the cost of the study engineer holding the configuration software of every make on the project. (9.1.2)

NOTE The study is the only authority for a setting; the acceptance testing standard says so from the field's side, and this standard says so from the engineer's. The sheet the technician signs and the table the engineer sealed have to be the same numbers, and the five-day revision is what keeps a device that will not take its setting from being left at a value no one calculated. (9.2.5)

NOTE An arc-flash label is a clearing time dressed as a number of calories, and the clearing time is a setting from this study. A label calculated on a design setting that the field never applied, or on a setting this study revised the week after, is wrong by exactly the difference, which is why the settings are issued to the arc-flash engineer as a dated deliverable and not read off a drawing. (9.3.3)

NOTE A curve sheet is the argument for a setting, and a sheet that leaves off the fault current markers or the damage curves is a drawing of two curves that do not touch, which proves nothing about the system they are installed in. The reviewer reads the markers first. (10.2.5)

NOTE Coordination is lost one device at a time, and almost never by a decision: a breaker replaced from stock with the same frame and a different trip unit, a fuse of the right amperage and the wrong speed, a relay firmware update that renamed the curve. None of them look like a reason to call an engineer, which is why the list is written in the words of the work rather than the words of the study. (11.2.5)