One of the most confusing subjects faced by utility distribution engineers is distribution neutral grounding. This confusion is compounded by utility mergers and the combining of construction practices. More and more, you'll hear questions like: Is good grounding really necessary? Does poor grounding have advantages? What is the best grounding? When is grounding important and when isn't it? This article will attempt to answer some of these questions, as well as demonstrate that while good grounding is usually preferred, it is sometimes unimportant or even detrimental.
There are a number of ways to ground capacitor banks. Engineers normally use grounded wye banks, but this may not always be the optimum choice. Typically, grounded wye banks are used on 4-wire multigrounded systems only. A grounded wye bank on an ungrounded system creates a ground source that can interfere with sensitive relaying and contribute to overvoltages during ground faults. It's true that grounded wye banks are generally easy to clear because there is adequate ground current. On the other hand, ungrounded banks have currents limited to 300% of normal phase currents by the impedance of the other two legs.
One alternative involves connecting a 3-phase capacitor in delta, wye-ungrounded, or wye-grounded. Delta or ungrounded wye banks offer the greatest possibility of neutral inversion or resonant conditions when one or two conductors on the source side of the bank are open. Consequently, it can be a problem to locate these banks on the load side of a switch or fuse.
Swells are steady-state overvoltages caused by faults on adjacent phases. The duration of these overvoltages depends on the protection practices used by the utility. Swells can result in power quality problems as well as arrester failures.
Our studies show that a ground footing resistance of less than 1 ohm is required for a typical 4-wire system to achieve an arbitrary swell limit of 20% (this is the value used for arrester applications by many utilities). A footing resistance of 25 ohms produces overvoltages (near the end of the line) of about 1.31 per unit for the same system. Using a ground footing resistance of 25 ohms does reduce overvoltages for faults within about 5 miles of the substation, as compared to 100 ohms. However, for faults beyond 5 miles, overvoltages are not reduced and are virtually equal.
Our study results also indicate that using the standard 4-grounds-per-mile is not sufficient to keep overvoltages down to the desired level (see Fig. 1). If the number of grounds increases to 8 per mile, there will be a reduction of about 2% with a footing resistance of 25 ohms. Augmenting the number of grounds per mile does not have a big effect on reducing swells. This is especially true when there are many grounds on the system. Even when soil resistivity ranges from 100 ohm-meters to 1000 ohm-meters, no changes occur in the magnitude of the swells.
Finally, let's examine the effects of substation grounding and neutrals on swells. Substation grounding impedances of 0.5, 1.0, 2.0, and 3.0 show minimal effects on swells caused by faults out on the feeder.
Neutrals, however, play a major role in effective grounding. A fault that occurs 10 miles from a substation can cause swells of 1.33 per unit if a broken neutral exists on any part of a system. Even faults occurring 1.5 miles away can cause swells up to 1.5 per unit if a broken neutral exists. The size of the neutral conductor appreciably reduces swells, whereas good grounding hardly affects the voltage at all. This indicates that the neutral is more important than the grounding.
Unbalanced load current flows in the ground and the neutral wire. The ground current creates most of the magnetic field associated with EMFs. Current in the neutral tends to reduce this magnetic field. Under typical conditions, approximately 50% of the return current flows in the earth; the other 50% flows in the neutral. One could make the case that poor grounding forces more current in the neutral and thereby reduces EMFs. Measurements taken by one of the authors on actual systems show that ground impedance is far less of a factor than what many studies show.
Ground rod footing resistance slightly affects fault current levels for close-in faults, but it has little effect on faults that occur more than 4 miles or 5 miles from the substation (see Fig. 2, on page 24). Footing resistance in close-in areas is not an issue because close-in fault magnitudes are almost always sufficient for proper protection operation. Footing resistance, therefore, is not important in the area of overcurrent protection.
Most cases of stray voltage arise from onsite problems, but some can result from a poor utility return path (earth and neutral wire). Stray utility voltage occurs when the return current (or unbalanced 3-phase current) returns via the neutral wire and the ground and produces a voltage that passes to the customer's premises through the distribution transformer connection. The flow of current in these paths is complex and depends on many factors, including distances from substations, number of grounds, footing-resistance values, and neutral sizes. While good ground footing resistances near the affected customer are important, the problem is more affected by the magnitude of the return current and the size of the neutral conductor. Often, reducing the ground footing resistance near the customer proves ineffective for this reason.
Arrester grounding is not as critical as most engineers believe. We've conducted studies that show ground resistance between 0 ohms and 250 ohms have little effect on flashover rates after arresters are put on every phase, tower, or pole (see Fig. 3). As arrester spacing increases, grounding does have a relatively minor influence.
The problem with arresters used for direct stroke protection is that they will most likely fail anyway because of the energy of the stroke. Poor grounding may help the arrester survive because the arrester closest to the lightning hit does not absorb all the energy and shares it with adjacent arresters.
Ground resistance and ground spacing are important factors to consider when using shield wires. Shield wires can be very beneficial if low ground resistances can be achieved. For example, simulations on a standard distribution system revealed that no flashovers occur with a ground resistance of 0 ohms (see Fig. 4). If the ground impedance increases to 25 ohms, about 22% of the hits would cause a flashover; with a ground footing impedance of 100 ohms, over 82% of the direct hits would cause the line to flashover. When you use shield wires, it's essential to put grounds on every span to achieve good protection. During field tests, a sampling of 50 feeders with static wire protection and a significant percentage of poles without static grounds (>15%) revealed a dramatic difference in performance (>50% reduction in lightning related flashovers) when grounds were added to these poles.
There are two types of faults — low- and high-impedance. A high impedance fault is considered to be a fault that has a high Z due to the contact of the conductor to the earth (i.e., Zf is high). By this definition, you would still classify a bolted fault at the end of a feeder as a low-impedance fault.
Low-impedance or bolted faults can be high in current magnitude (10,000A or higher) or low (300A at the end of a long feeder). Normal protective devices will detect all low-impedance faults. These faults are such that the calculated value of fault current assuming a bolted fault closely approximates actual fault current. Most detectable faults show fault impedances close to 0 ohms. This implies that the phase conductor either contacts the neutral wire or the arc to the neutral conductor has a low impedance.
A 15-year-old study from the Electric Power Research Institute (EPRI) indicates that the maximum fault impedance for detectable faults is 2 ohms or less. Fig. 5 shows that 2 ohms of fault impedance influences the level of fault current — depending on the location of the fault. As you can see, 2 ohms of fault impedance considerably decreases the level of fault current for close-in faults but has little effect on faults some distance away. Based on these studies, fault impedances of 30 ohms, 40 ohms, and 50 ohms (used in many computer programs and by many utilities) lack justification.
High-impedance faults are faults that are low in value (generally less than 100A because of the impedance between the phase conductor and the surface on which the conductor falls). Fig. 6 illustrates that most surface areas, whether wet or dry, do not conduct well. If an 8-ft ground rod sunk into the earth results in an impedance of 100 ohms or greater more often than not, then a conductor simply lying on a surface cannot be expected to have a low impedance. High-impedance faults do not contact the neutral and do not arc to the neutral, and they are undetectable by conventional means.
Studies show that good grounding, arbitrarily taken to be 25 ohms, rarely adds to good system performance. Although good grounding is important when considering the effectiveness of shield wire protection, it's far less important for determining fault levels and arrester protection. All in all, the need for good grounding depends exactly on what the distribution engineer is trying to accomplish.
Jim Burke is a consultant for ABB in Raleigh, N.C. You can reach him at jim.j.burke@us.abb.com.
Mike Marshall is a project engineer for Carolina Power and Light in Asheville, N.C. He can be reached at mike.marshall2@pgnmail.com.