System Grounding Study and Effective Grounding Verification

1. Executive Summary

The question that started this engagement was deceptively short: is this system effectively grounded? It was asked by an Interconnection Customer whose 74.9 MW inverter-based generating facility had been issued an interconnection agreement containing a clause requiring the facility to "provide an effectively grounded source at the point of interconnection at all times the facility is energized." The Customer's engineering, procurement and construction contractor had priced the plant without a grounding transformer, had specified 80 percent rated surge arresters and 100 percent insulation level medium-voltage cable, and had assumed that because the utility's 138 kV system was solidly grounded, the 34.5 kV system behind it was too. None of those three assumptions survived analysis.

Keentel Engineering Solutions was retained by the Customer, with the host utility participating as a reviewing party, to answer the grounding question completely rather than partially. The engagement ran from Month 0 to Month 9 and deliberately treated grounding as two separate disciplines that share a word and nothing else . <em>System grounding</em> concerns how the system neutral is connected to earth, and it determines ground fault current magnitude, the overvoltage that appears on the unfaulted phases during a ground fault, the ability to detect and locate that fault, and therefore arrester ratings, cable insulation levels and relay settings. <em>Equipment and substation grounding</em> concerns the conductive grid buried in the earth, and it determines ground potential rise, touch voltage and step voltage, and therefore personnel safety. The two are coupled only through the fault current that the first produces and the second must discharge. Conflating them is the most common and most expensive error in this field, and the study report was structured to make the separation impossible to miss.

The system grounding analysis found that the host utility's 34.5 kV system was not effectively grounded at the point of interconnection. The as-found ratio of zero-sequence to positive-sequence reactance, X0/X1, was 4.70 against the criterion of 3.0 or less, with R0/X1 of 0.60. Worse, when the utility's 34.5 kV source breaker opened to clear a line-to-ground fault, the plant became an ungrounded island because every inverter step-up transformer presented a delta winding to the 34.5 kV system. Electromagnetic transient simulation of that island produced a sustained 1.63 per unit voltage on the unfaulted phases, a coefficient of grounding of 0.94, and an arrester energy duty that exceeded the class rating of every arrester on the circuit within 400 milliseconds.

A 34.5 kV zig-zag grounding transformer, located inside the plant's point-of-interconnection breaker so that it remains with the island, brought X0/X1 to 2.60, the coefficient of grounding to 0.77 and the unfaulted-phase voltage to 1.34 per unit. That permitted 80 percent rated arresters with 22 kV maximum continuous operating voltage instead of 100 percent rated arresters at 29 kV, and 100 percent insulation level cable instead of 133 percent, releasing 2.1 million dollars of avoided capital against a 310 thousand dollar grounding transformer installation. The cost of the answer was a rise in available single line-to-ground fault current at the point of interconnection from 1,150 A to 4,380 A, which broke the utility's existing ground overcurrent coordination and required a revised directional ground scheme delivered as part of the study.

The substation grounding analysis, performed to IEEE Std 80 for the utility's 138/34.5 kV substation, produced a grid resistance of 0.40 ohms, a ground potential rise of 3,760 V, a mesh voltage of 512 V against a tolerable touch voltage of 654 V, and a step voltage of 198 V against 2,080 V tolerable. Fall-of-potential testing after construction measured 0.41 ohms. Because ground potential rise exceeded the 300 V threshold, all metallic telecommunications entering the station were replaced with an all-dielectric fiber-optic link per IEEE Std 367 and IEEE Std 487.

Figures presented are representative of the delivered study and have been generalized to protect client confidentiality.

Study at a Glance

2. Background and Study Drivers

The Interconnection Customer had reached the design freeze milestone on a 74.9 MW alternating-current photovoltaic facility with a 20 MW / 80 MWh battery energy storage system sharing the same collection network. The facility exports through a 2.5 mile 34.5 kV overhead and underground tie as part of the grid interconnection engineering scope to the host utility's 138/34.5 kV substation, where it terminates on the 34.5 kV bus behind a dedicated point-of-interconnection breaker.

Three separate events converged to trigger the study.

The interconnection agreement clause. The executed agreement contained a standard but often-ignored requirement that the generating facility present an effectively grounded source at the point of interconnection whenever energized, with effective grounding defined by reference to the ratios X0/X1 and R0/X1. The Customer's contractor had read the clause as satisfied by the presence of a grounded-wye winding somewhere in the plant. In fact every inverter step-up transformer in the collection system was 34.5 kV delta on the high side and 0.69 kV grounded-wye on the low side, a configuration chosen for zero-sequence isolation of inverter switching harmonics and for third-harmonic circulation, and there was no zero-sequence source anywhere on the plant side of the point of interconnection.

An arrester failure on an adjacent circuit. During the design period the host utility experienced a cascade of surge arrester failures on a neighboring 34.5 kV circuit that also served an inverter-based facility. Nine arresters failed within a two-second window following a single line-to-ground fault, and two 34.5 kV cable terminations failed subsequently. The utility's post-event analysis attributed the failures to ground fault overvoltage in the ungrounded island created when the source breaker opened. That event turned a paperwork clause into a live engineering condition, and the utility subsequently declined to energize any new inverter-based facility on its 34.5 kV system without a grounding study.

A substation expansion. The 138/34.5 kV substation required a new 34.5 kV bay, a second 138 kV line position and a grid extension. The utility's existing ground grid drawings dated from an era when the station carried roughly half its present fault duty, no soil resistivity data existed in the record, and there was no evidence that a formal IEEE Std 80 analysis had ever been performed. The expansion made a grid study unavoidable, and it was efficient to perform it in the same engagement, provided the two disciplines were kept analytically separate.

The commercial stake was concrete. If the 34.5 kV system could not be shown effectively grounded, the plant would require 100 percent rated arresters throughout, 133 percent insulation level collection cable, and higher basic impulse insulation levels on the pad-mounted transformers. The contractor's change order for that package was priced at 2.1 million dollars against a schedule impact of eleven weeks on cable delivery. If instead effective grounding could be established by a grounding source, the marginal cost was one grounding transformer and its protection, but the utility then had to accept a materially higher ground fault duty on its own system and re-derive its ground relay settings. Neither party wished to make that trade on assertion.

3. Study Objectives and Scope of Work

Ten numbered objectives defined the engagement.

1. Establish and document the distinction between system grounding and equipment grounding for both the design team and the reviewing utility, and structure all deliverables accordingly.
2. Compare all six practical system neutral grounding methods against the technical and commercial criteria that matter for this facility, and justify the selection.
3. Determine, by sequence-network analysis, whether the 34.5 kV system is effectively grounded at the point of interconnection in every credible system state, including the islanded state.
4. Quantify the ground fault overvoltage exposure of the ungrounded island by electromagnetic transient simulation, including arrester energy duty.
5. Size and specify a grounding source that satisfies effective grounding without imposing excessive ground fault duty on the host utility.
6. Determine the consequential arrester ratings, cable insulation levels and equipment insulation coordination.
7. Assess ferroresonance and neutral instability risk in the as-found and mitigated configurations, and specify mitigation.
8. Perform an IEEE Std 80 substation grounding design for the expanded 138/34.5 kV substation, including soil resistivity survey, two-layer soil modeling, grid current derivation, conductor sizing, and touch and step voltage verification.
9. Re-derive the utility's ground overcurrent protection to remain selective at the higher ground fault duty, and specify the plant's ground fault detection and transfer trip.
10. Validate the study against field measurement at commissioning.

Scope and Deliverables

The deliverable set comprised a system grounding method selection report, an effective grounding and insulation coordination report, a ground fault overvoltage and ferroresonance electromagnetic transient report, an IEEE Std 80 substation grounding design report, a ground protection coordination report, the grounding transformer specification, the ground grid construction specification, the telecommunications isolation specification, and a field test procedure with acceptance criteria.

4. Regulatory and Standards Basis

The facility interconnects to a bulk electric system operator in the continental United States under a generator interconnection agreement derived from FERC Order No. 2003 and its successors, with the queue process governed by Order No. 2023 and Order No. 2023-A. The grounding requirement itself is not a FERC requirement; it originates in the host utility's facility connection requirements published under NERC FAC-001 , and the study demonstrating compliance is the kind of study contemplated by FAC-002 .

Effective grounding is defined for US practice principally in IEEE Std C62.92.1 , <em>Guide for the Application of Neutral Grounding in Electrical Utility Systems, Part I — Introduction</em>, which establishes the classification of grounding methods by the ratios X0/X1 and R0/X1 and provides the coefficient of grounding curves used throughout this study. IEEE Std C62.92.2 extends the treatment to synchronous generator systems and C62.92.4 to distribution systems. IEEE Std 142 , the Green Book, supplies the industrial and commercial system grounding practice used for the low-voltage and station service portions. Arrester selection follows IEEE Std C62.11 for the device ratings and IEEE Std C62.22 for the application, in particular the temporary overvoltage capability curves against which the calculated 1.34 per unit condition was checked. Insulation coordination follows IEEE Std 1313.1 and IEEE Std 1313.2 .

Substation grounding follows IEEE Std 80 , <em>Guide for Safety in AC Substation Grounding</em>, with field measurement per IEEE Std 81 , <em>Guide for Measuring Earth Resistivity, Ground Impedance and Earth Surface Potentials of a Grounding System</em>, and connector qualification per IEEE Std 837 , <em>Qualifying Permanent Connections Used in Substation Grounding</em>. Telecommunications protection follows IEEE Std 367 for calculation of the ground potential rise and induced voltage presented to a telecommunications facility, and IEEE Std 487 for the protection and isolation of wire-line circuits serving power stations.

The National Electrical Safety Code (ANSI C2) governs the grounding of the utility's supply system, including grounding electrode requirements, the grounding of surge arresters, and the multi-grounded neutral practice that supplies part of the split factor used in the grid current derivation. NFPA 70 (NEC) Article 250 governs the station service and low-voltage systems inside the control building, where the 480/277 V system is solidly grounded wye with a main bonding jumper and an equipment grounding conductor scheme, and Article 250 Part VIII governs the direct-current and communications grounding. OSHA 29 CFR 1910.269 governs worker protection grounding, specifically the requirement that temporary protective grounds establish an equipotential zone at the work location, a requirement whose adequacy depends on the same grid parameters computed in the IEEE Std 80 analysis. NFPA 70E governs work practice in the low-voltage station service.

Because the facility is inverter-based and exceeds 20 MVA aggregate at a connection above 60 kV once referred through the substation, the post-Order No. 901 standards apply to its performance. NERC PRC-029-1 , adopted by the NERC Board in October 2024 and filed with FERC in November 2024, becoming effective in 2026, governs voltage and frequency ride-through, and it interacts with this study at one specific point: the plant must ride through the retained voltages produced by ground faults on the utility system, and the unfaulted-phase overvoltage during those faults must not exceed the inverters' declared withstand. IEEE Std 2800-2022 provides the underlying performance definitions.

Standards and Regulatory Register

5. System Modeling and Data Development

5.1 The two disciplines, and why the data sets are different

The study maintained two model families because the two grounding disciplines ask different questions of different physics.

System grounding is a sequence-network problem. It asks how much current flows when a phase conductor contacts earth, and what voltage appears on the two phases that did not. Its inputs are transformer winding connections and impedances, line and cable positive- and zero-sequence impedances, source equivalents, and the zero-sequence capacitance of the cable network. Soil resistivity enters only weakly, through the earth-return impedance of overhead lines. Its outputs are fault currents, sequence voltages, coefficient of grounding, arrester duty and relay quantities.

Equipment and substation grounding is a field problem in a conducting half-space. It asks where the current that has entered the earth goes, and what potential difference appears between a person's feet and a person's hand. Its inputs are soil resistivity as a function of depth, grid geometry, conductor size, burial depth, ground rod configuration, surface layer material, and the proportion of fault current that actually enters the earth at that station. Transformer winding connections enter only through that proportion. Its outputs are grid resistance, ground potential rise, mesh voltage, step voltage and transferred potential.

Data was therefore collected in two streams, and the report was written so that no reader could take a number from one stream and apply it in the other. The most consequential example is the fault current. The system grounding study reports a single line-to-ground fault current of 4,380 A at the 34.5 kV point of interconnection. The substation grounding study uses a grid current of 9.4 kA derived from the 138 kV bus fault. These are different currents at different voltages for different purposes, and a design that used either in place of the other would be wrong by a wide margin.

5.2 Data sources and gap closure

The host utility supplied 138 kV positive- and zero-sequence source equivalents at the substation, the 138/34.5 kV transformer test report, the existing station grounding transformer nameplate and impedance, the 34.5 kV line construction standards, and the existing ground grid drawings of record. The Customer supplied the collection system design, inverter step-up transformer specifications and test reports, cable data including shield construction and grounding practice, and the inverter manufacturer's declared fault current contribution and voltage withstand.

Five material gaps were closed explicitly.

Soil resistivity was absent. No measurement existed for the substation site. Because every quantity in IEEE Std 80 scales with resistivity, and because a two-layer structure changes both grid resistance and surface potentials in ways that an average value cannot represent, a Wenner four-pin survey was commissioned across eight traverses at electrode spacings from 1.5 ft to 300 ft, with two traverses at right angles to each other over the future grid footprint. The measured apparent resistivity curve inverted cleanly to a two-layer model of 165 ohm-meters upper over 58 ohm-meters lower with an upper layer thickness of 8.9 ft , a structure consistent with a weathered overburden over a moist, more conductive substrate. The inversion residual was 4.1 percent.

The existing station grounding transformer impedance was uncertain. The nameplate recorded a zero-sequence impedance in percent on an unstated base. The value was reconciled by staged single line-to-ground fault records from two historical events on the 34.5 kV circuit, from which the zero-sequence path impedance at the substation bus was back-calculated. The resulting Z0 was 4.61 + j36.14 ohms referred to the point of interconnection, and the reconciliation agreed with the nameplate interpretation the utility's protection group had assumed within 8 percent.

Zero-sequence capacitance of the collection system was not documented. It governs the ungrounded-island behavior and the ferroresonance assessment. It was computed from the cable geometry, insulation permittivity and length schedule for all 41.2 circuit miles of 34.5 kV collector cable and 2.5 miles of tie, giving 14.8 microfarads per phase total, equivalent to a zero-sequence capacitive reactance of 179 ohms per phase and a total charging current of 111 A.

The existing grid's as-built condition was unverified. Drawings of record showed a conductor size and spacing but no evidence of the actual installed condition after decades of service. Six excavation test pits and a continuity survey were specified. Two of the six pits revealed 4/0 copper corroded at bolted connections that predated IEEE Std 837 qualification, with measurable resistance across the joint. Those findings changed the mitigation scope.

The split factor was not documentable from records. The proportion of ground fault current that returns through overhead shield wires and neutrals rather than entering the earth cannot be assumed. It was computed from the shield wire impedances of the four 138 kV lines terminating at the station, the 34.5 kV multi-grounded neutral where present, and the cable sheath returns, using the ladder-network method of IEEE Std 80 and cross-checked in the grounding software.

Model Data Sources and Gap Closure

5.3 Model build and benchmarking

Four models were maintained against a single controlled parameter register.

An ASPEN OneLiner model carried the utility's 138 kV network for four buses either side of the substation, the 138/34.5 kV transformers, the 34.5 kV bus, the tie and the plant collection system reduced to a single equivalent, with full positive-, negative- and zero-sequence representation. It produced the sequence impedances, the effective grounding ratios, fault currents at every location, and the relay coordination quantities.

An ETAP model carried the unreduced plant, all 84 inverter step-up transformers, the full collector network segment by segment, the station service system, and the low-voltage grounding arrangement. It was used for the plant-side fault duty, the grounding transformer duty, the station service and NEC Article 250 verification, and the ground fault arc energy estimates.

A PSCAD/EMTDC model carried the 34.5 kV system in the time domain with saturable transformer representation for both the inverter step-up units and the grounding transformer, distributed-parameter cable models with explicit sheath representation, arrester models with energy accumulation, breaker models with pole-scatter and current-chopping behavior, and averaged inverter models with the manufacturer's fault ride-through and current limit logic. It produced the ground fault overvoltage results, the arrester energy duty and the ferroresonance assessment. Positive-sequence and phasor models cannot produce any of these, because all three depend on saturation, on the interaction of capacitance with magnetizing inductance, or on the accumulation of energy over a transient.

A CDEGS model carried the substation grounding grid, the two-layer soil, the shield wires and neutrals as an explicit conductive network rather than as a lumped split factor, the fence, the cable trench system and the control building rebar. It produced grid resistance, the surface potential contours, mesh and step voltages, transferred potentials and the fall-of-potential test prediction against which the field measurement was later compared.

Three benchmarks preceded any reported result. The ASPEN three-phase and single line-to-ground fault currents at the 138 kV bus reconciled to the utility's own published fault duty within 1.4 percent and 2.1 percent respectively. The PSCAD model, run with a linear source and no saturation, reproduced the ASPEN single line-to-ground fault current at the point of interconnection within 2.6 percent, confirming that the sequence impedances had been transferred correctly before any non-linear behavior was introduced. The CDEGS grid resistance computed with a uniform soil of 90 ohm-meters reproduced the closed-form Sverak expression within 3.8 percent, confirming the geometry entry before the two-layer model was applied.

The report stated the limits plainly. The phasor models cannot represent the ungrounded-island transient, arrester energy or ferroresonance. The electromagnetic transient model cannot economically sweep the several hundred fault positions needed for relay coordination. The grounding model represents soil as horizontally layered and laterally uniform, which the site is not exactly, and its computed touch voltages carry an uncertainty of roughly plus or minus 12 percent on that account alone. The inverter models are averaged and do not represent switching-frequency behavior, which is immaterial to grounding but would matter for a harmonic study.

6. Study Methodology and Assumptions

6.1 System configuration

System Configuration Summary

The configuration that creates the problem is visible in one line of that table. The 138/34.5 kV source transformers are grounded-wye on the 138 kV side and delta on the 34.5 kV side. A delta winding is not a zero-sequence source. The only ground source on the utility's 34.5 kV system was a station zig-zag grounding transformer at the substation bus, installed decades earlier and sized to limit ground fault current rather than to establish effective grounding. On the plant side, every one of the 84 inverter step-up transformers presents a delta to the 34.5 kV collection system. The plant therefore contains no zero-sequence source whatsoever, and when the point-of-interconnection breaker opens, the 34.5 kV island that remains — the plant, its collection cable, the tie and whatever length of utility circuit is still connected — has no galvanic connection to earth except through the fault itself and the distributed capacitance of the cable.

6.2 Analytical approach

The system grounding analysis proceeded in five steps. First, the six practical grounding methods were compared against the criteria that matter for this system, to establish which methods were even candidates. Second, the sequence networks were built and the effective grounding ratios computed for every credible system state: maximum and minimum source, one and two source transformers, one and two 34.5 kV circuits, plant online and offline, and the islanded state. Third, the coefficient of grounding was determined from the C62.92.1 curves for each state, and the governing state identified. Fourth, the islanded state was carried into electromagnetic transient simulation, because in that state the sequence-network method breaks down: an ungrounded system's zero-sequence impedance is capacitive, the neutral floats, the behavior depends on saturation and on the arc, and the quantity of interest is an accumulated energy rather than a phasor. Fifth, the mitigation was sized and the analysis repeated.

The substation grounding analysis followed the IEEE Std 80 sequence without shortcuts: field measurement of soil resistivity, inversion to a layered model, determination of the maximum grid current, conductor sizing for the fault duration, preliminary grid design, computation of grid resistance and ground potential rise, computation of mesh and step voltages, comparison against tolerable limits, iteration, and then treatment of the special cases — fence, transferred potential, telecommunications, and the effect of the grid on worker protective grounding.

6.3 Cases and scenarios

Case and Scenario Matrix

Each case was run with the ground fault at seven positions: the 138 kV bus, the 34.5 kV substation bus, three points along the tie, the plant 34.5 kV bus, and a collector feeder mid-point. Fault resistance was taken as zero for duty and overvoltage, and swept from 0 to 40 ohms for relay sensitivity.

6.4 Acceptance criteria

Acceptance Criteria

Two assumptions were carried explicitly. The fault duration for shock safety was taken as 0.75 s , corresponding to the utility's zone-2 backup clearing time rather than the primary clearing time, on the reasoning that a stuck breaker is precisely the condition in which a person is most likely to still be in contact. The fault duration for conductor thermal sizing was taken as 1.0 s , a longer value, because conductor damage is cumulative and because the grid must survive a duty it may see repeatedly.

7. Analysis and Results

7.1 System grounding methods compared

There are six practical ways to connect a three-phase system neutral to earth, and the choice among them is one of the few decisions in power system design that simultaneously fixes fault current, overvoltage, detectability, arc energy, service continuity, arrester rating and cable insulation class. Each method optimizes some of those and penalizes others; none optimizes all.

Ungrounded (capacitively coupled). No intentional connection between any system neutral and earth. The system is nonetheless coupled to earth through the distributed phase-to-ground capacitance of lines, cables and equipment, and it is that capacitance which determines ground fault current. A single line-to-ground fault produces a current equal to three times the per-phase charging current — here 111 A for the plant collection system — flowing through the fault and returning through the capacitance of the unfaulted phases. The system can continue to operate with the fault present, which is the entire historical justification for the method. But the neutral shifts fully to the potential of the faulted phase, so the unfaulted phases sit at full line-to-line voltage to earth continuously, a coefficient of grounding of 1.0. Worse, an intermittent or restriking arcing ground fault can pump charge into the system capacitance cycle after cycle and produce transient overvoltages of 3 to 6 per unit. Fault detection requires zero-sequence voltage measurement rather than current, and fault location is notoriously difficult. Ungrounded operation is now rare in US utility practice and is confined mainly to small industrial systems below 5 kV where continuity of process is paramount.

High-resistance grounded. The neutral is connected to earth through a resistance chosen so that the resistive ground fault current slightly exceeds the system's total capacitive charging current, typically limiting total ground fault current to 10 A or less. That single design rule is what suppresses the transient overvoltage: the resistance damps the charge-pumping mechanism of the arcing ground fault and holds transients to roughly 2.5 per unit. Steady-state overvoltage on the unfaulted phases remains at essentially full line-to-line value, so the coefficient of grounding is still about 1.0 and full insulation and 100 percent arresters are required. Ground fault current is far too small to operate an overcurrent relay, so detection is by neutral voltage and location is by pulsing or by a tracing signal. The method is standard US practice for 480 V and 600 V industrial systems and for medium-voltage generator neutrals up to about 15 kV, where continuity is valuable and the system is small enough that charging current stays low. It does not scale: on a system with 111 A of charging current the required resistor current would exceed 111 A and the "high-resistance" designation ceases to be meaningful.

Low-resistance grounded. The neutral is grounded through a resistor sized to allow a definite and useful ground fault current, most commonly 200 A, 400 A, 600 A or 1,000 A. That current is large enough to operate ground overcurrent relays selectively and to permit conventional fault location, and small enough to limit ground fault arc damage and to reduce ground potential rise. Steady-state overvoltage on the unfaulted phases is close to line-to-line value because the zero-sequence impedance is high and resistive, so the coefficient of grounding usually exceeds 0.8 and 100 percent rated arresters and higher insulation levels are required. The resistor has a short-time rating, typically 10 s, and cannot support sustained operation with a fault. Low-resistance grounding is the dominant US practice for industrial and commercial medium-voltage distribution at 4.16 kV and 13.8 kV, and for many utility 34.5 kV subtransmission systems that are not required to be effectively grounded.

Reactance grounded. The neutral is grounded through a reactor. Because the zero-sequence impedance remains inductive, the reactance value can be chosen to place X0/X1 anywhere on the spectrum. When X0/X1 is held at 3.0 or below the system is effectively grounded and behaves for overvoltage purposes like a solidly grounded system, while ground fault current is reduced relative to solid grounding. When X0/X1 exceeds 3.0 the system is not effectively grounded and the overvoltage benefit is lost. Reactance grounding is applied where a solidly grounded transformer would produce a ground fault current higher than the three-phase fault current — a real and common condition on delta-wye transformers with low zero-sequence impedance — and it is the basis of the grounding transformer application selected in this study.

Resonant grounded (Petersen coil, ground fault neutralizer). The neutral is grounded through a tuned reactor whose inductive current at fundamental frequency very nearly cancels the system's capacitive charging current at the fault, reducing residual fault current to a few amperes. Most single line-to-ground arcing faults then self-extinguish without any breaker operation, which is the method's outstanding advantage and the reason for its adoption in high wildfire-risk applications and in systems with a high proportion of transient overhead faults. The price is high: the coil must be retuned as the network is switched, requiring an automatic tuning controller and continuous measurement of neutral displacement; the unfaulted phases sit at full line-to-line voltage during the fault so the coefficient of grounding is 1.0 and full insulation is required; and detection of a persistent fault requires a residual-current or wattmetric scheme rather than simple overcurrent. Resonant grounding is long-established in European practice and has been adopted in the United States on a growing but still limited number of distribution systems, principally for wildfire mitigation.

Solidly grounded. The neutral is connected directly to earth with no intentional impedance. Zero-sequence impedance is then set by the transformer and line impedances alone, and X0/X1 is typically between 0.7 and 3.0, so the system is effectively grounded. Unfaulted-phase overvoltage is held to about 1.25 to 1.4 per unit, permitting 80 percent rated arresters and 100 percent insulation level cable. Ground fault current is high, often equal to or exceeding the three-phase fault current, which maximizes ground fault arc energy, maximizes ground potential rise at the station, and requires the ground grid to be designed accordingly. Every fault must be cleared; there is no ride-through of a ground fault. Solid grounding is universal US practice for transmission at 69 kV and above, for four-wire multi-grounded distribution at 12.47 kV and 24.9 kV, and for low-voltage wye systems under NEC Article 250.

System Grounding Method Comparison

Two rows in that comparison decided this project. Arrester rating required and cable insulation level are not free choices; they are consequences of the coefficient of grounding, and the coefficient of grounding is a consequence of the neutral grounding method. Only the reactance-grounded and solidly grounded columns permit 80 percent arresters and 100 percent cable, and only those two columns therefore permit the equipment package the Customer had already priced.

7.2 What "effectively grounded" actually means

The term is defined in US practice by IEEE Std C62.92.1 and reproduced, with the same numbers, in the NEC and in most utility interconnection requirements. A system is effectively grounded at a given point when, for all system conditions:

X0 / X1 <= 3.0 and R0 / X1 <= 1.0

where X1 and X0 are the positive- and zero-sequence reactances and R0 the zero-sequence resistance, all as seen looking into the system from the point in question, and all evaluated with the system in the configuration that produces the worst ratios.

The two ratios exist for one reason: together they bound the coefficient of grounding , defined as the ratio of the highest root-mean-square line-to-ground voltage appearing on an unfaulted phase during a ground fault, to the nominal line-to-line voltage of the system. When both criteria are met, the coefficient of grounding cannot exceed 0.80 . That is the whole meaning of the 80 percent figure, and it is worth stating precisely because it is routinely misquoted. It does not mean that the voltage rises to 80 percent of normal. It means that the unfaulted phases, which normally sit at 1/sqrt(3) or 57.7 percent of line-to-line voltage, rise during a ground fault to no more than 80 percent of line-to-line voltage. On a 34.5 kV system that is a rise from 19.9 kV to no more than 27.6 kV, a factor of 1.39 in per-unit terms.

The direct consequences are two.

Arrester rating. A surge arrester must survive the temporary overvoltage that appears across it during a ground fault elsewhere on the system, for as long as that fault persists. An arrester whose duty-cycle voltage rating is 80 percent of the system line-to-line voltage — an "80 percent rated" or "effectively grounded" arrester — is applied only where the coefficient of grounding is 0.80 or less. Where it is higher, a 100 percent rated arrester is required, whose higher maximum continuous operating voltage buys temporary overvoltage capability at the cost of a higher protective level and therefore a smaller protective margin for the insulation it guards.

Cable insulation level. Medium-voltage cable is manufactured in three insulation levels. 100 percent level cable is applied where the fault is cleared within one minute, which in practice means an effectively grounded system with fast ground fault clearing. 133 percent level cable, with proportionally thicker insulation, is applied where the fault may persist for up to one hour, which is the impedance-grounded case. 173 percent level cable is applied where the fault may persist indefinitely, which is the ungrounded and resonant-grounded case. For 35 kV class cable the wall thickness rises from 345 mils at 100 percent level to 420 mils at 133 percent, with the corresponding increases in diameter, bending radius, duct fill, reel length, splice cost and price.

Worked sequence-impedance arithmetic. The as-found sequence impedances at the 34.5 kV point of interconnection, in the minimum-source configuration with one source transformer and one 34.5 kV circuit in service, were:

• Z1 = Z2 = 1.08 + j7.69 ohms
• Z0 = 4.61 + j36.14 ohms

The effective grounding ratios follow directly:

• X0 / X1 = 36.14 / 7.69 = 4.70 , against a criterion of 3.0 or less — fails
• R0 / X1 = 4.61 / 7.69 = 0.60 , against a criterion of 1.0 or less — passes

Only one of the two criteria must fail for the system to be not effectively grounded, and X0/X1 failed by 57 percent. The single line-to-ground fault current follows from the same impedances. With the pre-fault line-to-neutral voltage E taken as 34.5 kV / sqrt(3) = 19,920 V:

• Z1 + Z2 + Z0 = (1.08 + 1.08 + 4.61) + j(7.69 + 7.69 + 36.14) = 6.77 + j51.52 ohms
• Magnitude = sqrt(6.77^2 + 51.52^2) = 51.96 ohms
• 3I0 = 3 x 19,920 / 51.96 = 1,150 A

For comparison the three-phase fault current at the same location is 19,920 / |1.08 + j7.69| = 19,920 / 7.77 = 2,565 A. Ground fault current at 45 percent of three-phase current is the signature of a system that is not effectively grounded.

Entering the C62.92.1 coefficient of grounding curves at X0/X1 = 4.70 and R0/X1 = 0.60 gives a coefficient of grounding of 0.88 for the grid-connected condition, already above the 0.80 threshold. The governing condition, however, is not the grid-connected one.

After installation of the grounding transformer, in the maximum-source configuration with both source transformers, both 34.5 kV circuits and the plant contributing:

• Z1 = Z2 = 0.41 + j2.91 ohms
• Z0 = 1.75 + j7.57 ohms
• X0 / X1 = 7.57 / 2.91 = 2.60 — passes
• R0 / X1 = 1.75 / 2.91 = 0.60 — passes
• Z1 + Z2 + Z0 = 2.57 + j13.39 ohms, magnitude 13.63 ohms
• 3I0 = 3 x 19,920 / 13.63 = 4,380 A

and the coefficient of grounding from the curves is 0.77 , corresponding to a highest unfaulted-phase voltage of 0.77 x 34.5 = 26.6 kV, or 1.34 per unit of the nominal line-to-neutral value. The system is effectively grounded, the 80 percent arrester application is valid, and 100 percent insulation level cable is valid. The four-fold increase in ground fault current is the price, and section 7.7 addresses what it cost in protection engineering.

Effective Grounding Results by System State

The table carries the finding that mattered. The as-found system failed the effective grounding criteria in every grid-connected state, and in the islanded state it had no zero-sequence source at all: the zero-sequence impedance was capacitive rather than inductive, the criteria do not apply because they presume an inductive zero-sequence network, and the coefficient of grounding rose to 0.94. That islanded state is the governing case, and it is the case a purely phasor-based study will miss, because a fault study run on a case file in which the point-of-interconnection breaker is closed will never reveal it.

7.3 Ground fault overvoltage on the inverter-based interconnection

Ground fault overvoltage, universally abbreviated GFOV, is the single most commercially significant grounding problem in US inverter-based generation interconnection, and it arises from a sequence of events that is entirely ordinary.

A single line-to-ground fault occurs on the utility's 34.5 kV circuit between the substation and the plant. The utility's line relay operates and its source breaker opens. The fault is still present. On a conventional system the circuit would now be dead. Here it is not, because the generating facility is still connected to the faulted circuit and is still producing current. What remains energized is an island consisting of the plant, its collection system, the tie line and the faulted section, and that island contains no zero-sequence source of any kind, because every transformer facing the 34.5 kV system presents a delta winding. The island is, by construction, an ungrounded system with a ground fault on it .

In an ungrounded system with a ground fault, the system neutral shifts to the potential of the faulted phase. The two unfaulted phases, which were at 19.9 kV to earth, are now at the full line-to-line voltage of 34.5 kV to earth, a factor of 1.73. Every arrester, every cable termination, every pad-mounted transformer bushing and every instrument transformer on those two phases sees that voltage. Metal-oxide arresters do not fail instantly at 1.73 per unit; they conduct, heat, and fail after a period that depends on their energy rating and their temporary overvoltage capability. For a typical 34.5 kV distribution-class arrester that period is a fraction of a second to a few seconds. The plant's own protection may eventually detect the condition and trip, but "eventually" is measured against arrester thermal runaway, and the arresters usually lose.

The mechanism explains the adjacent-circuit event described in section 2 exactly: nine arresters failing within two seconds of a ground fault, followed by cable termination failures at the weakened points. It also explains why the utility's interconnection requirement exists. The utility is not protecting the generator; it is protecting its own distribution assets and its other customers from an overvoltage that the generator creates and that the utility cannot clear, because the source breaker is already open.

Electromagnetic transient results. The islanded condition was simulated in PSCAD/EMTDC with saturable transformer models, distributed-parameter cable, arrester models with energy accumulation, and averaged inverter models carrying the manufacturer's ride-through and current-limit logic. The fault was applied at 0.100 s on the tie line, the source breaker opened at 0.183 s on the first current zero after the relay operate time, and the simulation ran to 2.0 s.

Ground Fault Overvoltage Simulation Results

The as-found sustained value of 1.63 per unit is below the theoretical 1.73 per unit of a perfectly ungrounded system because of the resistance in the fault path, the damping contributed by the inverters' current-limited behavior, and the small zero-sequence loading of instrument transformers and arrester leakage. It is not below it by enough to matter. The arrester energy duty of 4.7 kJ per kV of maximum continuous operating voltage exceeded the class rating of the installed 34.5 kV arresters within 0.38 s, which is well inside the time required for the plant's anti-islanding protection to detect the condition on the as-found scheme.

Three options were evaluated to introduce a zero-sequence source into the island.

Option A: a grounding transformer at the plant. A zig-zag or grounded-wye/delta bank connected to the plant 34.5 kV bus, inside the point-of-interconnection breaker so that it stays with the island when that breaker opens. This is the standard US solution. Its advantages are that it is independent of the main step-up transformers, that it can be sized precisely to the required zero-sequence impedance rather than accepting whatever the step-up transformer offers, that it can be switched or removed for maintenance with a defined operating restriction, and that its neutral offers a convenient location for ground fault detection and for a neutral resistor should current limiting ever be wanted. Its disadvantages are the added equipment, the added losses, the requirement for its own protection, and the fact that it is a permanently energized asset whose failure removes the effective grounding.

Option B: a grounded-wye / grounded-wye main step-up transformer. Making the collection-side winding of the plant's step-up units grounded-wye rather than delta creates a zero-sequence path through the transformer itself. This appears elegant and is frequently proposed, but it carries two well-known penalties. A grounded-wye/grounded-wye transformer with no delta winding has no path for third-harmonic magnetizing current, so the flux waveform distorts, the core operates less efficiently and third-harmonic voltage appears on both windings; the standard remedy is a buried delta tertiary, which reintroduces the equipment being avoided. It also transfers zero-sequence voltage and current between systems: a ground fault on one side produces zero-sequence current on the other, which desensitizes ground relaying on both and can create spurious operation. Finally, grounded-wye/grounded-wye units with cable-fed windings are the classic ferroresonance candidate under single-phase switching, discussed in section 7.6.

Option C: a wye-grounded high side with a delta tertiary. A three-winding step-up transformer with a grounded-wye 34.5 kV winding, the delta as a buried tertiary providing the third-harmonic path and defining the zero-sequence impedance. This is a robust and conventional solution for larger interconnections, but it requires the main step-up transformers to be respecified and reordered, which at the design freeze milestone carried a fourteen-month delivery impact and a cost far above the alternatives.

Option A was selected. The decisive consideration beyond cost was the ability to choose the zero-sequence impedance rather than inherit it, because the zero-sequence impedance is the variable that resolves the central tension of the whole problem.

7.4 Sizing the grounding transformer: the tension between grounding and fault duty

Adding a zero-sequence source to a system does two things at once, and they pull in opposite directions.

It lowers X0/X1 and therefore lowers the coefficient of grounding, which is the objective. But it also lowers the zero-sequence impedance seen by every ground fault on the system, which raises ground fault current everywhere. Higher ground fault current means more ground fault arc energy, higher ground potential rise at the substation, higher duty on the ground grid, higher thermal duty on cable shields and on the grounding transformer itself, and — most awkwardly for the host utility — a redistribution of ground fault current that changes what its existing ground relays see. If the grounding transformer is made too strong, the utility's ground relays at the substation see less of the fault current, because more of it is supplied from the plant end, and the relays lose sensitivity for faults they previously covered. That is the desensitization problem, and it is the reason utilities are ambivalent about the very grounding source they require.

The sizing therefore has an upper and a lower bound. The lower bound on the grounding transformer's zero-sequence impedance is set by fault duty and by relay desensitization. The upper bound is set by the effective grounding criterion: the impedance must be low enough that X0/X1 in the island stays at or below 3.0.

Grounding Transformer Sizing Options

The selected value of 13.8 ohms per phase is the highest impedance — that is, the weakest grounding source — that satisfies the effective grounding criterion with margin in every state including the island. Choosing the weakest adequate source is the correct optimization, because every ohm removed below that point buys nothing in overvoltage terms and costs fault duty.

The rating follows from the duty. For a bolted single line-to-ground fault at the grounding transformer terminals with the utility source removed, the bank alone supplies:

• 3I0 = 3 x 19,920 / 13.8 = 4,330 A

which is very nearly the whole of the 4,380 A available at the point of interconnection with the utility source connected, confirming that the bank is the dominant zero-sequence source in the mitigated system. The specified bank is a 34.5 kV zig-zag grounding transformer, 400 A continuous neutral current, 4,400 A for 10 seconds, zero-sequence impedance 13.8 ohms per phase , with a 10-second short-time rating per IEEE Std C57.32 chosen to envelope the calculated 4,380 A duty against the maximum backup clearing time of 0.75 s with margin for repeat operations. The continuous rating of 400 A, at 9 percent of the short-time rating, covers the standing neutral current arising from system unbalance and from the residual third-harmonic circulation, which was measured at 11 A after energization.

A zig-zag connection was chosen over a grounded-wye/delta bank for three reasons: it uses less core and winding material for the same zero-sequence impedance because each limb carries windings from two phases in opposition; it has no secondary winding to protect, fuse or load; and its zero-sequence impedance is defined by the interwinding geometry and is stable, whereas a grounded-wye/delta bank's zero-sequence impedance depends on the delta winding's short-circuit impedance and is more variable in manufacture. The zig-zag's disadvantage — that it cannot supply station service from a secondary — did not apply, because the plant already had station service.

The bank is protected by a three-phase overcurrent relay on its phase terminals for internal faults and by a neutral overcurrent relay in its ground connection, with the neutral element time-coordinated to remain above the maximum system ground fault contribution so that it does not operate for external faults. A key specification requirement, written explicitly because it is often missed, is that the grounding transformer must not be switched independently of the plant . Its disconnect is interlocked so that the plant cannot be energized with the grounding transformer isolated, because a plant energized without its grounding source is precisely the as-found condition the study exists to eliminate.

7.5 Insulation coordination consequences

With the coefficient of grounding established at 0.77, the arrester and insulation package was re-derived.

Arrester selection at 34.5 kV. The maximum continuous operating voltage of an arrester must equal or exceed the maximum continuous line-to-ground voltage at its location. On a 34.5 kV nominal system operating within ANSI C84.1 Range A, the maximum operating voltage is 36.2 kV, giving 20.9 kV line-to-ground. The as-found requirement, driven by a coefficient of grounding above 0.8, was a 100 percent rated arrester: 36 kV duty-cycle rating, 29 kV MCOV . With effective grounding demonstrated, an 80 percent rated arrester: 27 kV duty-cycle rating, 22 kV MCOV is applicable. The 22 kV MCOV exceeds the 20.9 kV continuous requirement with 5 percent margin.

The temporary overvoltage check is the one that must not be skipped. The calculated ground fault overvoltage of 1.34 per unit corresponds to 26.6 kV line-to-ground, which is 1.21 times the 22 kV MCOV of the selected arrester. The manufacturer's temporary overvoltage capability curve for that arrester, with prior duty, permits 1.30 times MCOV for 1 second and 1.25 times for 10 seconds. Against a maximum backup clearing time of 0.75 s the selected arrester has capability at 1.30 times MCOV, a margin of 7.4 percent over the calculated 1.21. That margin was the acceptance basis, and it is thin enough that the study recommended, and the utility adopted, a reduction of the 34.5 kV backup clearing time from 1.20 s to 0.75 s as a condition of the grounding transformer's installation.

The protective margin was then confirmed against the insulation. The selected 80 percent arrester has a lightning impulse protective level of 88 kV at 10 kA. The 34.5 kV pad-mounted transformers and cable terminations carry a 150 kV basic lightning impulse insulation level. The protective margin is (150 / 88) - 1 = 70 percent, comfortably above the 20 percent minimum of IEEE Std 1313.2. Had the 100 percent rated arrester been required, its protective level of 112 kV would have reduced the margin to 34 percent, still acceptable but materially poorer, and on the 25 kV class cable terminations originally considered it would have been marginal.

Cable insulation level. With effective grounding demonstrated and ground fault clearing within 0.75 s, 100 percent insulation level 35 kV class cable is applicable: 345 mils of insulation wall. Without it, 133 percent level cable at 420 mils would have been required across the entire 41.2 circuit miles of collector cable and the 1.0 mile underground portion of the tie. The difference is not only the insulation cost. The larger diameter changes duct fill, reduces the length per reel and therefore adds splices, increases the minimum bending radius and therefore the trench and vault geometry, and reduces ampacity slightly through the increased thermal resistance of the thicker wall, which in two collector circuits would have forced a conductor size increase.

Insulation Coordination Package

The 138 kV package was unaffected because the 138 kV system is solidly grounded with X0/X1 of 1.42, a coefficient of grounding of 0.75, and 80 percent rated arresters were already correct there. That contrast — an effectively grounded 138 kV system feeding a 34.5 kV system that was not effectively grounded, through a delta winding that blocks any zero-sequence connection between them — is the misconception the contractor had made, and it is worth stating as a rule: grounding does not pass through a transformer. Each voltage level's grounding must be established on its own terms, by a source connected to that level.

7.6 Ferroresonance and neutral instability

Ferroresonance is a non-linear resonance between the saturable magnetizing inductance of a transformer and a series capacitance, most commonly the phase-to-ground capacitance of the cable feeding it. It requires three ingredients: a saturable inductance, a series capacitance of the right order, and a source connected to the combination through fewer than three phases. On a grounded system with three-phase gang-operated switching it is rare. On an ungrounded or impedance-grounded system with single-phase switching devices — fused cutouts, single-pole reclosers, or simply a blown fuse — it is common.

The mechanism is straightforward once stated. When one or two phases are opened, the transformer's magnetizing inductance is no longer connected to a stiff source. It is instead connected in series with the phase-to-ground capacitance of the still-energized phases through the transformer windings. If the resulting series circuit has a natural frequency near a subharmonic or harmonic of 60 Hz, and if the inductance is driven into saturation so that its effective value swings as the flux traverses the knee of the magnetizing curve, the circuit can lock into a sustained non-sinusoidal oscillation. The consequences are neutral displacement, sustained overvoltage on the unfaulted or de-energized phases of 2 to 4 per unit, characteristic audible noise from the core, arrester failure, and in severe cases transformer damage.

The as-found system was a candidate on every count. The 34.5 kV system was impedance-grounded through a distant, high-impedance station bank, so neutral displacement was unconstrained. The tie line included a 1.0 mile cable section with substantial phase-to-ground capacitance. The plant's 84 pad-mounted step-up transformers presented saturable magnetizing inductance behind delta windings. And the utility's 34.5 kV construction standard permitted single-phase fused cutouts at the tie takeoff.

The assessment used the standard screening ratio, the ratio of the transformer's magnetizing reactance to the cable's zero-sequence capacitive reactance, expressed in the US convention as the connected transformer capacity per unit length of cable. Screening identified one exposed configuration and one only: a single pad-mounted step-up transformer left connected to a lightly loaded collector cable lateral through a single-phase device during construction or maintenance switching. PSCAD verification of that configuration produced a sustained period-3 subharmonic oscillation with a peak phase-to-ground voltage of 2.41 per unit and a fundamental-frequency component of 1.38 per unit, persisting indefinitely once established. The same configuration with the grounding transformer in service produced no sustained oscillation, because the zero-sequence path provided by the grounding bank clamps the neutral and prevents the flux excursion that drives the mechanism.

Three mitigations were specified and all three adopted, because ferroresonance is cheap to prevent and expensive to experience.

First, the grounding transformer itself is the primary mitigation , and this is one of its underappreciated benefits: an effectively grounded system is very difficult to drive into ferroresonance because the neutral cannot displace. Second, single-phase switching devices were prohibited on any cable-fed transformer in the collection system and on the tie; three-phase gang-operated switches and three-phase reclosers were specified in their place, and the utility replaced the fused cutouts at the tie takeoff with a three-phase device. Third, an energization and de-energization procedure was written requiring that switching always be performed at the three-phase device nearest the source, never at a single-phase device on the load side, and requiring transformers to be energized with load connected where practical, since load damping suppresses the oscillation.

Ferroresonance Assessment Results

7.7 Ground protection re-coordination

The rise in available ground fault current from 1,150 A to 4,380 A at the point of interconnection was not a side effect to be noted; it became a power system studies protection engineering problem that had to be solved before the utility would accept the grounding transformer.

Three difficulties arose. First, the utility's existing ground overcurrent relay at the 34.5 kV substation was set with a pickup of 120 A and a time dial chosen against the as-found maximum ground fault current, and at 4,380 A that setting operated in 0.21 s, faster than the downstream plant protection and therefore non-selective. Second, and more subtly, the grounding transformer at the plant end now supplies ground fault current toward the substation for faults on the utility side of the point of interconnection, which means the utility's relay sees less of the total fault current than it used to for faults near the plant end, while seeing more for faults near the substation. A non-directional element cannot resolve that. Third, the grounding transformer is itself a source of zero-sequence current for faults anywhere on the utility system, including faults on adjacent circuits at the same substation bus, which the plant's own protection must not trip for.

The scheme delivered resolved all three.

Ground Protection Scheme

Directional supervision at the point of interconnection is provided by a zero-sequence directional element polarized from the grounding transformer's neutral current, with a negative-sequence directional element as a secondary polarizing source for faults where the zero-sequence quantity is small. Polarizing from the grounding transformer neutral is the natural choice here and one of the practical advantages of a dedicated grounding bank over a grounded-wye/grounded-wye step-up transformer, which offers no equally clean polarizing quantity.

Coordination was verified over the full range of fault positions and fault resistances. The minimum coordination interval across all cases was 0.34 s , above the 0.30 s acceptance criterion. Sensitivity was verified against a 40 ohm fault resistance at the remote end of the utility circuit under minimum-source conditions, where the residual current falls to 388 A, still above the revised 240 A pickup with a 1.6 factor. The 59N neutral overvoltage element provides the backstop: it detects a ground fault anywhere on the 34.5 kV system by neutral displacement alone, independent of current magnitude and independent of direction, and its 0.50 s definite time setting sits below the arrester temporary overvoltage capability with margin.

Direct transfer trip from the utility's point-of-interconnection breaker to the plant was retained even with the grounding transformer installed. The grounding transformer eliminates the overvoltage consequence of an unintended island; it does not eliminate the island. Transfer trip remains the deterministic means of ensuring the plant does not energize a utility circuit that the utility believes to be de-energized, which is an operator safety requirement independent of any overvoltage consideration.

7.8 Substation grounding: the second discipline

Everything to this point concerns system grounding. What follows concerns the conductive grid in the earth, and shares with the foregoing only one number: the fault current.

### 7.8.1 Soil resistivity survey and two-layer modeling

A Wenner four-pin survey was performed per IEEE Std 81 across eight traverses, two of them orthogonal, at electrode spacings of 1.5, 3, 5, 10, 15, 25, 40, 60, 100, 150, 200 and 300 ft. The Wenner arrangement places four equally spaced electrodes in a line, injects current between the outer pair and measures voltage across the inner pair; the apparent resistivity is 2 pi a R where a is the spacing and R the measured resistance, and the measurement is weighted toward soil at a depth of roughly the electrode spacing. Sweeping the spacing therefore sweeps depth.

The measured apparent resistivity curve fell monotonically from 158 ohm-meters at 1.5 ft spacing to 66 ohm-meters at 300 ft, the classic signature of a resistive upper layer over a more conductive lower layer. Inversion produced a two-layer model of 165 ohm-meters upper over 58 ohm-meters lower, with an upper layer thickness of 8.9 ft , with a root-mean-square inversion residual of 4.1 percent.

The two-layer structure matters more than a casual reader expects, and in opposite directions for the two quantities of interest. Grid resistance is dominated by the deeper soil, because current spreading from a large grid penetrates well below the grid depth; the low 58 ohm-meter lower layer therefore gives a low grid resistance and a low ground potential rise, which is favorable. Surface potentials, however, are dominated by the shallow soil, because touch and step voltage are determined by the potential gradient in the first meter; the higher 165 ohm-meter upper layer therefore raises mesh and step voltages, which is unfavourable. A single averaged resistivity would have got both wrong: using 66 ohm-meters throughout would have underestimated mesh voltage by 41 percent, and using 158 ohm-meters throughout would have overestimated grid resistance by a factor of 2.3 and driven a substantially over-designed grid.

### 7.8.2 Grid current: the number that is most often wrong

The current that matters for grounding design is not the fault current. It is the portion of the fault current that actually flows from the grid into the earth, and getting from one to the other requires three separate corrections. Each is routinely omitted, and each omission is conservative in one direction only.

Correction one: use the ground fault current, not the three-phase current. Ground potential rise is produced by zero-sequence current returning through earth. A three-phase fault produces no zero-sequence current at all and contributes nothing to ground potential rise. Using the three-phase fault current, which is often the larger number and the one most readily available, is not conservative — it is simply the wrong quantity, and at this station it would have been wrong in the unconservative direction for one configuration and the over-conservative direction for another. The design current here is the single line-to-ground fault current at the 138 kV bus, 23,785 A symmetrical , taken with the station at maximum source configuration.

Correction two: the split factor. Not all of that current returns through the earth. A substantial fraction returns metallically, through the overhead shield wires of the transmission lines, through the multi-grounded neutrals of distribution circuits, and through cable sheaths, all of which are bonded to the grid at one end and to remote ground electrodes at the other. Only the remainder enters the soil at the station and produces ground potential rise. The split factor Sf is the ratio of grid current to total fault current, and it is computed from the impedances of those metallic return paths in parallel with the earth return. At this station, with four 138 kV lines each carrying two overhead shield wires, a partially multi-grounded 34.5 kV system, and bonded cable sheaths, the computed split factor was 0.38 . Sixty-two percent of the fault current returns metallically and never enters the earth. Ignoring the split factor would have produced a ground potential rise of 9,900 V instead of 3,760 V and would have driven a grid roughly 2.4 times the size actually required.

Correction three: the decrement factor. The symmetrical fault current is an rms phasor quantity. The actual current in the first cycles contains a decaying direct-current component whose magnitude depends on the X/R ratio at the fault point and on the instant of fault initiation, and which increases the effective heating and the effective rms value over the fault duration. The decrement factor Df accounts for it, and it depends on both X/R and the duration: for short durations it is large, for long durations the direct-current component has decayed and the factor approaches unity. At this station the X/R at the 138 kV bus is 15.2 and the shock design duration is 0.75 s, giving Df = 1.04 .

The design grid current is therefore:

• IG = 3I0 x Sf x Df = 23,785 x 0.38 x 1.04 = 9,400 A

Grid Current Derivation

The growth allowance deserves a note. The utility's transmission planning forecast a rise in 138 kV bus fault duty to 26.4 kA at the ten-year horizon. The grid was verified against that value as a sensitivity rather than designed to it, because grid extension is straightforward and over-designing the initial grid for a forecast fault duty is poor capital allocation. At 26.4 kA the mesh voltage rises to 568 V, still below the 654 V tolerable limit.

### 7.8.3 Conductor sizing

Grid conductors must survive the fault current without fusing and without annealing, and must retain adequate cross-section after a service life of corrosion. IEEE Std 80 provides the sizing relation derived from the Onderdonk equation, which in its practical US form is:

• A\_kcmil = I x Kf x sqrt(tc)

where I is the fault current in kiloamperes, tc the fault duration in seconds, and Kf a constant that depends on the conductor material and on the maximum allowable temperature, which is itself set by the weakest connection in the path , not by the conductor. That last point is the one that governs practice: bare hard-drawn copper can reach 1,084 degrees Celsius before fusing, but a brazed joint is limited to 450 degrees and a bolted or compression connection to 250 degrees. Designing a grid to the conductor's limit and then joining it with bolted connectors produces a grid that fails at the joints.

For this design, with exothermically welded connections qualified to IEEE Std 837, Kf = 7.06 for commercial hard-drawn copper at a 1,084 degree limit. With the full symmetrical fault current of 23.8 kA — the split factor is deliberately not applied to conductor sizing, because a conductor near the fault may carry the whole current before it divides — and a thermal design duration of 1.0 s:

• A = 23.8 x 7.06 x sqrt(1.0) = 168 kcmil minimum

The selected conductor is 4/0 AWG bare copper, 211.6 kcmil , which provides a 26 percent margin over the thermal minimum. The margin is not decorative: it covers the corrosion allowance over a 40-year service life, mechanical robustness during backfill and future excavation, and the possibility of a longer clearing time under a stuck-breaker condition. Ground risers to equipment are 4/0 copper; equipment bonding jumpers are 2/0; the fence bonding conductor is 2/0.

All connections are exothermic welds qualified per IEEE Std 837 , which subjects candidate connectors to a sequence of fault current, freeze-thaw, corrosion and mechanical tests and requires that the connection remain electrically and mechanically sound after them. The excavation findings described in section 5.2 — bolted connections on the legacy grid showing measurable resistance across the joint after decades in soil — are exactly the failure mode IEEE Std 837 exists to prevent, and they were the basis for a recommendation that all accessible legacy bolted connections in the retained grid be cut out and replaced with exothermic welds during the expansion.

### 7.8.4 Grid resistance, ground potential rise, mesh and step voltage

The grid as designed covers 387 ft by 387 ft, a footprint of 3.44 acres, with 4/0 copper conductors on a uniform 28 ft spacing in both directions, buried 20 inches deep, plus 64 ground rods of 10 ft length distributed around the perimeter and at equipment locations. Total buried conductor length including rods is 12,240 ft.

Substation Grounding Design and Results

Ground potential rise is the product of grid current and grid resistance:

• GPR = IG x Rg = 9,400 x 0.40 = 3,760 V

The tolerable limits follow the IEEE Std 80 body-current criterion, in which the tolerable voltage is the product of the body resistance path and the current that the heart can survive for the shock duration. For a 70 kg body and a shock duration ts:

• E\_touch,70 = (1000 + 1.5 x Cs x rho\_s) x 0.157 / sqrt(ts)
• E\_step,70 = (1000 + 6.0 x Cs x rho\_s) x 0.157 / sqrt(ts)

with 1000 ohms the assumed body resistance, 1.5 and 6.0 the foot-resistance multipliers for the touch and step contact geometries, rho\_s the surface layer resistivity and Cs the surface layer derating factor. The 0.157 constant becomes 0.116 for a 50 kg body.

The surface layer derating factor merits explanation because it is the most cost-effective single measure in substation grounding. A layer of crushed rock placed on the surface raises the resistance in series with the feet, and therefore raises the voltage a person can tolerate, without changing the grid at all. Its effectiveness depends on its own resistivity, its thickness, and the resistivity of the soil beneath it, through the IEEE Std 80 reflection expression:

• Cs = 1 - 0.09 x (1 - rho / rho\_s) / (2 x hs + 0.09)

With rho = 165 ohm-meters upper-layer soil, rho\_s = 2,500 ohm-meters for the specified crushed rock, and hs = 0.102 m (4 inches), Cs = 0.71. With ts = 0.75 s the tolerable values are 654 V touch and 2,080 V step for a 70 kg body , and 493 V touch and 1,568 V step for a 50 kg body .

Against those limits the computed grid performance is:

• Mesh voltage 512 V against 654 V tolerable, a margin of 21.7 percent
• Step voltage 198 V against 2,080 V tolerable, a margin of 90.5 percent

Step voltage passing by an enormous margin while touch voltage passes narrowly is the normal result for a well-designed grid, and it is worth stating why: step voltage involves the potential difference between two points one meter apart on the surface, both of which sit above the same buried grid at similar potential, while touch voltage involves the difference between the surface potential at the middle of a mesh and the potential of a grounded structure the person is holding, which is the full grid potential. Touch voltage is almost always the governing criterion, and a design that checks only step voltage is not a design.

One result required an explicit decision. The mesh voltage of 512 V passes the 70 kg criterion but exceeds the 50 kg criterion of 493 V . IEEE Std 80 permits the 70 kg body weight to be used where access is restricted to qualified personnel, and the utility's practice is to design substation interiors on that basis. The interior of the fenced station was therefore accepted at the 70 kg criterion. The perimeter fence and the exterior of the gates , however, are accessible to the public, and were designed to the 50 kg criterion: the fence is bonded to the grid at 50 ft intervals, an additional grid conductor is installed 3 ft outside the fence line at 20 inches depth, the crushed rock surface layer is extended 5 ft beyond the fence, and the resulting touch voltage at the fence exterior is 268 V against the 493 V limit. Stating that decision explicitly in the report, rather than allowing a single body weight to be applied silently across the whole site, was a deliberate act of engineering documentation.

### 7.8.5 Transferred potential, telecommunications and corrosion

A grid at 3,760 V above remote earth does not confine that potential to the fenced area. Any conductive path leaving the station carries station potential outward, or brings remote potential inward, and the resulting difference appears across whatever is at the far end. Four transfer paths were assessed.

Fence. The fence is bonded to the grid, so it sits at grid potential, and the hazard is a person standing on remote earth outside touching it. That case was resolved by the perimeter conductor, extended surface layer and 50 kg design basis described above.

Rails and pipes. A spur rail served the transformer unloading pad and passed under the fence to a point 1,115 ft from the station. Bonded to the grid, it would carry 3,760 V to that point. Bonded to remote earth, it would import remote potential into the station. The design used an insulated rail joint 20 ft outside the fence, with the interior section bonded to the grid, the exterior section isolated, and warning signage at the break. Two buried metallic services, a water line and a gas service, were fitted with isolating joints at the property boundary in coordination with their owners.

Neutral and shield wire connections. These are intentional transfer paths and are the mechanism of the 0.38 split factor. They carry station potential to every downstream pole ground, which is normal and accepted US practice, and their effect on customer-side touch voltage was verified for the first five structures of each departing circuit.

Telecommunications. This is the case with a specific standards requirement. A metallic telecommunications pair entering the station is bonded to the station ground at one end and to the telephone company's remote ground at the other. During a ground fault the full ground potential rise appears across the insulation of that pair and across any equipment or person connected to it. IEEE Std 367 provides the methods for calculating the ground potential rise and the induced longitudinal voltage that a power station presents to a telecommunications facility, and establishes the concept of the power influence zone: the region within which the calculated GPR exceeds a threshold, conventionally 300 V rms , beyond which special protection is required. IEEE Std 487 then specifies the protection and isolation options: high-voltage isolating transformers, neutralizing transformers, optical isolators, or complete isolation by an all-dielectric fiber-optic link.

At 3,760 V this station is more than twelve times the 300 V threshold. All metallic telecommunications entering the station were eliminated. The specification requires an all-dielectric self-supporting fiber-optic cable with no metallic strength member, armour or moisture barrier, terminated on optical equipment inside the control building with no conductive path to the telephone company plant, and the existing copper pairs abandoned and physically removed to the first pole outside the influence zone. This is the least ambiguous and, at current costs, usually the cheapest of the IEEE Std 487 options.

Corrosion. The grid is a buried copper structure adjacent to galvanized steel structures, reinforced concrete foundations and, at this site, a steel spur rail. Copper is cathodic to steel and to zinc, so a galvanic cell exists wherever they are bonded and share an electrolyte, and the steel corrodes. The design measures were: exothermic welds throughout to eliminate the crevice corrosion that attacks bolted joints, welded copper-to-steel connections at structures rather than mechanical clamps, sacrificial magnesium anodes on the two structures identified as most exposed, and an inspection and testing regime requiring grid resistance measurement and connection sampling at ten-year intervals. Soil chemistry testing returned a pH of 6.4, a chloride content of 78 ppm and a sulphate content of 210 ppm, classifying the site as mildly corrosive and not requiring the more aggressive measures.

7.9 NEC, NESC and OSHA interfaces

Three code frameworks apply at this site simultaneously and to different parts of it, and the report addressed each boundary explicitly because they are commonly confused.

The NESC (ANSI C2) governs the utility's supply system: the substation grid itself, the grounding of surge arresters and equipment, the multi-grounded neutral of the 34.5 kV system, grounding electrode requirements and conductor ampacity for grounding conductors, and the clearances that determine where a person can stand relative to energized parts. The IEEE Std 80 design satisfies and exceeds the NESC's grounding provisions, which set minimum requirements rather than performance criteria.

The NEC (NFPA 70) governs the station service and low-voltage systems inside the control building and the plant's own auxiliary systems. Article 250 applies in full: the 480/277 V station service is a solidly grounded wye system with a single main bonding jumper at the service disconnect, a grounding electrode conductor to the station grid, equipment grounding conductors sized per Table 250.122, and separately derived system grounding for the two 480-208/120 V transformers. The boundary between NEC and NESC jurisdiction falls at the service point, and the report drew it on a drawing rather than describing it in text.

OSHA 29 CFR 1910.269 governs worker protection, and it connects to this study through the equipotential zone requirement. When workers apply temporary protective grounds and a fault occurs, the current flowing through those grounds develops a voltage across them and across the earth between the worker's feet, and the requirement is that the worker be within an equipotential zone in which that difference is tolerable. Whether that is achieved depends on the same grid parameters computed here. The study delivered a specific product for this purpose: a set of calculated worker touch and step voltages at the eight most-used maintenance work positions in the station under the design fault, all of which fell below the 70 kg tolerable limits, together with a requirement for a portable equipotential grounding mat at two positions where the calculated values approached the limit.

8. Sensitivity and Scenario Analysis

Six parameters were varied to establish which conclusions were robust and which depended on assumptions.

Sensitivity Analysis Results

Four findings emerged from the sweep.

The effective grounding conclusion is insensitive to almost everything. The coefficient of grounding depends on impedance ratios, not magnitudes, and those ratios are set by the grounding transformer and the transformer and line impedances, all of which are known to within a few percent from test reports. Varying the inverter fault contribution across its full plausible range changed the ground fault current by 4 percent and the coefficient of grounding not at all, because inverters are not a zero-sequence source behind a delta winding regardless of how much positive-sequence current they produce. That is a useful result to be able to state, because it removes the single most uncertain input in an inverter-based study from the grounding conclusion entirely.

The substation grounding conclusion is sensitive to two inputs and only two. Upper-layer soil resistivity and split factor together account for nearly all the variance in mesh voltage. Both were measured or computed rather than assumed, which is why the sweep is reassuring rather than alarming. Had either been assumed, the design would have carried real risk: at an upper-layer resistivity of 218 ohm-meters, only 32 percent above the measured value, the design fails.

The split factor is the more dangerous of the two, because it is the one most often taken from a table or a rule of thumb. A designer who assumed 0.50 rather than computing 0.38 would have over-designed by roughly 30 percent; a designer who assumed 0.25 would have produced a grid that fails. It is also the parameter most likely to change during the station's life, since removing a transmission line or replacing a shield wire changes it directly. The report recommended that the split factor be recomputed whenever a line terminating at the station is added, removed or reconductored.

Ferroresonance risk is binary rather than graded. Sweeping cable length and transformer size across the plausible range produced either no sustained oscillation or a fully developed one, with a narrow transition. That is characteristic of the phenomenon and it is why screening ratios are used as go/no-go tests rather than as margin calculations, and why the mitigation was specified as a prohibition on single-phase switching rather than as a limit on cable length.

9. Findings and Root Cause Assessment

Findings Register

The root causes reduce to three, and they are worth separating because they call for different remedies.

A conceptual error about what grounding means. Findings F1, F2 and F3 all trace to the belief that a system connected to a solidly grounded system is itself grounded. It is not. A delta winding is an open circuit to zero-sequence current, and every voltage level requires its own zero-sequence source. This error is common enough that the study report opened with a page devoted to it, and the interconnection review meeting spent more time on that page than on any result.

A legacy design that was correct when made and was never revisited. Findings F2, F5 and F6 describe a station and a system designed to the practice of its era, with a delta 34.5 kV source, a current-limiting station grounding bank, bolted grid connections and no soil data. None of this was negligent at the time. What changed was the arrival of generation on the 34.5 kV system, which converted a system that never had to be effectively grounded into one that does, and the growth of fault duty, which converted a grid that was adequate into one that had never been verified.

Settings and specifications derived against superseded conditions. Findings F8 and F9 are the ordinary consequence of any change in system configuration, and they are listed not because they were surprising but because they are the part of the work that a study which stopped at "install a grounding transformer" would have left undone, and which would then have surfaced as a misoperation or an arrester failure after energization.

10. Mitigation Options and Recommendations

The system grounding options were compared on technical merit, cost order of magnitude, schedule and residual risk. Costs are order-of-magnitude figures for comparison only.

Mitigation Options Comparison

Option D deserves comment because it is the option most developers instinctively reach for: if the system is not effectively grounded, simply buy equipment rated for it. It fails here for two reasons. It does not satisfy the interconnection agreement, which requires an effectively grounded source rather than equipment able to tolerate the absence of one. And it protects only the Customer's own equipment; the utility's arresters, its other customers' equipment and the utility's cable on the affected circuit remain exposed to 1.63 per unit, and the utility has no means to clear the condition once its breaker is open.

Option E likewise fails, and it is the option most often proposed as a compromise. Transfer trip does reduce the exposure time, but the simulation showed arrester class energy exceeded within 0.38 s from breaker opening, and a transfer trip scheme including relay operate time, communications latency and plant breaker interrupting time cannot reliably act faster than roughly 0.15 to 0.25 s, leaving little margin and none at all if the communications channel is degraded. Transfer trip is a necessary complement to a grounding source, not a substitute for it.

Option A was recommended and implemented , with three specification conditions: the bank is located inside the point-of-interconnection breaker so that it remains with any island; its disconnect is interlocked against plant energization; and the utility's 34.5 kV backup clearing time is reduced from 1.20 s to 0.75 s to restore arrester temporary overvoltage margin.

The substation grounding recommendations were: extend the grid to the new bay at 28 ft spacing with 4/0 copper and exothermic connections; add 64 ground rods; install the 4 inch, 2,500 ohm-meter crushed rock surface layer across the yard and 5 ft beyond the fence; install the perimeter conductor 3 ft outside the fence with fence bonding at 50 ft intervals; cut out and replace accessible legacy bolted connections; install insulating joints in the spur rail and the two metallic services; replace all metallic telecommunications with an all-dielectric fiber-optic link; and adopt the calculated worker equipotential zone values with portable mats at two positions.

11. Implementation Support and Field Validation

Keentel supported implementation from Month 5 to Month 9 and validated the study against measurement rather than against agreement.

The grounding transformer was factory tested with a zero-sequence impedance measurement per IEEE Std C57.12.90, returning 13.6 ohms per phase against the specified 13.8, within the 5 percent tolerance. Its temperature rise test at the 400 A continuous rating and its 10 second short-time verification were witnessed.

Grid construction was inspected at three hold points: after conductor installation and before backfill, when conductor routing, spacing, depth and weld quality were verified against the design; after rod installation, when individual rod resistances were spot-measured; and after surface layer placement, when thickness was verified at 24 locations against the 4 inch specification, since a surface layer thinner than specified silently invalidates the derating factor and therefore the touch voltage compliance.

Three field tests validated the analysis.

Fall-of-potential grid resistance test per IEEE Std 81. Current was injected between the grid and a remote current electrode with the potential probe traversed along a line at an angle to the current lead, and the grid resistance read at the plateau. The 61.8 percent rule — that the true resistance is read when the potential probe is placed at 61.8 percent of the distance to the current electrode — derives from an analytical result for hemispherical electrodes in uniform soil with the potential probe collinear with the current lead, and it is exact only under those conditions. On a large grid in layered soil it is an approximation whose error grows with grid size relative to probe spacing. The test therefore used a current electrode at 2,740 ft, five times the 547 ft grid diagonal, with the potential lead at 30 degrees to the current lead to reduce mutual coupling, and the result was confirmed by the slope method as well as by the 61.8 percent reading. Measured grid resistance was 0.41 ohms against a calculated 0.40 ohms , an agreement of 2.5 percent.

Surface potential measurement. Touch and step voltages were measured at twelve positions with a current injection of 42 A and scaled to the design grid current, using a 1,000 ohm shunt to represent body resistance and a weighted electrode to represent foot contact. Measured touch voltages scaled to 9,400 A ranged from 388 V to 547 V against calculated values of 361 V to 512 V, the measurements running 5 to 8 percent above calculation, which is the expected direction given lateral soil non-uniformity the layered model cannot represent. All measured values remained below the 654 V tolerable limit.

Staged ground fault. A staged single line-to-ground fault at the plant 34.5 kV bus during commissioning, cleared by the point-of-interconnection breaker, produced a residual current of 4,290 A against a calculated 4,380 A , an agreement of 2.1 percent, and a measured neutral displacement voltage consistent with a coefficient of grounding of 0.78 against the calculated 0.77. The 67N, 51G and 59N elements all operated with the expected timing, and the recorded oscillography was retained as the commissioning evidence for the interconnection grounding condition.

Field Validation Against Simulation

12. Results and Value Delivered

Outcomes Scorecard

The 2.1 million dollar figure is the contractor's priced change order for the 133 percent insulation level cable, the 100 percent rated arrester package and the associated termination and duct changes, avoided by demonstrating effective grounding. Against it stand the 310 thousand dollar installed cost of the grounding transformer, its foundation, protection and interlocking, and the engineering cost of the study. The net capital benefit is approximately 1.7 million dollars.

That figure understates the value, because it counts only what was avoided and not what was prevented. The as-found configuration carried an arrester energy duty exceeding the class rating within 0.38 s of every ground fault followed by a breaker opening. On the adjacent circuit, the same condition destroyed nine arresters and two cable terminations in a single event. There is no defensible way to price the failures that did not occur, and the report did not attempt to, but the utility's own event cost for the adjacent circuit incident, including outage, replacement and investigation, was on the order of 400 thousand dollars for one event.

13. Lessons Learned and Engineering Insights

Grounding is two disciplines, and the word is the only thing they share. The most valuable single act in this engagement was structuring the report so that system grounding and equipment grounding could not be confused. The confusion is not merely semantic. It produces designers who size a ground grid using three-phase fault current, protection engineers who believe a solidly grounded transmission system grounds everything behind it, and developers who think a grounded-wye winding on the low side of an inverter transformer is a ground source for the collection system. Each of those errors was present at the start of this project.

The islanded case is the governing case, and a fault study will not find it. Every conclusion about arrester duty and coefficient of grounding was driven by the state in which the utility's breaker is open and the plant is still energized. A conventional fault study is run on a case file in which that breaker is closed. If the study scope does not explicitly require the islanded state to be examined, it will not be examined, and the answer produced will be both defensible and wrong.

Choose the weakest adequate grounding source. The instinct when adding a grounding transformer is to make it strong, because strong grounding sounds safer. It is not. Every ohm of zero-sequence impedance removed below the value that satisfies the criterion buys no reduction in overvoltage and buys a real increase in fault duty, arc energy, ground potential rise and relay desensitization. Sizing the bank at the highest impedance that satisfies X0/X1 <= 3.0 with margin in every state is the correct optimization, and it is the opposite of what an intuition trained on "more grounding is better" will suggest.

The split factor deserves calculation, not a table lookup. It varied the mesh voltage result by more than any other input, it is invisible in the fault study output, and it changes whenever the station's line terminations change. A grounding design that documents a split factor without documenting how it was derived should not be accepted.

R0/X1 is not a secondary criterion. The two effective grounding criteria are usually quoted together and then only the first is checked. A zero-sequence network that is resistance-dominated can produce a coefficient of grounding well above 0.8 at an X0/X1 that looks acceptable, because resistive zero-sequence impedance shifts the neutral without limiting its displacement. On systems grounded through resistors, through long lines with high earth-return resistance, or through legacy banks of uncertain construction, R0/X1 is the criterion that fails first.

The field test is not a formality. Fall-of-potential agreement within 2.5 percent and staged fault agreement within 2.1 percent are what converted a calculation into an accepted compliance demonstration. The measurements also carried a real caution: touch voltages measured 5 to 8 percent above calculation, consistently and in one direction, which is the signature of lateral soil variation that a layered model cannot represent. A design whose calculated margin is under 10 percent should be regarded as unverified until measured.

14. Keentel Capability Summary

• System neutral grounding method selection and justification across all six practical methods, for utility, industrial and generation interconnection applications
• Effective grounding verification to IEEE Std C62.92.1, including sequence-network development, coefficient of grounding determination and consequential insulation coordination
• Ground fault overvoltage analysis for inverter-based resource interconnections, including electromagnetic transient simulation of the islanded condition and arrester energy duty
• Grounding transformer sizing, specification, protection and interlocking, including the fault duty and relay desensitization trade-off
• Insulation coordination to IEEE Std 1313.1/1313.2 and arrester application to IEEE Std C62.11/C62.22, including temporary overvoltage capability verification
• Ferroresonance and neutral instability screening and electromagnetic transient verification
• Substation grounding design to IEEE Std 80, including soil resistivity survey design, multi-layer inversion, grid current derivation, conductor sizing, touch and step voltage analysis and transferred potential assessment
• Grounding system field testing to IEEE Std 81, including fall-of-potential, slope method and surface potential measurement, with staged fault test design and witness
• Telecommunications GPR assessment and isolation specification to IEEE Std 367 and IEEE Std 487
• Ground protection coordination, directional ground scheme design and settings development
• NESC, NEC Article 250 and OSHA 1910.269 grounding compliance interfaces, including worker equipotential zone verification
• Modeling in ASPEN OneLiner, ETAP, PSCAD/EMTDC and CDEGS/WinIGS class grounding software

15. Frequently Asked Questions

16. Glossary of Terms and Abbreviations

17. Confidentiality and Use Statement

This case study has been prepared for informational purposes. All client identities, project locations, contract details, and proprietary data have been withheld or generalized. Technical parameters, study results, and figures presented are representative of work performed by Keentel Engineering Solutions and have been adapted so that no individual project, owner, or facility can be identified. Nothing in this document constitutes a design recommendation for any specific installation. Any reuse of the methodologies described requires project-specific engineering analysis by a qualified professional engineer.