Gas-Insulated Substations for Industrial & Utility Power

A field-tested design review, FAQ, and four project case studies — anchored in a landmark 138 kV GIS cogeneration install

Why Gas-Insulated Substations Still Win the Hard Sites

What a 138 kV cogeneration substation from the golden age of GIS teaches us about compact, low-maintenance high-voltage design in 2026.

The problem every crowded plant eventually hits

Sooner or later, an industrial site runs out of room. A cogeneration unit gets added next to an already-dense process area. A data hall needs 200 MW where there used to be a parking lot. A utility must rebuild a downtown substation on the same fenced footprint it has occupied since the 1960s. In every one of these situations the electrical engineer faces the same wall: a conventional air-insulated substation (AIS) at transmission voltage simply needs a lot of land, a lot of vertical clearance, and a lot of exposed energized steel that people, cranes, and weather can all reach.

The most instructive answer to that problem is not a new one. A widely cited IEEE Petroleum & Chemical Industry Committee paper — R. L. Doughty’s “Design of a 138 kV Gas Insulated Substation for a Large Industrial Cogeneration Facility” (PCIC-89-13) — documented a gas-insulated substation (GIS) built inside a petrochemical plant to tie a 120 MVA gas-turbine cogeneration unit into a 138 kV system. Decades later it remains one of the clearest end-to-end walkthroughs of *why* an engineering team chooses GIS, and *what* that choice actually commits you to. Keentel Engineering revisits it here because almost every design tension it describes is live again today, only now with an added regulatory dimension around the insulating gas itself.

A GIS packs the same 138 kV switching, isolating, grounding, metering, and protection functions of an open-air yard into an SF₆-filled enclosure roughly the footprint of a shipping container. On a constrained site, that density is not a luxury — it is what makes the project buildable at all.

The project in one paragraph

A General Electric Frame 7 combustion gas turbine driving a 120 MVA generator (0.85 power factor, 13.8 kV) was installed at a Gulf-region petrochemical plant. An 80/119 MVA step-up transformer lifted the generator output to 138 kV. There was no room for an outdoor yard and no appetite for stringing overhead lines across a plant full of tall process equipment and mobile cranes. So the team specified a compact GIS in a dedicated building, fed the 138 kV connection out through underground low-pressure oil-filled (LPOF) cable running about 600 feet in a concrete duct bank to the existing plant substation, and sized the whole arrangement as a ring bus so a second cogeneration unit could be added later without a rebuild. Two years after commissioning, the unit was running at better than 98% availability.

Eight reasons the team chose GIS and why they still hold

The paper lays out a decision rationale that reads like a checklist any modern owner can reuse. The eight drivers were:

1. Compact footprint that made the installation physically possible and left room for a future second unit.
2. Elimination of overhead 138 kV line hazards to plant vehicular traffic — critically, mobile cranes working near a process unit.
3. No exposed energized conductors or bushings for personnel to contact.
4. Reduced impact of airborne pollutants on the electrical system — a real concern in a petrochemical atmosphere that coats and corrodes open insulators.
5. ~30% lower electrical maintenance cost for the 138 kV equipment, with minimal cable maintenance expected once the LPOF system was commissioned.
6. Reliability equal to or better than a conventional substation.
7. A cleaner visual profile for the facility.
8. Life-cycle cost equal to or below a conventional substation, despite the higher initial price.

That last point is the one decision-makers most often get wrong. GIS almost always costs more to buy. The case for it is made on land, safety, availability, and lifetime maintenance — not on the sticker. When those four are constrained, GIS frequently wins the total-cost argument even though it loses the first-cost argument.

Conventional AIS vs. GIS at a glance

The design decisions worth copying

Underground cable instead of overhead line

The transition from the GIS to the plant substation was made with LPOF cable rather than an overhead line. The team accepted a higher initial cost for the cable in exchange for removing the overhead-line hazard entirely, eliminating personnel exposure, and cutting the facility’s exposure to airborne contamination. The 138 kV connection ran as SF₆ bus out of the building, transitioned to 1250 kcmil LPOF cable outdoors above a cable pit, then entered a concrete duct bank. The cable pit was deliberately split into two compartments so a fault or oil fire on one supply cable could not easily damage the other — a small civil detail with a large reliability payoff.

Ring-bus configuration, built half now and half later

The GIS was arranged as a ring bus sized for two incoming 138 kV supplies and two load feeders, but only the portion needed for the first unit was installed, with provision to complete the ring when a second cogeneration unit arrives. A ring bus gives each element two paths to source and lets any single breaker be maintained without dropping load — high reliability without the cost of a full breaker-and-a-half scheme. Designing the ultimate configuration up front, then staging the build, is exactly the kind of foresight that keeps a later expansion from becoming a demolition job.

Synchronizing through the GIS, with reclosing deliberately blocked

In normal operation both 138 kV breakers were closed, and either could be used to synchronize the turbine to the system. Automatic reclosing was intentionally disabled for two reasons: to prevent out-of-synchronism reclosing onto the generator, and to reduce the probability of an internal fault escalating to an enclosure burn-through. This is a good reminder that a GIS changes the fault calculus — an arc that would flash harmlessly to air in an open yard is confined inside an aluminum enclosure, so protection and operating philosophy must respect that confinement.

Protection: layered, overlapping, and fast

Underground cable instead of overhead line

The transition from the GIS to the plant substation was made with LPOF cable rather than an overhead line. The team accepted a higher initial cost for the cable in exchange for removing the overhead-line hazard entirely, eliminating personnel exposure, and cutting the facility’s exposure to airborne contamination. The 138 kV connection ran as SF₆ bus out of the building, transitioned to 1250 kcmil LPOF cable outdoors above a cable pit, then entered a concrete duct bank. The cable pit was deliberately split into two compartments so a fault or oil fire on one supply cable could not easily damage the other — a small civil detail with a large reliability payoff.

Ring-bus configuration, built half now and half later

The GIS was arranged as a ring bus sized for two incoming 138 kV supplies and two load feeders, but only the portion needed for the first unit was installed, with provision to complete the ring when a second cogeneration unit arrives. A ring bus gives each element two paths to source and lets any single breaker be maintained without dropping load — high reliability without the cost of a full breaker-and-a-half scheme. Designing the ultimate configuration up front, then staging the build, is exactly the kind of foresight that keeps a later expansion from becoming a demolition job.

Synchronizing through the GIS, with reclosing deliberately blocked

In normal operation both 138 kV breakers were closed, and either could be used to synchronize the turbine to the system. Automatic reclosing was intentionally disabled for two reasons: to prevent out-of-synchronism reclosing onto the generator, and to reduce the probability of an internal fault escalating to an enclosure burn-through. This is a good reminder that a GIS changes the fault calculus — an arc that would flash harmlessly to air in an open yard is confined inside an aluminum enclosure, so protection and operating philosophy must respect that confinement.

Protection: layered, overlapping, and fast

The protection scheme is a clinic in defense-in-depth for a generation intertie:

• High-impedance instantaneous differential relays protected the 138 kV GIS bus between the step-up transformer and the breakers.
• Harmonic-restrained differential protected the step-up transformer, guarding against false trips on magnetizing inrush.
• An overall differential zone wrapped the whole machine — generator (from the neutral side), 13.8 kV isolated-phase bus, step-up transformer, and 138 kV SF₆ bus up to the breakers — as a backup blanket over the individual zones.
• The line side, underground cable, and connected 138 kV line used a directional-comparison unblocking scheme with frequency-shift carrier, backed by a permissive overreaching transfer-trip using audio-tone equipment.
• Each generator breaker had breaker-failure protection that, on failure, isolated the turbine, tripped the remaining generator breaker, and reached back to the 138 kV backup breakers.

The lesson for owners: in a GIS you cannot walk the yard and visually confirm a problem, and you cannot tolerate a lingering internal arc. That pushes the design toward fast, redundant, overlapping protection with clear backup for every primary function.

Insulation coordination: the study that changed the purchase order

The most technically rich part of the paper is the surge-protection study, and it carries a warning every specifier should hear. The GIS was specified to IEC standards at 145 kV maximum continuous voltage, 650 kV BIL, and 275 kV power-frequency withstand. The step-up transformer windings, though, were only rated 550 kV BIL. To verify that the metal-oxide arresters actually protected both, the team commissioned an EMTP (Electromagnetic Transients Program) study modeling lightning strokes both far from and near the cable-to-overhead-line transition.

The results were sobering. With arresters only at the incoming cable, a nearby stroke drove the calculated safety factor below the IEC minimum of 1.20 at the transformer, and the 550 kV transformer BIL was exceeded in the worst case. The fix was to add SF₆-enclosed metal-oxide arresters on the SF₆ bus, right at the transformer primary bushings (98 kV rms maximum continuous operating voltage). Re-running the study with those arresters raised the GIS safety factor to 1.62 and the transformer to 1.74 — comfortably above the 1.50 target. The purchase order was amended accordingly.

Takeaway: arrester placement is not a formality you can bolt on at the incoming line and forget. In a GIS, steep-front surges travel fast through low-impedance bus, and the equipment they threaten most may be the transformer behind the switchgear. Model it before you buy it.

The building around the switchgear is part of the design

A GIS is not just apparatus — it is apparatus plus a building plus a gas-handling regime, and the paper treats all three as one system. The SF₆ building was a steel-frame, precast-concrete structure roughly 50 ft long by 26 ft wide by 26 ft high, finished as a near clean-room: white walls, dust-suppressing floor coating, and full HVAC operating *before* the GIS ever arrived so the equipment was received into a clean, protected environment.

Because SF₆ is about five times heavier than air, colorless, and odorless, it pools in low spots and can displace breathable air. The building answered that with a ventilation system providing constant fresh-air makeup, floor-level SF₆ detectors that trigger exhaust fans and dampers to sweep the floor, an alarm in the control room, and a flashing warning light at the entrance.

Below-grade cable pits required a confined-space (vessel entry) procedure, and any maintenance on arced compartments demanded special handling because SF₆ decomposition products are toxic. None of this is optional folklore — it is the safety envelope that makes an enclosed high-voltage gas system livable for the people who operate it.

The civil interface nobody expects: differential settlement

One of the paper’s most quietly valuable sections concerns foundations. A rigid SF₆ bus connecting a heavy step-up transformer to the GIS cannot tolerate much relative movement.

The team allowed only about 0.25 inches of differential settlement between the transformer and GIS foundations and roughly 0.50 inches of vertical bus movement, using a flexible bellows at the transformer connection to absorb vibration and thermal growth, and a saddle support that let the bus flex as a cantilever. Because ordinary shallow foundations could not guarantee those tolerances, civil engineers specified under-reamed (belled) footings for the building, the GIS, and the transformer, and took benchmark elevation readings after installation to track settlement over time. The point generalizes: a GIS forces the electrical and civil disciplines to design to each other’s tolerances, far more tightly than an AIS ever does.

Key ratings from the installation



For reference, the headline equipment ratings that defined the design:

Reading the 1989 design through a 2026 lens

Everything above still applies. What has changed is the gas. SF₆ is an extraordinary insulator and arc-quencher, but it is also the most potent greenhouse gas known, with a global-warming potential on the order of 23,000–25,000 times CO₂ and an atmospheric lifetime measured in millennia. Regulators have moved. The EU’s F-gas Regulation (EU) 2024/573 phases SF₆ out of new switchgear on a voltage-tiered schedule — new equipment up to 24 kV from 2026, the 52–145 kV class (which includes 138 kV) restricted from 2028, and higher voltages from 2032, with narrow exemptions where no suitable alternative exists. In the U.S., SF₆ use and emissions face growing EPA and state-level reporting and reduction pressure.

The industry’s response is a new generation of insulating media: fluoronitrile (C4-FN) mixtures marketed as “clean-air” or “g³” solutions, fluoroketone (C5-FK) blends, and technical/vacuum approaches at lower voltages — all aiming to cut GWP by 99%+ while preserving GIS’s compactness. For an owner specifying a 138 kV GIS today, the Doughty design principles are unchanged; the specification simply gains a new line item: which insulating gas, at what handling and end-of-life cost, under which regulatory horizon. That is precisely the kind of trade-off Keentel helps clients frame before the purchase order is cut.

What Keentel takes from this paper

• Specify for the whole life, not the first invoice. GIS wins on land, safety, availability, and maintenance — build the business case there.
• Design the ultimate one-line, build the first stage. A ring bus sized for expansion turns a future rebuild into a bolt-on.
• Model your surges before you buy. An EMTP/insulation-coordination study can move an arrester and save a transformer.
• Treat the building, gas handling, and foundations as part of the electrical design — not as someone else’s scope.
• Put the gas question on the table early. With SF₆ phase-downs underway, insulating-medium choice now shapes lifecycle cost and compliance.

Have a constrained site, a generation intertie, or an aging yard you cannot expand? That is exactly the conversation this paper was written for — and the one our team has most weeks. Talk to Keentel Engineering about whether GIS is the right answer for your footprint.

Four GIS Applications, Four Different Constraints

How gas-insulated design solves the space, safety, and reliability problem across very different sectors.

Note: The following are anonymized, representative engagements. Locations and identifying details have been generalized; technical parameters are illustrative of typical projects in each sector and are consistent with the design principles discussed above.



Case Study 1 — Downtown Utility Rebuild on a Frozen Footprint

The challenge

A utility needed to rebuild a 1960s-era air-insulated substation that fed a growing downtown load pocket. The site was hemmed in on all four sides by streets and commercial buildings — acquiring adjacent land was impossible, and the existing yard could not be enlarged. Load growth demanded more capacity and a second transmission source yet the substation had to keep serving the neighborhood throughout construction.

The Keentel approach

The team specified a compact indoor GIS that fit the entire high-voltage lineup into a two-story equipment building on the original parcel, with underground cable getaways replacing the overhead line terminations that had dominated the old yard. A staged cutover kept the legacy AIS energized while the GIS was built alongside it, then transferred feeders bay by bay. Because the GIS is sealed and indoors, the design also removed the utility’s chronic problems with urban salt, grime, and pigeon-related flashovers on open insulators.

The outcome

Capacity and a redundant source were added on the same footprint, with a cleaner streetscape and no overhead transmission over city sidewalks. Projected maintenance dropped sharply because the switchgear no longer weathered the urban atmosphere, and the utility gained a template it now reuses for other landlocked downtown sites. An SF₆-free (fluoronitrile) option was evaluated against the phase-down timeline and held as the preferred medium for the next phase.

Case Study 2 — Hyperscale Data Center Campus Intertie

The challenge

A hyperscale operator needed to bring hundreds of megawatts onto a campus where every acre was earmarked for revenue-generating data halls, not electrical yard. Uptime was non-negotiable — the intertie had to support concurrent maintenance with zero load interruption — and the utility’s timeline for energization was aggressive.

The Keentel approach

The design used a GIS switching station in a footprint a fraction of the AIS equivalent, freeing land for additional halls. The bus was arranged for N-1 redundancy with concurrent maintainability so any breaker or transformer bay could be isolated without dropping the campus, mirroring the ring-bus philosophy of building the ultimate configuration and staging capacity as new buildings come online. Fast, redundant differential and breaker-failure protection matched the facility’s tolerance for downtime — effectively zero.

The outcome

The campus secured its full power block on minimal land, with a switching arrangement that supports live expansion as each new data hall is commissioned. The compact GIS also compressed the construction schedule, helping the operator hit its energization date. The reduced maintenance profile and enclosed design fit a lights-out site with a small on-site electrical staff.

Case Study 3 — Offshore-Fed Renewable Collector Substation

The challenge

An offshore wind development landed its export cables at a coastal site subject to relentless salt fog, wind-driven moisture, and corrosion — an environment that punishes exposed air-insulated equipment. The collector substation also had a compact coastal parcel with strict visual and environmental constraints from local stakeholders.

The Keentel approach

A fully enclosed GIS was the natural fit: sealing the live parts from the marine atmosphere directly addressed the corrosion and contamination problem that would otherwise drive constant maintenance and flashover risk. Underground cable terminations brought the export circuits into the GIS, keeping the site low-profile to satisfy visual concerns. The enclosed arrangement also simplified environmental permitting relative to a sprawling open yard on a sensitive shoreline.

The outcome

The substation delivered a weatherproof, low-visual-impact tie-in that keeps the wind farm exporting reliably through coastal storms, with dramatically reduced exposure-driven maintenance. As with the other engagements, a reduced-GWP insulating medium was evaluated up front so the asset aligns with tightening emissions rules over its multi-decade life.

Case Study 4 — Remote Mining & Minerals Processing Complex

The challenge

A minerals processing complex sat at high altitude in a remote, seismically active, dust-laden region. Thin air reduces the dielectric strength of air-insulated equipment (forcing larger clearances), airborne dust coats insulators, seismic activity threatens tall rigid structures, and the site’s remoteness made frequent maintenance visits costly and slow. The plant’s process loads also could not tolerate extended outages.

The Keentel approach

A GIS resolved several constraints at once. Because the insulation is inside a pressurized enclosure rather than relying on ambient air, altitude derating and dust contamination largely fall away, and the compact, low-mass arrangement is far easier to seismically qualify than a tall open yard. Echoing the cogeneration reference case, the design paid close attention to the civil interface — foundations and bus flexibility tuned to seismic and settlement conditions — and to layered protection that keeps process loads stable. The low-maintenance, sealed design directly answered the remoteness problem.

The outcome

The complex received a robust, low-maintenance high-voltage supply that shrugs off altitude, dust, and seismic events, with long maintenance intervals suited to a site where every service trip is expensive. The result is exactly the value proposition the 1989 cogeneration paper first articulated — compact size, improved safety, and reduced maintenance — carried into a modern, hostile-environment application.

About this briefing

This document is a Keentel Engineering technical-content piece reviewing IEEE PCIC-89-13, “Design of a 138 kV Gas Insulated Substation for a Large Industrial Cogeneration Facility” (R. L. Doughty, E. I. du Pont de Nemours and Company). Technical figures attributed to “the reference project” are drawn from that paper. Regulatory and alternative-gas context reflects the EU F-gas Regulation (EU) 2024/573 and industry sources current as of July 2026. Case studies are anonymized, representative composites for illustration.

Gas-Insulated Substations: In-Depth FAQ

Straight answers to the questions engineers, project managers, and owners ask most often when weighing GIS — grounded in the 138 kV cogeneration case and current practice.