Grid and Transmission: A Practical Reference
Generation gets the headlines; the grid decides what actually gets built. This guide catalogs 34 technologies across seven classes, from 765 kV overhead line to the smart meter on a house, with the voltage class each one serves, what it costs in time rather than dollars, and how much of the schedule is engineering against how much is right-of-way.
Overhead lines
conductors, structures, and how hard you can push them6 technologiesThe overhead AC line is how almost all bulk power moves: bare aluminum conductors strung between insulator-hung structures, using air as the insulation. Voltage is what makes it work. Power is volts times amps, and losses go as current squared, so a 765 kV line carries several times the power of a 345 kV line on similar conductors with a fraction of the loss per MW. A single 765 kV circuit moves roughly 2–4 GW. Everything else in this sheet exists because an overhead line is either not enough, not allowed, or not possible.
Strengths & weaknessesNothing is close on cost per MW-mile, air self-heals after a flashover, and a damaged line can be repaired in days. Capacity scales with voltage, so the same corridor can be uprated later. The weaknesses are all about the corridor rather than the wire. AC power flow follows impedance rather than contracts, so power takes unscheduled paths through neighboring systems; series reactance limits how much a long AC line can carry well below its thermal rating; and a new right of way needs agreement from every landowner and jurisdiction it crosses. Losses run roughly 2–4% per 1,000 km at 765 kV, higher at lower voltage.
When to useDefault to overhead AC for anything on land under about 600 km. It is cheaper, faster, and more repairable than any alternative, and the whole protection and operations toolkit already assumes it. Reach past it in three situations: when the route is longer than roughly 600 km, where HVDC's lack of reactive charging current wins; when the line has to cross water or dense city, where cable is the only option; and when you cannot get a new corridor, which is the common case in the US, and where reconductoring an existing line usually beats a new one. If the project schedule assumes a greenfield right of way in under seven years, check the assumption before anything else.
Key numbersCommon classes 138, 230, 345, 500, and 765 kV, with 800–1,100 kV UHV in China and India · a 765 kV circuit carries roughly 2–4 GW · losses about 2–4% per 1,000 km · roughly $2–6M per mile for a 500 kV double-circuit line in the US, higher in difficult terrain · right of way 45–60 m wide at EHV · typical development-to-energization 7–12 years, most of it permitting.
ExamplesThe 765 kV AEP network in the eastern US, the highest AC voltage in commercial North American service; State Grid's 1,000 kV UHV AC lines in China; the TransWest Express and Grain Belt Express projects, both of which took over a decade to permit; the Ten West Link, whose schedule was set by federal land review rather than construction.
Economic profileCost per MW-mile falls with voltage, which is why utilities standardize on a few classes and build the highest one the load justifies. What has changed is that construction is no longer the long pole. On a typical US project, engineering and construction take two to four years and siting, permitting, and litigation take five or more, so the interest cost of the delay can rival the steel. That imbalance is the reason reconductoring and dynamic ratings have become interesting: they buy capacity inside a corridor that is already approved.
VideosNational Transmission Needs Study (US Department of Energy) · Long-Term Reliability Assessment (NERC)
Aluminum conductor steel-reinforced is the wire on most overhead lines built since the 1930s. Aluminum strands carry the current because aluminum has about 61% of copper's conductivity at a third of the weight and a fraction of the price. A galvanized steel core carries the mechanical tension, because aluminum on its own would stretch and sag. The two jobs are separated on purpose, and the ratio between them is what distinguishes the variants. Related types swap the core or the aluminum: AAAC is all aluminum alloy, ACAR mixes alloys, ACSS uses fully annealed aluminum that can run hotter.
Strengths & weaknessesIt is cheap, universally available from many mills, and its behavior over 60 years of service is thoroughly documented, which matters when a utility is planning a 40-year asset. Stringing crews and hardware are standardized around it. The weaknesses are thermal. Aluminum anneals and loses strength above about 100 °C, so the conductor is normally limited to 75–100 °C, and at that limit it sags into the clearance envelope. Sag is what actually caps a line's rating on a hot still day, not the metal's ability to carry current. The steel core also carries no useful current and adds weight that the structures have to hold up.
When to useUse ACSR for any new line where the rating is set by economics rather than by an existing structure, and for like-for-like replacement on lines whose structures were designed for it. It is the right default and the cost baseline every advanced conductor is measured against. Move to a high-temperature low-sag conductor when the constraint is the corridor rather than the wire: an existing line that needs more capacity, a river crossing where clearance is tight, or a route where new structures are unaffordable. Paying two to three times more per meter is worth it only when it avoids new structures or a new right of way.
Key numbersAluminum conductivity about 61% of copper at roughly one third the weight · continuous operating limit typically 75–100 °C · full annealing of hard-drawn aluminum begins near 100 °C · common sizes carry 500–1,500 A per conductor, bundled two to four per phase at EHV · conductor is a small share of line cost, and structures and right of way are most of it.
ExamplesNearly every transmission line built in North America between 1930 and 2010; the standard Drake, Rail, and Bluebird sizes that utilities specify by codeword; ACSS variants used where a line has to run hotter without changing structures.
VideosAdvanced conductors, also called high-temperature low-sag, replace the steel core with a composite and the hard-drawn aluminum with a fully annealed grade. A carbon-fiber or ceramic-fiber core has a thermal expansion coefficient close to zero, so the conductor barely sags as it heats, and annealed aluminum can run to 180–200 °C without losing strength. That combination roughly doubles the current a given conductor size can carry within the same clearance envelope. The lighter core also leaves room for more aluminum in the same diameter, which lowers resistance and cuts losses at normal load.
Strengths & weaknessesDoubling capacity without touching structures or rights of way is a genuinely different kind of option, and lower resistance means 25–40% less line loss at the same current, which pays back on energy alone. The catch is handling. Composite cores are notch-sensitive and can be damaged by ordinary stringing practices, so crews need training and different hardware, and a mishandled section can fail later. Cost per meter runs two to three times ACSR, splices and dead-ends are proprietary, and long-term field data covers about two decades rather than six.
When to useReach for an advanced conductor whenever the binding constraint is the corridor. Reconductoring an existing 345 kV line to twice its rating in 18–36 months beats a new line that takes a decade, and the Berkeley and GridLab analysis found this route could quadruple the rate of US capacity expansion. It also suits long spans and crossings where sag governs. Stay with ACSR when structures are being replaced anyway, since a new line can just be built at a higher voltage or with more conductors, and when the utility has no crews trained on composite cores and no appetite to build that capability for one job.
Key numbersContinuous operation to 180–200 °C against 75–100 °C for ACSR · roughly 2x the ampacity of the ACSR it replaces in the same envelope · 25–40% lower losses at equal current · two to three times ACSR's cost per meter, though conductor is a minority of project cost · reconductoring typically takes 18–36 months against 7–12 years for a new line.
ExamplesCTC Global's ACCC with a carbon and glass fiber core, the most widely deployed type; 3M's aluminum matrix composite conductor; TS Conductor's carbon-core design; the Illinois and Minnesota reconductoring dockets, where utilities have proposed advanced conductors instead of new corridors.
Economic profileThe economics rest on avoided permitting rather than on the wire. A reconductoring project stays inside an approved right of way, which removes the single largest schedule and cancellation risk in transmission. The Berkeley and GridLab study put the consumer saving from a nationwide reconductoring program at $85 billion by 2035. The counterargument is that a reconductored line is still an AC line with the same impedance and the same power-flow behavior, so it relieves thermal congestion and does nothing about stability limits.
VideosAccelerating transmission capacity expansion by using advanced conductors in existing right-of-way (PNAS, via PMC) · Reconductoring with Advanced Conductors Can Accelerate the Rapid Transmission Expansion Required for a Clean Grid (GridLab and UC Berkeley)
Reconductoring means pulling the old conductor off an existing line and stringing new conductor on the same structures. It is a construction practice rather than a technology, and it has been ordinary utility maintenance for decades. What changed is the arithmetic: with advanced conductors the replacement can double the line's rating instead of merely restoring it, so a job that used to be about replacing worn wire is now a way to add capacity. Structures usually need checking and sometimes reinforcing, since a heavier or hotter conductor changes the loads and the clearances.
Strengths & weaknessesIt uses an approved right of way, which sidesteps the part of transmission development that actually fails. Timelines run 18–36 months against 7–12 years, and the work is familiar to line crews. The limits are physical and operational. Structure condition sets what is possible, and a line built in 1965 may need foundation or steel work that erodes the savings. The line has to be out of service during the work, so the surrounding system has to carry the load, and on a constrained system there may be no outage window. Doubling thermal rating also does nothing for a line limited by stability or voltage rather than heat.
When to useReconductor when a specific line is thermally constrained, its structures are sound, and an outage is schedulable. That describes a large share of the US network. Check the binding constraint first: if flowgate limits come from stability or voltage collapse, more ampacity buys nothing and the answer is series compensation, a STATCOM, or storage. Also compare against dynamic line rating, which is cheaper and faster and often unlocks 10–30% on the same line without an outage. The two stack: rate the line dynamically first, and reconductor when that runs out.
Key numbers18–36 months from decision to energization · up to 2x rating with advanced conductors, roughly 1.2–1.4x with a larger conventional conductor · typically 30–60% of the cost per MW of building a new line · needs a line outage, so system conditions gate the schedule · structure reinforcement is the usual cost surprise.
ExamplesUtility filings in Illinois, Minnesota, and Texas proposing advanced-conductor reconductoring in place of new corridors; the Netherlands' TenneT program of uprating existing 380 kV routes; numerous river-crossing spans reconductored to gain clearance rather than capacity.
VideosStructures hold conductors in the air at a safe distance from the ground, from each other, and from anything that might grow or drive underneath. Three families cover almost everything. Lattice towers are bolted steel angle, light for their strength and cheap in material, but slow to assemble and wide at the base. Tubular steel monopoles take a much narrower footprint and go up in a day, at higher steel cost. Wood and concrete poles serve lower voltages and distribution. The choice is driven by right-of-way width, terrain access, and how quickly the line has to be built.
Strengths & weaknessesLattice is the cheapest way to hold a heavy EHV bundle across a long span, and its open frame takes wind well. Monopoles win where land is expensive or the public objects to the look, and they can be set with a crane in remote terrain that a bolting crew cannot reach easily. Both are commodity steel, exposed to fabrication capacity and tariff policy. The real constraint is foundations: rock, wetland, and permafrost each demand a different design, and geotechnical surprises are the most common source of cost overrun on a line project. Helicopter placement is routine in mountains and expensive everywhere.
When to useChoose lattice for long EHV lines across open country where right-of-way width is available and material cost dominates. Choose monopoles in developed corridors, along highways and railways, and anywhere the public process is sensitive to visual impact, since a narrower base and cleaner profile genuinely helps a permit. For a rebuild, check whether the existing structures can carry a heavier conductor before assuming they must be replaced, because keeping them is what makes reconductoring fast. If new structures are needed anyway, the decision reopens the voltage question, since going up a class is often cheaper than a second circuit later.
Key numbersTypical EHV span 300–400 m, up to 1–2 km at major river crossings · lattice right of way 45–60 m, monopole often 25–40 m · structure and foundation are typically 30–50% of line cost · monopole erection in about a day against several days for lattice · foundations are the most common source of cost overrun.
ExamplesThe Big Eddy to Knight 500 kV river crossing in Oregon, built on lattice; monopole rebuilds along interstate corridors in the US northeast; China's Zhoushan crossing towers, the tallest transmission structures in the world at 380 m.
VideosA line's rating is set by how hot the conductor gets, and that depends on ambient temperature, wind, and sun as much as on current. Static ratings assume a conservative worst case, typically 40 °C still air with full sun, all year. Dynamic line rating measures or forecasts the real conditions and calculates the rating that actually applies right now. Implementations range from ambient-adjusted ratings, which use a temperature forecast alone, to sensor-based systems that measure conductor tension, sag, or the local wind field on the span that governs.
Strengths & weaknessesThe upside is that wind cools a conductor hard, and wind is highest exactly when wind generation is highest, so the extra capacity tends to arrive when the system wants it. Typical gains run 10–30% and can exceed 50% on windy days, for a cost measured in hundreds of thousands of dollars rather than hundreds of millions. Nothing needs an outage or a permit. The problems are operational rather than physical. Operators have to trust a rating that moves, protection settings and market software must accept time-varying limits, and a rating that increases must never be relied on for a contingency that occurs when the wind drops. Sensors on remote spans need power and communications.
When to useDeploy dynamic ratings on lines that are congested for a limited number of hours a year, where the congestion cost is real and a rebuild is years away. It is the cheapest capacity in this sheet and the fastest to install. Do not use it where the constraint is stability or voltage, and do not plan firm capacity around it, since the conservative case is what backs up a contingency. In the US, FERC Order 881 has made ambient-adjusted ratings mandatory for near-term operations, so the baseline has moved and full dynamic rating is the increment above it.
Key numbersStatic ratings typically assume 40 °C still air with full sun · typical dynamic gain 10–30%, above 50% in high wind · installed cost of a monitored line in the hundreds of thousands of dollars · no outage and no permit needed · FERC Order 881 requires ambient-adjusted ratings for operations within the next 10 days.
ExamplesPPL Electric Utilities' deployment in Pennsylvania, which won industry recognition for congestion savings; National Grid's UK sensor programs; EPRI's dynamic line rating test facility; LineVision and Ampacimon as the main sensor vendors.
VideosFERC Order No. 881: Managing Transmission Line Ratings (ISO New England)
Underground & subsea
cable systems and their thermal limits4 technologiesUnderground transmission cable replaces air insulation with cross-linked polyethylene extruded over the conductor, wrapped in semiconducting screens, a metallic sheath, and an outer jacket. Cross-linking turns thermoplastic polyethylene into a thermoset that holds its shape at operating temperature, which is what allows 90 °C conductor ratings. Cables go in ducts, direct-buried in a thermal backfill, or in a tunnel. Because the earth is a poor conductor of heat, how the cable sheds heat into its surroundings, not the copper itself, usually decides the rating.
Strengths & weaknessesCable is invisible, immune to weather and vegetation, and it is often the only thing that gets permitted through a city. It has no exposure to lightning or ice. The costs are severe. Installed cost runs 5–15 times an equivalent overhead line, faults take weeks to locate and repair instead of hours, and cable has high shunt capacitance, so on AC above roughly 50–80 km the charging current consumes the conductor's own capacity and reactive compensation is needed at both ends. Joints are the reliability weak point and are made by hand in the field.
When to useUse cable where an overhead line cannot go: dense urban areas, airport approaches, protected landscapes, and short crossings where a route is otherwise blocked. It is also the standard for the last few kilometers into a city substation. Do not use it to avoid a public fight over a long rural line unless the buyer understands the multiple. And beyond about 50–80 km on AC, stop and evaluate DC, since the charging current problem disappears entirely on DC and the cost crossover moves in DC's favor.
Key numbersConductor rated to 90 °C continuous, 105 °C emergency · available to 500 kV AC, with 525 kV DC types in service · installed cost roughly 5–15x an overhead line of the same rating · AC length limit around 50–80 km before reactive compensation dominates · fault repair measured in weeks · joints every 500–1,000 m, each made by hand.
ExamplesLondon Power Tunnels, a multi-billion-pound program putting 400 kV circuits under the city; the Randstad 380 kV underground sections in the Netherlands; almost every substation getaway in a dense downtown; 35 kV distribution feeders, where XLPE has entirely replaced paper-insulated lead-covered cable.
VideosEnergy Infrastructure (Europacable) · An Introduction to Medium and Low Voltage Cables in Distribution Networks (Europacable)
Subsea cable is what connects grids across water, and beyond a few tens of kilometers it is always DC. Two insulation systems compete. Mass-impregnated cable wraps the conductor in many layers of paper tape soaked in a high-viscosity compound; it has been in service since the 1950s, works to 600 kV and great depth, and has the longest record. Extruded XLPE DC cable is lighter, cheaper to make, and easier to joint, and has moved up to 525 kV with 2,600 MW ratings. Cable is laid from a specialist vessel and usually buried in the seabed by a plow or jetting tool.
Strengths & weaknessesDC removes the charging-current limit that caps AC cable length, so a link can run 700 km or more, and it gives the operator direct control of how much power flows and in which direction. Subsea routes avoid landowners entirely for most of their length. The problems are supply chain and repair. A handful of manufacturers and even fewer laying vessels exist worldwide, so slots are booked years ahead, and a single project can absorb a factory's output for a year. A fault means finding the break, bringing a repair vessel, and lifting the cable, which takes months and costs tens of millions.
When to useSubsea DC is the answer whenever power has to cross water further than an AC cable can reach, which covers offshore wind export, island interconnection, and cross-border links. It is also worth considering for long land routes where a right of way is unobtainable and a buried DC cable can follow a road or rail corridor. Plan procurement first and engineering second, since cable and vessel slots, not design, set the schedule. Budget for spare cable and a repair agreement from the start, because a link with no spare is a link with a months-long outage waiting to happen.
Key numbersMass-impregnated to 600 kV; extruded XLPE DC to 525 kV and about 2,600 MW · routes to 700+ km in service · installed cost roughly $1.5–4M per km depending on rating and depth · lead time for cable and vessel commonly 3–5 years · repair takes months and needs a specialist vessel.
ExamplesNorth Sea Link, 720 km between Norway and the UK; NordLink; Viking Link between the UK and Denmark; the NeuConnect interconnector; the many offshore wind export cables now being built in the North Sea and off the US east coast.
VideosAn Introduction to High Voltage Direct Current (HVDC) Subsea Cables Systems (Europacable)
A gas-insulated line is a rigid aluminum tube with a coaxial aluminum conductor held on insulators inside it, filled with pressurized gas. It is the same construction as gas-insulated switchgear stretched out over kilometers. Because the insulating medium is gas rather than polymer, the capacitance per meter is close to an overhead line's rather than a cable's, so a GIL can run long distances on AC without the charging-current problem that limits XLPE. Original designs used SF6; current ones use SF6 and nitrogen mixtures or fluoronitrile blends to cut the greenhouse impact.
Strengths & weaknessesGIL carries overhead-line power underground: a 420 kV system moves around 2,000 MVA, buried or in a tunnel, with low losses and no length limit from capacitance. It has no fire load and a very small magnetic field outside the enclosure. Against that, it is expensive, several times an equivalent XLPE cable, it needs a wide trench or a tunnel because the tubes are large and rigid, and the gas is a liability: SF6 has a global warming potential around 23,500 times CO2, so leak monitoring and end-of-life recovery are permanent obligations. Installation is a precision welding job, not a cable pull.
When to useConsider GIL for short high-power underground sections where a cable cannot carry the rating, typically the last kilometers into a city substation, tunnels through mountains, and getaways from a hydro plant or a GIS substation. It suits situations where the full overhead rating has to continue underground with no derating. Do not use it for long routes, where cost rules it out, and do not choose an SF6-filled system where the utility has a fluorinated-gas phase-down commitment. If the length is a few hundred meters and the rating is moderate, XLPE cable is far cheaper.
Key numbers420 kV systems rated around 2,000 MVA, matching an overhead circuit · losses roughly half those of an equivalent cable at full load · capacitance per meter close to an overhead line, so no practical AC length limit · SF6 global warming potential about 23,500x CO2, which is why nitrogen mixtures and fluoronitrile blends are now standard · cost several times XLPE cable per meter.
ExamplesThe Kelsterbach GIL at Frankfurt airport, one of the longest directly buried installations; the Limberg II pumped-storage connection in Austria; 420 kV fluoronitrile-insulated GIL sections installed in England and Scotland; tunnel installations at hydro plants where an overhead getaway is impossible.
VideosGas-Insulated Transmission Lines: Underground Power Transmission Achieving a Maximum of Operational Safety and Reliability (Jicable) · Electric Power Systems Partnership (US EPA)
A high-temperature superconducting cable carries current in tapes of rare-earth barium copper oxide cooled to about 70 K by circulating liquid nitrogen inside a vacuum-insulated pipe. Below its critical temperature the conductor has essentially zero DC resistance, so a cable a few inches across can carry the current of a much larger conventional circuit at medium voltage. A useful side effect is fault-current limiting: when a fault drives current past the critical value the tape goes resistive in milliseconds and chokes the fault, which is why several installations were justified on protection rather than capacity.
Strengths & weaknessesVery high power density is the point. A 12 kV superconducting cable can carry what would otherwise need a 138 kV circuit and a substation, which matters when a downtown has no room for either. Losses in the conductor are near zero, and the magnetic field outside is negligible. The costs are the cryogenics. A refrigeration plant has to run continuously, consuming power and needing maintenance, and if it stops the cable stops. Tape is expensive, joints are specialized, and the installed base is a handful of demonstrations rather than a fleet, so utilities have no long-run reliability data to plan against.
When to useConsider it only where the constraint is space, not distance: a dense urban core where new duct banks and substations are impossible, or a plant where a bus tie has to carry enormous current across a short run. The fault-limiting behavior makes it interesting where two substations would otherwise be tied together and raise fault duty beyond breaker ratings. Do not consider it for long routes, where cryogenics per kilometer make no sense, and do not plan around cost curves, since tape production is still small. For essentially every other duty, conventional cable at a higher voltage is cheaper and better understood.
Key numbersOperates near 70 K with liquid nitrogen · roughly zero DC conductor resistance, so current density is orders of magnitude above copper · Chicago's Resilient Electric Grid installation is a 200 m cable rated 3,000 A at 12 kV, about 62 MVA · fault current limiting acts within milliseconds · continuous refrigeration load is the standing operating cost.
ExamplesComEd's Resilient Electric Grid project in downtown Chicago, commissioned in 2021 and the first permanent North American installation; the AmpaCity project in Essen, Germany, a 1 km medium-voltage cable replacing a 110 kV line; NEDO's demonstration in Japan; LS Cable's Korean installations.
VideosHVDC converters
turning AC into DC and back3 technologiesLine-commutated converter HVDC is the original and still the highest-power form of DC transmission. Banks of thyristors, switched in sequence, rectify AC to DC at one end and invert it back at the other. Thyristors turn on when gated and turn off only when the AC line voltage reverses, which is where the name comes from: the AC system commutates the valves. That dependency shapes everything about the technology. It needs a reasonably strong AC system at both ends, it draws large reactive power, and it requires big harmonic filter yards.
Strengths & weaknessesLCC handles the biggest ratings anywhere: 800 kV and 1,100 kV UHV links move 8–12 GW over 2,000–3,000 km, and converter losses are the lowest of any HVDC type at about 0.7% per station. Thyristors are rugged and tolerate overloads. The limitations follow from line commutation. The converter consumes 50–60% of its rated power in reactive power, so filter and capacitor yards are large and expensive; it cannot start into a dead network, so black start needs help; and it is prone to commutation failure when the receiving AC voltage dips, which briefly interrupts power transfer.
When to useChoose LCC for the largest point-to-point transfers over long distances into strong AC systems, which in practice means China's UHV program, India's national links, and North American ties like Pacific DC. It is also the right answer when losses over thousands of kilometers dominate the economics. Avoid it for offshore wind, weak grids, and anything that needs to energize a passive network or feed a converter station with no local generation, since it physically cannot. Those duties belong to VSC, which costs more per MW and buys independence from the AC system.
Key numbersRatings to 800 kV and 1,100 kV, with links of 8–12 GW · converter losses around 0.7% per station, the lowest of any HVDC · reactive consumption 50–60% of rated power, needing large filter yards · overhead DC line losses roughly 3% per 1,000 km · converter stations typically 3–5 years to deliver, with the full link 7–10 years.
ExamplesChangji-Guquan in China, 1,100 kV and 12 GW over 3,300 km, the largest transmission link ever built; the Pacific DC Intertie between Oregon and Los Angeles, running since 1970; India's Raigarh-Pugalur 800 kV link; Itaipu's 600 kV bipoles in Brazil.
Economic profileHVDC pays back the cost of two converter stations by saving on line, losses, and control. The break-even against AC is roughly 600–800 km for overhead lines and 40–70 km for cable, and the cable case is what has driven most recent projects. Converter cost scales with rating rather than with distance, so long links have better economics than short ones, and a 3,000 km UHV bipole spreads its terminal cost over enormous throughput. That is the arithmetic behind China building dozens of them.
VideosHigh-Voltage Direct Current (HVDC) (Hitachi Energy) · National Transmission Needs Study (US Department of Energy)
Voltage-source converter HVDC uses IGBTs, which turn off on command, so the converter no longer depends on the AC system to commutate. Modern designs are modular multilevel converters: hundreds of small submodules in series, each switched in or out to build a staircase that approximates a sine wave. The output is clean enough that the huge harmonic filter yards of an LCC station disappear. Because the converter controls its own output voltage and angle, it can set real and reactive power independently, and it can energize a network that has no generation of its own.
Strengths & weaknessesIndependence from the AC system is what makes VSC the default for anything new. It works into weak or passive grids, provides reactive support like a STATCOM whether or not power is flowing, can black-start a dead network, and reverses power flow without reversing DC voltage polarity, which is what makes multi-terminal schemes practical. The station footprint is roughly half an LCC's. The costs: converter losses run about 1% per station against 0.7%, IGBT stacks cost more per MW, and top ratings are lower, in the 2–3 GW range against 8–12 GW for UHV LCC.
When to useUse VSC for offshore wind export, city infeeds, links into weak grids, any scheme with more than two terminals, and anywhere the converter's reactive support has value in its own right. That covers nearly all HVDC being ordered in Europe and the US today. LCC still wins for the largest overhead point-to-point transfers into strong systems, where its lower losses and higher ratings matter more than flexibility. Whichever you pick, procure early: converter transformers and IGBT supply are both bottlenecked, and delivery drives the project schedule more than engineering does.
Key numbersTypical ratings 1–3 GW per link at 320–525 kV DC · converter losses about 1% per station · reactive power controllable across the full range, independent of active power · station footprint roughly half an LCC's · lead times of 3–5 years for converters, and longer where a supplier's order book is full.
ExamplesThe German offshore grid connections at 320 and 525 kV, which made VSC the standard for North Sea wind; Champlain Hudson Power Express into New York City; INELFE between France and Spain; Kriegers Flak, which links Danish and German grids through a wind farm.
VideosReliability Guideline: BPS-Connected Inverter-Based Resource Performance (NERC)
Almost every HVDC link built so far runs point to point: one converter in, one converter out, one contract. A multi-terminal scheme connects three or more converters to a shared DC network, so power can be routed rather than merely shipped. The attraction is offshore. A meshed DC grid in the North Sea would let wind farms feed whichever shore needs power, and would let the same cables carry trade between countries when the wind is low. Making it work needs DC circuit breakers, converters from different vendors that interoperate, and a protection philosophy that does not exist yet in mature form.
Strengths & weaknessesA shared DC network uses far fewer cables and converters than a set of radial links doing the same job, and it turns single-purpose wind export assets into interconnectors that earn between weather events. Redundancy improves because losing one terminal need not lose the whole scheme. The obstacles are interoperability and protection. A DC fault propagates across the whole DC network in milliseconds with no natural current zero to help clear it, so either every converter must ride through or DC breakers must act faster than any AC breaker. Vendors' control systems have historically not worked together, which is why nearly every scheme is single-supplier.
When to usePlan for multi-terminal where several offshore projects will connect in the same sea area over a decade, since designing radial links one at a time forecloses the option. Regulators in the North Sea have started requiring it for exactly that reason. Do not commit a first-of-a-kind meshed scheme to a critical delivery date, and do not assume equipment from two vendors will interoperate without a demonstrated multi-vendor test program. For a single wind farm to a single shore, a point-to-point VSC link is cheaper, simpler, and available now.
Key numbersThree or more converter terminals on a shared DC network · a DC fault spreads across the network in milliseconds, against tens of milliseconds for AC clearing · European offshore plans envisage tens of GW of meshed capacity by the 2030s · almost all operating schemes today are single-vendor · interoperability testing is the pacing item, not converter hardware.
ExamplesThe Zhoushan and Zhangbei multi-terminal projects in China, the largest operating examples; Kriegers Flak, which links Denmark and Germany through an offshore wind farm; the North Sea Wind Power Hub and related European programs; the InterOPERA project, set up specifically to make multi-vendor HVDC work.
VideosHVDC apparatus
the hardware a DC grid still needs2 technologiesAn AC breaker gets help from physics: current passes through zero a hundred or a hundred and twenty times a second, and the arc goes out at one of those crossings. DC current never does, so a DC breaker has to force a zero itself. The workable design is hybrid. A low-loss mechanical path carries the current normally; on a fault, a small semiconductor commutating circuit pushes the current into a parallel power-electronic branch, the mechanical contacts open with almost no arc, and the semiconductor branch then blocks against a surge arrester bank that absorbs the stored inductive energy. All of it in a few milliseconds.
Strengths & weaknessesA working DC breaker is what allows a meshed DC grid to lose one line instead of losing everything, and hybrid designs have demonstrated clearing in 2–5 ms with normal running losses close to a mechanical switch. The problems are size, cost, and count. The energy-absorbing arrester stacks are large, a breaker can cost a meaningful fraction of a converter, and a meshed grid needs one at every line end. The installed base worldwide is small, so failure statistics do not exist. Some system designers avoid them entirely by using fault-blocking converters, which cost more in losses but need no breakers.
When to useYou need DC breakers only if you are building a meshed or multi-terminal DC network where a single line fault must not shut the whole scheme down. Point-to-point links do not need them, because the AC breakers at each end can clear the fault by de-energizing the converters. Compare the DC breaker route against full-bridge fault-blocking converters early, because the two are alternative answers to the same problem: breakers cost capital and space, blocking converters cost roughly an extra half point of efficiency forever.
Key numbersClearing in about 2–5 ms, against 40–80 ms for AC breakers · on-state losses near those of a mechanical disconnector in hybrid designs · surge arrester banks sized to absorb the DC network's stored inductive energy · commercial installations remain in the low tens worldwide · alternative is a full-bridge converter, which blocks faults at the cost of roughly 0.5 percentage points of efficiency.
ExamplesABB's hybrid HVDC breaker, the first demonstrated design; the Zhangbei project in China, the largest deployment of DC breakers in a meshed scheme; European research programs including PROMOTioN, which tested competing designs against a common specification.
VideosA back-to-back converter is an HVDC link with no line: rectifier and inverter sit in the same building, connected by a short DC bus. There is no transmission benefit at all. The point is that the two AC systems on either side are now electrically separate, so they can run at different frequencies, different phase angles, or simply out of step, and power still flows between them under full control. It is the standard way to tie asynchronous grids together and to place a controllable valve inside a synchronous grid that would otherwise route power wherever impedance sends it.
Strengths & weaknessesFull control of flow is the product. The operator sets megawatts, and unlike an AC tie the number is respected regardless of what the surrounding network does. Faults do not propagate across the link, so a disturbance on one side is contained. Modern VSC versions add reactive support on both sides. The costs are the same converter costs as any HVDC scheme, roughly 1–2% of throughput lost in conversion, plus the fact that you are paying converter prices for zero distance. A back-to-back station is expensive per MW compared with an AC tie that just needs a transformer.
When to useUse back-to-back when the two systems cannot be synchronized: the ERCOT ties to the eastern and western US, Japan's 50 Hz and 60 Hz halves, and most cross-border links in developing regions. Inside one synchronous system, use it where loop flow is causing real congestion and a controllable tie is worth the converter cost, which is the argument behind several proposed US seams projects. Do not use it where an AC tie would work and the flow does not need controlling, since the converter is pure added cost and loss.
Key numbersNo DC line, so no distance-related loss · conversion loss roughly 1–2% total across both converters · typical ratings 100–1,000 MW per station · sets power flow exactly rather than by impedance · isolates faults and frequency disturbances between the two systems.
ExamplesThe ERCOT direct-current ties, which let Texas trade with its neighbors without joining them; Japan's Higashi-Shimizu and Sakuma converters between the 50 Hz and 60 Hz systems; the Eagle Pass tie between Texas and Mexico; several Indian regional seam stations installed before the national grid was synchronized.
VideosSubstation apparatus
transformers, switchgear, and compensation5 technologiesA large power transformer steps voltage up at a generator and down at every level below it, and there is no substitute. Windings of copper or aluminum sit on a core of grain-oriented electrical steel, all immersed in insulating oil inside a steel tank with radiators and a tap changer. Ratings run from tens of MVA to over 1,000 MVA, and a large unit weighs 200–400 tons. Efficiency is around 99.5%, which sounds unremarkable until you multiply the remaining half percent by continuous operation for forty years.
Strengths & weaknessesThe technology is mature, extremely reliable, and long-lived: units routinely run 40 years and many US transformers in service are older. Losses are low and maintenance is mostly oil testing. The weakness is supply. Grain-oriented electrical steel comes from a handful of mills, large units are built to order rather than stocked, and demand has risen sharply. Lead times that were 12–18 months before 2020 are now commonly 80–120 weeks and longer for generator step-up units. There are over 80,000 distinct transformer designs in the US fleet, so spares are rarely interchangeable, and moving a 300-ton unit needs specialized rail or road transport.
When to useThere is no choice about whether to use one, only about how to procure it. Order early, treat delivery as the project's critical path, and standardize designs across a fleet so a spare can serve more than one site. Utilities facing a firm energization date increasingly buy a slot before the design is final. Where a replacement is urgent, look at refurbished units and mobile transformers, which trade efficiency and rating flexibility for availability in weeks. For a new load that can wait, ordering ahead of need is now normal practice rather than over-buying.
Key numbersRatings from tens of MVA to 1,000+ MVA · efficiency about 99.5% at rated load · weight 200–400 tons for a large unit · service life typically 40 years, with much of the US fleet already past 30 · lead times commonly 80–120 weeks, and around 144 weeks for generator step-up units in 2025 · Wood Mackenzie modeled a 30% supply shortfall for power transformers in 2025.
ExamplesThe US fleet's aging population, flagged repeatedly in DOE and NIAC assessments; new domestic capacity announcements from Hitachi Energy, Siemens Energy, and Prolec GE; utilities running shared spare programs such as the Grid Assurance and STEP reserves.
Economic profilePrice has roughly doubled since 2020, and delivery has become the binding constraint on interconnection and load growth, which changes how projects are financed. A data center or a wind farm can be fully permitted and financed and still wait two years for a step-up transformer. The supply chain runs on grain-oriented electrical steel, which has limited domestic production, so tariff and trade policy feed straight into grid timelines.
VideosTransformer supply bottleneck threatens power system stability as load grows (Utility Dive) · Power transformers and distribution transformers will face supply deficits of 30% and 10% in 2025 (Wood Mackenzie)
Gas-insulated switchgear puts the busbars, disconnectors, breakers, and instrument transformers of a substation inside grounded metal enclosures filled with pressurized insulating gas. Because the gas has several times the dielectric strength of air, clearances shrink dramatically, and a 400 kV bay that would occupy a field outdoors fits in a room. The whole assembly is sealed, so salt, dust, and humidity never reach live parts. Historically the gas was SF6; new equipment increasingly uses fluoronitrile blends or pure synthetic air with vacuum interruption.
Strengths & weaknessesFootprint is the reason GIS exists: roughly 10% of an equivalent air-insulated substation at EHV. Sealing gives it excellent performance in coastal, desert, and heavily polluted environments and very high availability. Against that, capital cost runs two to four times an air-insulated substation of the same rating, repairs mean opening a gas compartment and are slower, and internal faults are harder to locate. SF6 is the most potent greenhouse gas in industrial use, so leak rates, monitoring, and end-of-life recovery are now regulated obligations in Europe and increasingly elsewhere.
When to useUse GIS where land is expensive or unavailable: city substations, indoor installations, offshore platforms, underground caverns, and industrial sites with no spare yard. It is also the right choice in severely polluted or coastal air, where air-insulated equipment needs constant washing. Use air-insulated switchgear where land is cheap, which is most of the rural transmission system, since the cost multiple is hard to justify otherwise. When specifying new GIS, look hard at SF6-free options, because a fluorinated-gas phase-down turns a leaking fleet into a reporting and retrofit problem.
Key numbersFootprint roughly 10% of an equivalent air-insulated substation at EHV · capital cost two to four times air-insulated · available from medium voltage to 1,100 kV · SF6 global warming potential about 23,500x CO2, with regulated leak rates typically under 0.5% per year · SF6-free alternatives now rated to 145 kV and above.
ExamplesUrban substations under buildings in Tokyo, London, and New York; offshore wind platform switchgear, where GIS is the only practical choice; Siemens Energy's clean-air Blue portfolio and Hitachi Energy's EconiQ, both marketed as SF6-free at transmission voltages.
VideosGas-Insulated Switchgear: SF6-Free Blue Technology (Siemens Energy) · Gas-insulated switchgear (GIS) portfolio (Hitachi Energy)
A circuit breaker interrupts fault current, which at transmission voltage means quenching an arc carrying tens of thousands of amps. Every design waits for the natural AC current zero and then makes the gap recover faster than the returning voltage can restrike it. SF6 breakers blow compressed gas across the arc; vacuum breakers open contacts inside a sealed bottle, where there is nothing to ionize. Vacuum has owned medium voltage for decades and is now moving into transmission voltages as SF6 comes under regulatory pressure. Clearing takes two to three cycles, 40–50 ms on a 50 Hz system.
Strengths & weaknessesSF6 breakers interrupt very high currents in a compact envelope and have decades of service data, which is why they dominate above 72 kV. Vacuum breakers need no gas handling, have long contact life, and are essentially maintenance-free. The problems split by type. SF6 is a regulated greenhouse gas with leak reporting and end-of-life obligations. Vacuum bottles get physically difficult above about 145 kV, so transmission-class units use series bottles or hybrid designs, and the installed history at those voltages is short. All breakers have a finite number of fault interruptions before contact maintenance.
When to useSpecify vacuum at distribution and subtransmission voltages, where it is already the standard and cheaper over life. At 145 kV and above, choose SF6 where a proven long-life asset matters and no phase-down applies, and choose a fluoronitrile blend or clean-air vacuum design where it does. Check the interrupting rating against the system's fault duty including future generation, since adding inverter-based resources and stronger ties raises fault levels and can strand a breaker fleet. Where fault duty already exceeds ratings, a fault current limiter is often cheaper than replacing every breaker in the yard.
Key numbersClearing in two to three cycles, roughly 40–50 ms · interrupting ratings commonly 40–63 kA at transmission voltage · vacuum dominant to 72 kV and moving above 145 kV · SF6 leak rates regulated below about 0.5% per year in Europe · mechanical life of 10,000 operations, with far fewer at full fault current.
ExamplesThe SF6 puffer breakers standard in transmission yards worldwide since the 1970s; Siemens Energy's vacuum breakers rated to 145 kV using clean air insulation; GE Vernova's g3 fluoronitrile breakers deployed by National Grid in the UK; distribution reclosers, which are now almost universally vacuum.
VideosA long AC line's capacity is limited by its series inductive reactance, not by how hot the conductor gets. Putting capacitors in series with the line cancels part of that reactance, which shortens the line electrically. Compensating 40% of a line's reactance raises its stable transfer capability by roughly 40–70% and pulls flow onto that line and away from parallel paths. Thyristor-controlled series capacitors add a controllable reactor across part of the bank so the compensation level can be varied in real time, which also damps power oscillations.
Strengths & weaknessesIt is by far the cheapest way to raise the capability of an existing long line, typically a few percent of the cost of building a parallel circuit, and it fits inside an existing substation. Controlled versions damp inter-area oscillations that would otherwise limit transfers. The risk is subsynchronous resonance: a series-compensated line can exchange energy with the torsional modes of a nearby turbine-generator shaft and, in the worst case, damage it. The 1970 and 1971 Mohave shaft failures in Nevada are the reference cases. Any scheme therefore needs a subsynchronous study, and sometimes filters or protective relaying, before it goes in.
When to useUse series compensation on long EHV lines that are stability-limited rather than thermally limited, which describes most 500 kV corridors over about 300 km. It is the standard fix when a line's transfer limit is well below its thermal rating. Check for thermal generation with long shafts near the compensated line and run the subsynchronous study early, since it can rule the scheme out. If the line is thermally limited instead, series capacitors do nothing, and the answers are reconductoring or dynamic ratings.
Key numbersCompensation typically 25–70% of line reactance · raises stable transfer capability by roughly 40–70% at 40% compensation · costs a few percent of building a parallel line · thyristor-controlled versions vary compensation within a cycle and damp oscillations · subsynchronous resonance study is mandatory near steam turbine generators.
ExamplesThe Kayenta thyristor-controlled series capacitor in Arizona, the first of its type; extensive series compensation on Brazil's north-south interconnection and on Sweden's 400 kV network; the New York Power Authority's Marcy South project; India's 765 kV corridors, most of which are series-compensated.
VideosReactive power does not travel well, so voltage has to be managed locally, and the cheapest way is to switch fixed devices in and out. Shunt capacitor banks supply reactive power and raise voltage; they are what utilities add when a heavily loaded feeder or substation sags. Shunt reactors absorb reactive power and lower voltage; they are what a lightly loaded long line or cable needs, because its own capacitance generates reactive power and pushes voltage up. Both are simple, both are switched by a breaker, and both are sized in coarse steps.
Strengths & weaknessesCost per MVAr is the lowest of any reactive device by a wide margin, roughly a tenth of a STATCOM, and there is almost nothing to maintain. Reliability is excellent. The weaknesses come from being passive and discrete. A capacitor's output falls with the square of voltage, so it delivers least exactly when a sagging system needs it most, which is why capacitors alone cannot hold a system through a voltage collapse. Switching produces transients that stress nearby equipment, and control is stepwise rather than continuous, so voltage rides a sawtooth. Capacitors can also resonate with system inductance at harmonic frequencies.
When to useUse fixed and switched banks as the base layer of reactive support everywhere: they carry the steady-state reactive load at the lowest cost. Add reactors wherever long lines or cables run lightly loaded, and remember that cable getaways need them permanently. Reach for a STATCOM or SVC only for the dynamic increment, when the requirement is fast response after a fault or support during a voltage dip. The usual right answer is a layered design: banks for bulk MVArs, dynamic devices for the last fast part.
Key numbersCapacitor output falls with the square of voltage, so a 10% dip gives 19% less support · cost roughly a tenth of a STATCOM per MVAr · switching in discrete steps, commonly 25–150 MVAr per bank at transmission voltage · reactors are effectively mandatory on long cable circuits · resonance with system inductance requires a harmonic study on large banks.
ExamplesCapacitor banks on distribution feeders, the single most common reactive device on any grid; shunt reactors at both ends of long 400 and 500 kV lines and on nearly every EHV cable circuit; switched banks used seasonally to follow summer air-conditioning load.
VideosProtection & control
relays, telemetry, and substation automation4 technologiesProtective relays decide, in milliseconds, whether what they are measuring is a fault and which breaker should open. On transmission lines the workhorse is distance protection, which computes apparent impedance from voltage and current: a fault looks like a much lower impedance than load, and how low tells the relay roughly how far away it is. Differential protection compares current in against current out of a transformer, bus, or line, and trips on any difference. Overcurrent protection, the oldest scheme, simply trips above a threshold with a time delay that coordinates with devices downstream.
Strengths & weaknessesModern microprocessor relays are cheap for what they do, self-monitoring, and pack line protection, metering, fault recording, and communications into one box. They are the reason a modern grid clears a fault in three cycles. The difficulty is coordination rather than hardware. Settings must trip for every fault in a zone and never for anything outside it, across all system configurations, and a large utility maintains tens of thousands of settings files. Inverter-based resources break assumptions built into these schemes: they contribute little fault current, and their current is controlled by software rather than by physics, so distance elements can under-reach or misjudge direction.
When to useProtection is not optional, so the real choices are scheme and communications. Use line differential where a fiber channel exists between the two ends, since it is the most selective and fastest scheme available. Use distance protection with a communications-assisted trip where fiber is not available end to end. Reserve plain time-overcurrent for radial distribution. Wherever a line terminates near a large inverter-based plant, re-study the settings rather than reusing them, because that is where the recent misoperation reports cluster.
Key numbersFault clearing in two to three cycles, roughly 40–50 ms, including breaker time · distance relays typically set with a zone 1 reaching 80–90% of line length · line differential needs a fiber channel with sub-10 ms latency · inverter-based resources contribute roughly 1.1–1.5 times rated current against 5–6 times for a synchronous machine · a large utility manages tens of thousands of settings files.
ExamplesThe SEL-421 and equivalent transmission-class relays that define current practice; NERC's disturbance reports on inverter-based resource misoperation in the Western Interconnection, which drove new performance requirements; digital substations where relays exchange sampled values over Ethernet rather than copper.
VideosReliability Guideline: BPS-Connected Inverter-Based Resource Performance (NERC)
SCADA is the telemetry layer: remote terminal units in substations report analog measurements and breaker states every two to four seconds and accept control commands back. The energy management system sits on top and turns those numbers into a picture. A state estimator reconciles redundant, slightly inconsistent measurements into a single consistent network model, contingency analysis runs thousands of "what if this line trips" cases against it, and automatic generation control adjusts generator setpoints to hold frequency and interchange. Everything a control room does rests on that estimated model being right.
Strengths & weaknessesIt gives one operator a coherent view of thousands of devices, and it is the reason contingencies get caught before they happen rather than after. The weaknesses are refresh rate, model quality, and security. A four-second scan cannot see an oscillation that grows in a second, which is what synchrophasors were built for. State estimation depends on a network model that matches reality, and a mis-keyed line impedance or an unreported switch position quietly corrupts every downstream calculation, which is a documented contributor to the 2003 North American blackout. And a control system that can open breakers is a target, which is what happened in Ukraine in 2015.
When to useEvery transmission operator runs SCADA and an EMS; the questions are scope and refresh. Extend telemetry down into distribution when distributed generation makes the feeder no longer a passive load, which is where an advanced distribution management system starts to pay. Add synchrophasor measurement when the concern is dynamics rather than steady state. Treat model maintenance as an ongoing operational job rather than a project, since a stale model degrades every tool built on it. Keep the security boundary strict: the operational network should not share a path with the corporate one.
Key numbersSCADA scan rates of two to four seconds, against 30–60 samples per second for synchrophasors · state estimation typically runs every one to five minutes · contingency analysis screens thousands of cases per cycle · a model error propagates silently into every result · NERC CIP standards govern the security perimeter in North America.
ExamplesRegional control rooms at PJM, MISO, and National Grid ESO; the 2003 Northeast blackout, where a failed alarm system left operators without a working picture for over an hour; the 2015 Ukraine attack, in which intruders operated breakers through the utilities' own control systems.
VideosA phasor measurement unit samples voltage and current 30 to 120 times a second and stamps each measurement with GPS time, accurate to about a microsecond. Because every unit shares the same clock, measurements from hundreds of substations can be compared directly, and the phase angle between two ends of the system becomes a directly observed quantity rather than something a state estimator infers. That changes what operators can see: oscillations, angular separation, and the early signs of instability all live at timescales a four-second SCADA scan cannot resolve.
Strengths & weaknessesSynchrophasors make dynamics visible. They catch inter-area oscillations, verify generator models against real disturbances, and give post-event analysis a millisecond-resolution record instead of a set of sequence-of-events logs. The obstacles are data and use. A modest deployment produces terabytes a year, most of which nobody looks at, and the value comes from applications built on the stream rather than from the sensors. The other constraint is the clock: GPS timing is the single point of dependence, and spoofing or jamming corrupts the measurement in a way that looks like a real phase shift.
When to useDeploy synchrophasors where dynamics matter: long interconnections prone to inter-area oscillation, systems with a high share of inverter-based generation, and any network where model validation has become a recurring problem. Budget for the applications and the analysts, not just the units, since a deployment that produces archives nobody uses is common. Do not expect them to replace SCADA, which remains the control path. And plan alternate timing, because a system that depends entirely on GPS for measurement validity has a known single failure mode.
Key numbers30–120 samples per second, against a two to four second SCADA scan · time accuracy of about one microsecond, needed because one degree at 60 Hz is 46 microseconds · roughly 2,500 units across North America, up from a handful before 2009 · a single unit generates gigabytes per month · GPS is the usual time source and the usual dependency.
ExamplesThe North American SynchroPhasor Initiative, which coordinated deployment and standards after the 2003 blackout; the 2016 Blue Cut fire event in California, whose inverter behavior was reconstructed from synchrophasor data; oscillation detection tools now running in several US control rooms.
VideosTime Synchronization in the Electric Power System (NASPI and PNNL)
A conventional substation wires every current transformer, voltage transformer, and breaker back to a relay house through copper, one circuit per signal. A digital substation replaces most of that with Ethernet. IEC 61850 defines both a data model, so a breaker means the same thing to equipment from different vendors, and the protocols that carry it: GOOSE messages for fast trip and interlock signals, sampled values for digitized current and voltage. Merging units at the switchyard do the analog-to-digital conversion, and everything after that is a network.
Strengths & weaknessesCopper reduction is the visible win: a large substation can drop tens of kilometers of control cable, which cuts material, trenching, and commissioning time. Because devices describe themselves, configuration is generated from a standard file rather than from drawings, and testing can be automated. The costs are skills and lifecycle. Protection engineers now need network engineering, timing has to be distributed accurately enough for sampled values, and a cyber path exists where copper had none. Interoperability is better than it was, but "IEC 61850 compliant" still hides differences that show up during integration.
When to useChoose a digital architecture for greenfield substations, especially large ones, offshore platforms, and anywhere cable routing is expensive. Retrofit selectively: replacing relays with 61850-capable units while keeping copper to the yard captures much of the configuration benefit at a fraction of the disruption. Keep conventional wiring where the substation is small, the crew is not trained for it, or the site will be rebuilt within a decade. Whatever the choice, insist on a multi-vendor interoperability test before commitment, since that is where schedule risk lives.
Key numbersGOOSE messages delivered in under 4 ms for protection-class signals · sampled values commonly at 80 or 256 samples per cycle · tens of kilometers of control cable avoided in a large substation · timing accuracy of about one microsecond needed for sampled values · commissioning time typically cut by weeks through automated configuration.
ExamplesOffshore wind platform substations, where weight and cable savings decide the design; national digital substation programs in China and India; utility retrofits that adopt station-bus GOOSE messaging while leaving process-bus sampled values for later.
VideosStability & grid services
inertia, reactive power, and fast response5 technologiesA synchronous condenser is a large synchronous machine spinning on the grid with no prime mover and no load. Because it is a real rotating mass magnetically locked to system frequency, it supplies three things that power electronics do not naturally provide: inertia, which resists frequency change during the first seconds after a loss of generation; short-circuit current, which relays need in order to see a fault; and continuously variable reactive power, set by field excitation. Many installations are retired generators with the turbine disconnected, which is the cheapest way to get one.
Strengths & weaknessesIt is the only device on this list that supplies genuine inertia and strong fault current, and it is a hundred-year-old technology with no software behind it. Reactive output is smooth and, unlike a capacitor, does not collapse when voltage sags. The costs are mechanical: it is a spinning machine with bearings, a cooling system, and an excitation system that all need maintenance, and it consumes roughly 1–2% of its rating continuously just to keep turning. Response to a step is fast but not instant, on the order of a hundred milliseconds against a STATCOM's few milliseconds.
When to useAdd synchronous condensers where a network has retired most of its synchronous generation and now has low inertia and weak short-circuit strength, which is exactly what happened in South Australia and Ireland. They are the standard remedy when converters and relays start misbehaving because the system is too weak. Converting a retiring coal or gas unit is the cheapest route and keeps the interconnection. Do not use them where the need is purely fast reactive support, which a STATCOM does better and cheaper, and be aware that grid-forming inverters now provide part of the same service without a rotating mass.
Key numbersProvides real inertia, typically 2–6 MW-seconds per MVA of rating · short-circuit contribution of about 5–6 times rated current, similar to a generator · standing losses roughly 1–2% of rating · response to a voltage step in about 100 ms · converting a retired generator costs far less than a new machine and reuses the interconnection.
ExamplesSouth Australia's four synchronous condensers with added flywheels, installed after the 2016 system black event; EirGrid's fleet in Ireland, driven by very high wind share; several UK stability contracts that pay for inertia and short-circuit level; retired coal units in the US converted rather than demolished.
VideosBoth devices supply or absorb reactive power on command, and the difference is how. A static VAr compensator switches thyristor-controlled reactors and thyristor-switched capacitors, so it is a fast, continuously adjustable version of a capacitor bank, and like a capacitor its output falls with the square of voltage. A STATCOM is a voltage-source converter, the same hardware family as VSC HVDC, that synthesizes a controlled AC voltage behind a reactance. Because it is a current source, it holds its reactive current down to very low system voltage, which is where the two part company.
Strengths & weaknessesSTATCOMs respond in a few milliseconds, keep full output during a deep voltage dip, take a small footprint, and can add short-term overload capability. SVCs cost less per MVAr, use rugged thyristors, and have a long operating record. The weaknesses mirror those strengths. An SVC's output collapses exactly when the system is in trouble, so it needs oversizing for a voltage-support duty. A STATCOM costs more per MVAr, has converter losses of around 1%, and depends on control software whose interaction with nearby inverters has caused real oscillation events.
When to useUse a STATCOM where fast dynamic support during and after a fault decides the outcome: weak grids, large inverter-based plants, arc furnace and traction loads, and points where voltage collapse is the limiting contingency. Use an SVC where the duty is slower voltage regulation at lower cost and the system stays reasonably strong. In both cases put the bulk MVArs on mechanically switched banks and buy only the dynamic increment from power electronics, since a device that is ten times the cost per MVAr should not be doing steady-state work.
Key numbersSTATCOM response in about 2–5 ms; SVC in roughly one to two cycles · STATCOM holds rated reactive current down to very low voltage, while SVC output falls with voltage squared · converter losses around 1% for a STATCOM · cost roughly ten times a mechanically switched capacitor bank per MVAr · typical ratings 50–400 MVAr.
ExamplesSTATCOMs installed across the UK and Australian networks specifically to firm up weak grids around wind and solar; SVCs on long-standing transmission corridors and at arc furnace sites; converter-based reactive support built into offshore wind export systems.
VideosA grid-following inverter measures the grid's voltage waveform with a phase-locked loop and injects current in step with it. That works only if something else is setting the waveform, which is why a system made entirely of grid-following inverters cannot stand up. A grid-forming inverter instead behaves as a voltage source behind an impedance: it sets its own voltage magnitude and angle and lets current fall out of the physics, exactly as a synchronous machine does. Control laws vary, from droop to virtual synchronous machine to virtual oscillator, but the defining property is the same.
Strengths & weaknessesGrid-forming control gives fast frequency response, can supply synthetic inertia, supports black start, and lets a battery or a wind farm operate in a network with no synchronous generation at all. It is a software change on hardware that already exists, so the marginal cost is small. The limits are current and confidence. An inverter can only supply what its semiconductors survive, typically 1.1–1.5 times rated current, so it cannot deliver the fault current a synchronous machine does, which is a protection problem rather than a control one. Standards are young, and interoperability between vendors' control laws at high penetration is an area with more simulation than field experience.
When to useSpecify grid-forming capability on new battery storage and, where available, on wind and solar inverters, especially in systems already low on inertia. It is now a requirement in Australia, Great Britain, and several other markets, and the incremental cost at procurement is much lower than retrofitting later. It does not remove the need for short-circuit strength, so on a weak network expect to pair it with synchronous condensers rather than substitute for them. Treat vendor claims carefully: "grid-forming" covers several different control laws with different behavior during a fault.
Key numbersCurrent limit typically 1.1–1.5 times rating, against 5–6 times for a synchronous machine · inertial response in tens of milliseconds, faster than a real machine's · marginal cost at procurement is small compared with retrofit · mandated for new storage in Australia's National Electricity Market and in Great Britain's grid code · high-penetration multi-vendor behavior is mostly established in simulation.
ExamplesAustralia's Hornsdale and Riverina batteries running in grid-forming mode; the Dalrymple North battery, which has islanded and restarted part of the South Australian network; Kauai's high-solar island system in Hawaii; the DOE and industry specification effort that produced a common definition of the required behavior.
VideosSpecifications for Grid-forming Inverter-based Resources Version 1 (US Department of Energy)
A congested line is usually congested for a few hundred hours a year, and only under contingency. Storage placed at the downstream end can absorb that: when a line trips, the battery discharges within milliseconds and holds the system inside limits until generation redispatches, so the line's transfer limit can be raised without new steel. The industry calls it a virtual transmission line or storage as a transmission asset. It is a regulatory construct as much as an engineering one, because it means a battery is planned, paid for, and rate-based like a wire rather than bid into a market.
Strengths & weaknessesThe batteries can be sited and energized in one to two years against a decade for a line, they need no new corridor, and they solve contingency-driven limits directly. The same asset can often also provide inertia-like fast response and reactive support. The difficulties are regulatory and duty-cycle. If a battery is rate-based as transmission it usually cannot also sell into energy markets, which wastes most of its capability, and the dual-participation models that would fix this remain largely unresolved. Storage also cannot substitute for a line that is congested continuously, since it discharges and then needs to recharge.
When to useConsider it where a specific contingency, rather than steady-state flow, sets a transfer limit, and where the constraint is expected to persist for a decade or less. It is also useful as a bridge: install storage now, build the line later, and redeploy. Do not propose it where flows are congested around the clock, and do not assume the cost recovery is settled, because in most jurisdictions it is not. Check FERC or the local regulator's current position before writing it into a plan.
Key numbersDeployment in 12–24 months against 7–12 years for a new line · discharge response in milliseconds, fast enough for a contingency · effective only where the constraint is intermittent rather than continuous · dual participation as both transmission and market asset remains unresolved in most US markets · ISO New England received FERC approval in 2023 for storage as a transmission-only asset.
ExamplesAustralia's Waratah Super Battery in New South Wales, explicitly procured to raise transfer limits on existing lines; the Escondido and Pomona batteries installed after the Aliso Canyon gas leak; ISO New England's transmission-only storage framework; several MISO and CAISO proceedings on cost recovery.
VideosEnabling Principles for Dual Participation by Energy Storage as a Transmission and Market Asset (PNNL) · Energy storage underused as transmission asset amid unresolved questions, experts say (Utility Dive)
Demand response pays load to move or stop rather than paying generation to start. Programs range from industrial customers on interruptible tariffs, through aggregated residential thermostats and water heaters, to fast frequency response from data centers and electrolyzers. The service is the same in each case: reduce net demand on request, within a stated notice period, for a stated duration, a stated number of times a year. About 33 GW participates in US wholesale markets, roughly 6.5% of combined peak demand.
Strengths & weaknessesIt is the cheapest capacity available, needs no permit, and can be contracted in months. Load can respond in seconds, which is faster than a gas turbine can start, and aggregating many small devices smooths the delivery. The weaknesses are measurement and fatigue. Payment rests on a counterfactual baseline of what the customer would have used, which is estimable but manipulable, and every program spends effort on baseline methodology. Residential participation decays as people override, and industrial curtailment competes with production, so a program that calls too often loses its participants.
When to useUse demand response first when the need is a modest amount of capacity for a small number of hours a year, which describes most peak-driven constraints. It is also increasingly the price of interconnection for very large flexible loads: a data center that accepts curtailment during a handful of tight hours can connect years earlier than one that demands firm service. Do not plan around it for a sustained multi-day event, where participation falls off, and do not count it as firm without a baseline methodology and penalties that make it firm.
Key numbersAbout 33 GW enrolled in US wholesale markets, roughly 6.5% of combined RTO and ISO peak demand · response times from seconds to hours depending on program · typical event limits of 10–20 calls per year with 2–6 hour durations · cost far below new peaking capacity per kW-year · payment rests on an estimated counterfactual baseline.
ExamplesPJM's emergency and economic demand response programs, the largest in North America; ERCOT's load resources providing responsive reserve; UK Demand Flexibility Service payments to households; data center and crypto load curtailment agreements in Texas, and the flexible-load interconnection proposals that grew out of them.
Videos2024 Assessment of Demand Response and Advanced Metering (FERC)
Distribution & edge
medium and low voltage out to the meter5 technologiesThe distribution transformer is the last transformation before the customer, stepping a medium-voltage feeder at 4–35 kV down to 120/240 V or 400 V. In North America it is usually a pole-mounted can serving a handful of houses; elsewhere it is a pad-mounted or vault unit serving a larger group. Ratings run from 10 kVA to a few MVA. There are millions of them, they are built to a small number of standard designs, and until recently they were the most boring component in the system.
Strengths & weaknessesThey are cheap, standardized, and last 30–40 years with essentially no maintenance. Efficiency is high, though no-load losses run continuously for decades and are what efficiency standards target. The problems now are supply and sizing. Lead times moved from weeks to a year or more, and Wood Mackenzie modeled a 10% supply deficit for distribution units in 2025. Sizing assumptions are also breaking: units were specified for the diversified load of houses with a stove and an air conditioner, and a street where several neighbors add an EV charger and a heat pump can overload one that was correctly sized a decade ago.
When to useThere is no alternative for the last step down, so the decisions are sizing, monitoring, and procurement. Size for the electrified load rather than historical diversity, because replacing a failed unit takes longer than it used to. Monitor loading where electrification is concentrated, since transformer failure is now a common early symptom of clustered EV adoption. Keep a standardized inventory so a spare fits many sites, and expect to hold more spares than historical practice suggested.
Key numbersRatings from about 10 kVA to a few MVA · service life typically 30–40 years · no-load losses run continuously and are the target of efficiency rules · lead times of about a year, against weeks before 2021 · Wood Mackenzie modeled a 10% distribution transformer supply deficit in 2025 · a single fast EV charger can add half the diversified load of the houses a small unit serves.
ExamplesThe pole-mounted cans on nearly every North American residential street; pad-mounted units in underground developments; utility replacement programs triggered by EV clusters in California and the UK; the US efficiency standards that reshaped core material choice.
VideosVoltage falls along a loaded feeder and has to stay inside a legal band, typically 114–126 V on a 120 V nominal service in North America. Utilities hold it there with three tools: load tap changers on the substation transformer, step voltage regulators partway down long feeders, and switched capacitor banks that raise voltage by supplying reactive power. All three were designed for one-way flow from substation to customer. Rooftop solar reverses that on sunny afternoons, and voltage can now be highest at the far end of the feeder rather than the near end.
Strengths & weaknessesThe conventional kit is cheap, robust, and universally understood, and tap changers have decades of operating life. The weakness is that mechanical devices operate in discrete steps and wear out with use: a regulator rated for a few hundred thousand operations will not survive following a cloudy day's solar output minute by minute. Smart inverters solve that from the other end, absorbing or supplying reactive power continuously in response to local voltage with no moving parts, but their settings interact, and a feeder where every inverter fights the tap changer is a real and documented failure mode.
When to useUse tap changers and regulators as the backbone everywhere, and add capacitors for the reactive component of load. Turn on smart inverter volt-VAr functions once distributed generation is high enough to cause reverse flow, and coordinate the settings centrally rather than device by device. Where a feeder is long and rural, a regulator remains the practical answer. Where the problem is a dense cluster of rooftop solar in a suburb, inverter-based control plus feeder monitoring usually beats adding another mechanical device that will now cycle far more often.
Key numbersService voltage held to roughly 114–126 V on a 120 V nominal system · step regulators typically give plus or minus 10% in 32 steps · mechanical regulators rated for a few hundred thousand operations over a 30-year life · smart inverter volt-VAr response is continuous and has no wear mechanism · reverse flow from rooftop solar can raise end-of-feeder voltage above the substation's.
ExamplesStep voltage regulators on long rural feeders across North America; Hawaii's rules requiring advanced inverter functions after high rooftop solar caused voltage excursions; IEEE 1547-2018, which made volt-VAr capability a default expectation for new distributed generation; California Rule 21.
VideosAn advanced meter records consumption at 15-minute to hourly intervals and reports it over a communications network, usually a utility-owned mesh radio system or cellular. That is the visible function. The more useful ones are diagnostic: the meter reports voltage, detects and reports its own outage, and confirms restoration, so a utility learns which customers are out from the meters rather than from phone calls. About 128 million advanced meters were installed in the United States as of 2023, covering most residential customers.
Strengths & weaknessesInterval data makes time-varying rates possible, outage reporting shortens restoration, voltage data lets a utility see where a feeder is out of band, and remote connect and disconnect removes truck rolls. The costs are the network and the data. A full deployment is a large capital program, the head-end and meter data management systems are substantial software projects, and the analytics that justify the business case often lag the meters by years. Privacy is a real concern, since interval data reveals occupancy patterns, and communications security has to be maintained across millions of endpoints for 15 years.
When to useThe deployment decision is mostly made: most US and European utilities have rolled out or are rolling out. The live questions are what to do with it. Use meter voltage data to target where regulation is failing and where transformers are overloaded, which is the fastest-payback analytic. Use interval data to enable time-of-use and critical-peak pricing, which is what actually moves load. Where a utility is still choosing, weigh a mesh network against cellular on 15-year total cost rather than on installation price, since the communications choice is harder to change than the meters.
Key numbersInterval recording at 15 minutes to one hour · about 128.4 million advanced meters in the US as of 2023, up 9.1 million in one year · covers roughly three quarters of US electricity customers · meter life typically 15–20 years · outage detection at the meter is often faster than customer calls.
ExamplesFull-territory rollouts by most US investor-owned utilities; Great Britain's national smart meter program; Italy's Enel deployment, the first at national scale; rural cooperatives using AMI voltage data to find failing transformers before they fail.
VideosHow many smart meters are installed in the United States, and who has them? (US Energy Information Administration) · 2024 Assessment of Demand Response and Advanced Metering (FERC)
A microgrid is a defined section of distribution system with its own generation, its own controller, and a switch at the boundary that lets it separate from the utility and keep running. The distinguishing feature is that switch and the control that goes with it: when the grid fails, the microgrid detects it, opens, rebalances its own generation against its own load, and holds frequency and voltage on its own. When the grid returns, it resynchronizes and closes back in. Generation is typically solar, batteries, and a generator or fuel cell.
Strengths & weaknessesResilience is the product, and for a hospital, a base, or a wastewater plant that is worth real money. A microgrid can also earn between outages by shaving peaks, avoiding demand charges, and bidding into markets, which is what turns it from an expensive backup system into an investment. The costs are control and interconnection. Islanding, resynchronization, and protection all have to work with distributed sources whose fault current is small, and the utility interconnection agreement is often the longest part of the project. Per kW, a microgrid costs considerably more than grid supply.
When to useBuild one where an outage causes losses far larger than the premium: hospitals, data centers, military installations, water and wastewater, and remote communities where the alternative is diesel at $0.30–1.00/kWh. Also consider it where interconnection for new load will take years and on-site generation can bridge the gap. Do not build one for cost savings alone on a reliable grid, since the numbers rarely work. And design the protection scheme early, because on most projects it is the thing that delays commissioning.
Key numbersTypical size 100 kW to tens of MW · islanding transition in under a second for a seamless design · remote diesel-displacement cases compete against $0.30–1.00/kWh · IEEE 1547 and IEEE 2030.7 define the interconnection and controller requirements · payback normally depends on stacking resilience value with demand charge and market revenue.
ExamplesThe Borrego Springs microgrid in California, one of the earliest utility-scale examples; Alaskan and Canadian remote community systems displacing diesel; US military installation microgrids driven by mission assurance; hospital campuses that islanded successfully during hurricanes in Texas and Florida.
VideosA non-wires alternative meets a grid need with something other than new grid equipment. The classic case is a substation forecast to overload in three years for forty hours a year: instead of rebuilding it, the utility contracts batteries, demand response, targeted efficiency, and local solar to shave those hours. Making that work at scale needs a distributed energy resource management system, which forecasts what the resources will do, dispatches them against a distribution model, and verifies delivery. The technology is mostly software; the hard part is procurement and cost recovery.
Strengths & weaknessesDeferring a rebuild for a few hundred hours a year of relief is often much cheaper than the rebuild, and the resources arrive in months rather than years. Contracts are modular, so the utility can buy more if load grows faster and stop if it does not, which is real option value against an uncertain forecast. The weaknesses are performance risk and regulation. A wire is available 99.99% of the time and a portfolio of customer-owned devices is not, so contracts need penalties and headroom. Utilities also earn a return on capital rather than on contracts, so the regulatory model has to be adjusted or the incentive points the wrong way.
When to useUse a non-wires approach where the driver is a small number of hours, where load growth is uncertain enough that deferral has option value, and where enough customer-side resource exists in the right electrical location. Location matters: relief must be downstream of the constraint. Do not use it where the need is continuous capacity or where reliability standards demand firm deliverability, and be honest that it defers rather than eliminates the investment. Where load growth is fast and certain, build the wire.
Key numbersTypically targets 50–200 hours of constraint per year · deferral value depends on the utility's cost of capital and the length of the deferral · contracted resources deliver at 90–99% availability against a wire's 99.99% · deployment in 6–24 months · resources have to sit downstream of the constraint to count.
ExamplesCon Edison's Brooklyn-Queens Demand Management program, the best-documented US deferral case; National Grid's UK flexibility tenders; California's distribution deferral framework; the growing set of utilities running DERMS platforms to dispatch and settle these portfolios.
VideosBQDM program demonstrates benefits of non-traditional utility investments (Utility Dive)
No technologies match the current filters.
Try clearing a facet or broadening the search.
Glossary
Terms that show up in the technology explorer and are not obvious from outside the field. Numbers are typical values, not specifications.
| Term | What it means |
|---|---|
| Ampacity | The current a conductor can carry continuously without exceeding its temperature limit. It is set by heat balance rather than by the metal alone, so wind and ambient temperature change it by tens of percent through the day. |
| Ancillary services | The services that keep a grid stable rather than deliver energy: frequency regulation, reserves, reactive support, and black start. They are procured separately from energy, and a technology that cannot bid into them gives up the highest-value revenue on offer. |
| Black start | Restarting a dead grid from a unit that can start with no outside power. Traditionally hydro and diesel units; grid-forming inverters and VSC HVDC converters can now do it, which is one of the reasons they are specified. |
| Charging current | The current a cable's own capacitance draws even with no load at the far end. It scales with length, so above roughly 50–80 km an AC cable spends its entire rating charging itself, which is why long cables are DC. |
| Commutation | Handing current from one valve to the next in a converter. A line-commutated converter relies on the AC system's own voltage reversal to do it, so a voltage dip at the receiving end can cause a commutation failure and briefly interrupt power transfer. |
| Congestion | A condition where the cheapest available generation cannot reach the load because a transmission element is at its limit, so more expensive local generation runs instead. The cost difference is what a congestion charge measures and what most upgrades are justified against. |
| Contingency | A credible single failure the system must survive without cascading, usually the loss of any one line, transformer, or generator. Transfer limits are set by the worst contingency, not by normal conditions, which is why a line often runs well below its thermal rating. |
| Corona | Partial ionization of air at the conductor surface where the electric field is highest. It causes audible noise, radio interference, and a small continuous loss, and it is the reason EHV lines use bundled conductors: several smaller wires per phase lower the surface field. |
| DERMS | A distributed energy resource management system: the software a utility uses to forecast, dispatch, and settle customer-side batteries, generators, and flexible load against a model of the distribution network. It is what makes a non-wires alternative operable rather than merely contracted. |
| Distance protection | A relay scheme that computes apparent impedance from measured voltage and current. A fault looks like a much lower impedance than load, and how low indicates roughly how far away it is, so the relay can trip instantly for close faults and wait for distant ones. |
| Dynamic line rating | Setting a line's rating from actual weather rather than from a conservative fixed assumption. Because wind cools a conductor strongly and blows hardest when wind generation is high, the extra capacity tends to appear when the system wants it. |
| Fault current | The current that flows into a short circuit, which relays use to detect and locate faults. A synchronous machine supplies five to six times its rated current; an inverter supplies about 1.2 times under software control. That gap is why protection schemes are being revisited. |
| GOOSE | Generic Object Oriented Substation Event, the IEC 61850 message type used for fast trip and interlock signals between substation devices. It replaces a hardwired contact with an Ethernet multicast delivered in under 4 ms. |
| Grid-forming | An inverter control mode in which the device sets its own voltage magnitude and angle and lets current follow, the way a synchronous machine behaves. The alternative, grid-following, needs an existing waveform to lock onto and cannot hold up a system on its own. |
| HVDC | High-voltage direct current transmission. Converting to DC removes reactance and charging-current limits and gives direct control of power flow, at the cost of a converter station at each end. It becomes economic past roughly 600–800 km overhead and 40–70 km by cable. |
| Inertia | Rotational energy stored in generators spinning in step with the grid, which resists frequency change in the first seconds after a loss of generation. Inverters supply none inherently, so systems with high inverter share see frequency fall faster and need faster response. |
| Interconnection queue | The sequence in which proposed generators are studied for their effect on the network and assigned the cost of resulting upgrades. Queues now run several years in most US regions, and position in the queue often matters more to a project's timing than construction does. |
| Islanding | Operating a section of network separated from the main grid, supplied by its own generation. Intentional islanding is what a microgrid does during an outage; unintentional islanding is a hazard, because line crews may work on a circuit still energized by a customer's generator. |
| Load tap changer | A mechanism that changes a transformer's turns ratio while it is energized, holding downstream voltage as load varies. It is the primary voltage control on most distribution substations, and it wears with each operation, which matters when solar makes it operate far more often. |
| Loop flow | Power taking an unscheduled path through neighboring systems, because AC flow follows impedance rather than contracts. It congests lines belonging to parties who were not part of the transaction, and controlling it is one of the arguments for DC links inside a synchronous grid. |
| Reactive power | The component of AC power that oscillates between source and load without doing net work, measured in volt-amperes reactive. It is what holds voltage up, it does not travel far, and it has to be supplied close to where it is consumed. |
| Non-wires alternative | Meeting a grid need with contracted storage, demand response, or efficiency instead of new equipment. It works when the constraint binds for a limited number of hours and the resources sit downstream of it. |
| Power flow | The distribution of real and reactive power across a network, computed from generation, load, and impedances. In an AC system nobody chooses it directly, which is the underlying reason congestion appears where no one scheduled it. |
| Reconductoring | Replacing the conductor on an existing line while keeping the structures and the right of way. With advanced conductors it can double the line's rating in 18–36 months, which is why it has become a capacity strategy rather than maintenance. |
| Right of way | The strip of land a utility has the legal right to build and maintain a line on, typically 45–60 m wide at EHV. Acquiring a new one is the longest and least predictable part of a transmission project, and it is why reusing an existing corridor is worth a large premium. |
| Sag | How far a conductor hangs below its attachment points, which grows as the conductor heats and lengthens. Clearance to the ground under maximum sag, not the metal's current-carrying ability, is usually what caps a line's rating. |
| SCADA | Supervisory control and data acquisition: the telemetry layer that reports substation measurements and breaker states every two to four seconds and carries control commands back. Everything a control room computes rests on it. |
| Series compensation | Capacitors placed in series with a long line to cancel part of its inductive reactance, which makes the line electrically shorter and raises the power it can carry stably. It needs a subsynchronous resonance study wherever steam turbines are nearby. |
| SF6 | Sulfur hexafluoride, an excellent insulating and arc-quenching gas with a global warming potential about 23,500 times CO2. It made compact switchgear possible and is now being regulated out, with fluoronitrile blends and vacuum interruption as the replacements. |
| State estimation | Reconciling many redundant and slightly inconsistent measurements into one consistent picture of the network. Every contingency analysis and market calculation runs on its output, so a stale network model silently corrupts all of them. |
| Subsynchronous resonance | An exchange of energy between a series-compensated line and the torsional modes of a nearby turbine-generator shaft, at a frequency below the grid's. It damaged shafts at the Mohave plant in 1970 and 1971, and it is why series capacitors require a dedicated study. |
| Synchronous condenser | A synchronous machine spinning on the grid with no prime mover, supplying inertia, fault current, and adjustable reactive power. Converting a retiring generator is usually the cheapest way to get one, and it keeps the existing interconnection. |
| Synchrophasor | A voltage or current phasor measured 30 to 120 times a second and time-stamped from GPS to about a microsecond. Because every unit shares one clock, angles measured hundreds of kilometers apart can be compared directly, which makes oscillations visible. |
| System strength | How firmly the voltage waveform is held at a point in the network, in practice measured by the available fault current there. Weak points cause inverter controls to interact badly and protection to misjudge faults, and it is the constraint that limits new connections in high-renewable regions. |
| Thermal, stability, and voltage limits | The three different things that cap a line's transfer. Thermal means the conductor sags too far; stability means the system loses synchronism or damping first; voltage means the receiving end collapses. Each has a different fix, and using the wrong one buys nothing. |
| XLPE | Cross-linked polyethylene, the standard insulation for modern power cable. Cross-linking turns a thermoplastic into a thermoset that holds its shape at temperature, which is what allows a 90 °C conductor rating in a buried cable. |
How to choose a grid technology
Almost every grid decision is really a question about time, not engineering. A new 500 kV line is understood technology; what makes it a ten-year project is right of way, permits, and a transformer order. So the first question on any grid problem is not "what equipment solves this" but "what is the fastest thing that solves this inside a corridor I already have." That question orders this entire sheet.
The ladder, cheapest and fastest first
Given a constraint, work down this list and stop at the first rung that actually fixes it. Each step up costs roughly an order of magnitude more and takes several times longer than the one before it.
What is actually limiting the line
Three different limits get called "capacity" and they have completely different fixes. A thermal limit means the conductor sags into its clearance envelope; dynamic ratings and reconductoring help, series capacitors do nothing. A stability limit means the line cannot carry more without losing synchronism or damping; series compensation and fast controls help, a fatter conductor does nothing. A voltage limit means the receiving end collapses under load; reactive support helps, and so does anything that moves generation closer. Ask which one binds before choosing anything, because half of the wasted money in grid planning comes from fixing the wrong limit.
Engineering factors
| Factor | Why it matters |
|---|---|
| Voltage class | Power scales with voltage and losses with current squared, so going up a class carries several times the power on similar conductors. It also multiplies clearances, insulation, and equipment cost. |
| AC vs DC | DC has no reactance limit and no charging current, so it wins past roughly 600–800 km overhead and 40–70 km by cable. Below that, converter cost dominates and AC wins. |
| Short-circuit strength | Relays and converters both need fault current to work correctly. As synchronous machines retire, weak-grid behavior becomes the binding constraint on connecting more inverters. |
| Inertia | System inertia sets how fast frequency falls after a loss of generation. Inverters supply none unless they are grid-forming, which is why synchronous condensers came back. |
| Reactive power | It does not travel, so it has to be supplied near the load. Capacitors are cheapest per MVAr and weakest when voltage sags; STATCOMs are the opposite. |
| Fault duty | Adding ties and generation raises available fault current, and a breaker fleet rated 40 kA in 1990 can be under-rated today. Check this before adding a strong tie. |
| Protection and inverters | Distance and differential schemes were designed around machines that supply 5–6 times rated current into a fault. Inverters supply about 1.2 times, under software control, and misoperations follow. |
| Outage windows | Any work on an existing line needs the surrounding system to carry the load without it. On a constrained network there may be no window, which turns a cheap job into an impossible one. |
Economic and schedule factors
| Factor | Why it matters |
|---|---|
| Equipment lead time | Large power transformers run 80–144 weeks and HVDC converters three to five years. Order before design is final if the energization date is firm. |
| Right of way | The single largest schedule risk in transmission, and the reason anything that reuses an existing corridor is worth a large premium. |
| Cost allocation | Who pays for a line that benefits several regions is unresolved in most US markets, and it kills more interregional projects than engineering does. |
| Interconnection queue | Generation waits years for studies that mostly determine who pays for network upgrades. The queue, not the turbine, sets when a project connects. |
| Rate base vs contract | Utilities earn a return on capital, not on contracts, so non-wires alternatives face an incentive problem that has to be fixed in regulation rather than engineering. |
| Asset life | Lines last 50–80 years and transformers 40. A decision made now sets the network's shape past 2070, which argues for building the higher voltage class. |
| SF6 phase-down | Insulating gas with 23,500 times the warming potential of CO2 is being regulated out. New switchgear decisions should assume reporting obligations for the asset's whole life. |
Why lead times became the story
Before 2021 a utility could assume equipment would arrive when needed and spent its planning effort on siting. That assumption is gone. Power transformer demand rose 119% from 2019 while grain-oriented electrical steel capacity did not, and standard power transformers now average around 128 weeks with generator step-up units near 144. HVDC converter slots are booked years out, and so are subsea cable factory and vessel slots. The practical consequence is that procurement now belongs at the front of a project rather than after final design, and any schedule that assumes otherwise is optimistic by a year or more.
Core takeaway
Treat the existing corridor as the scarce resource and everything else as negotiable. Rating a line dynamically, compensating it, or reconductoring it delivers capacity in months to three years inside land that is already approved, and a new line delivers more capacity in seven to twelve years if it survives permitting at all. Build new lines where the need is large and durable, and exhaust the corridor you already own first, because that is where the time is.
Key questions for engineering decisions
- Is the binding limit thermal, stability, or voltage? Each has a different fix and they are not interchangeable.
- How many hours a year is the constraint actually binding, and does that favor storage or demand response over steel?
- What is the short-circuit strength at this point, and will it still support protection after the next retirement?
- Can the work be done inside an existing right of way, and is there an outage window to do it?
- What are the fault duties after this addition, and do the existing breakers still cover them?
Key questions for investment and business analysis
- What is the equipment lead time, and has a slot been reserved or only quoted?
- Which step in the schedule is longest: permitting, procurement, or construction? Optimize that one.
- Who pays, and has cost allocation actually been settled rather than assumed?
- If the load forecast is wrong by half, does the investment still make sense, or is a modular option worth its premium?
- Does the revenue model match how the asset is regulated, particularly for storage and non-wires portfolios?
Head-to-head: how do you get more capacity on this path
This is the choice a planner faces when a corridor is constrained. The options are ordered the way they should be evaluated, cheapest and fastest first. The tables after it cover the AC-versus-DC decision, what to put in a substation, and how to hold a low-inertia system together.
| Option | Capacity gain | Time | Siting | Pick it when |
|---|---|---|---|---|
| Dynamic line rating | 10–30%, more in wind | Months | None needed | The line is thermally limited for a limited number of hours. Cheapest capacity available, and it stacks with everything below. |
| Series compensation | 40–70% on a long line | 1–2 years | Inside the fence | The limit is stability rather than heat. Run the subsynchronous resonance study before committing. |
| Storage at the constraint | Relieves contingency limits | 1–2 years | Existing site | A specific contingency sets the limit and the binding hours are few. Check how your regulator treats cost recovery. |
| Reconductoring | Up to 2x | 18–36 months | Existing right of way | The line is thermally limited, the structures are sound, and an outage window exists. Usually the best large increment available. |
| Rebuild at higher voltage | 2–4x | 4–7 years | Existing corridor, widened | The corridor will be needed for decades and the load growth is certain. Build the higher class now rather than twice. |
| New AC line | New path | 7–12 years | New right of way | A genuinely new path is needed. Start the land work first and treat everything else as following it. |
| New HVDC link | New controllable path | 7–12 years | New right of way or subsea | The route is long, crosses water, joins asynchronous systems, or the flow has to be controlled rather than left to impedance. |
AC or DC, and which converter
Distance and the medium decide AC against DC. After that, the converter choice is decided by how strong the AC systems at each end are and whether the scheme will ever have more than two terminals.
| Option | Economic range | Losses | Needs a strong AC grid | Pick it when |
|---|---|---|---|---|
| Overhead AC | Under about 600–800 km | 2–4% per 1,000 km at 765 kV | n/a | Almost always, on land, at moderate distance. Cheapest per MW-mile and repairable in days. |
| AC cable | Under 50–80 km | Low, but charging current eats capacity | n/a | An overhead line cannot be built: city, airport, protected landscape. Expect 5–15x the cost. |
| LCC HVDC | Long overhead, 1,000 km+ | 0.7% per station | Yes, at both ends | The largest point-to-point transfers into strong systems. 800–1,100 kV moves 8–12 GW at the lowest loss available. |
| VSC HVDC | Any distance where DC wins | About 1% per station | No | Offshore wind, weak grids, city infeeds, and anything that may become multi-terminal. The default for new schemes. |
| Back-to-back converter | Zero distance | 1–2% total | No | The two systems cannot be synchronized, or an internal seam needs a controllable valve rather than an impedance path. |
| Subsea DC cable | Up to 700+ km | Low, plus converters | No | The route crosses water at all. Book cable factory and vessel slots before finishing the design. |
What goes in the substation
Reactive support, switching, and protection are separate purchases that get conflated. The rule of thumb: buy bulk MVArs mechanically and buy speed electronically, and never let a device that costs ten times more per MVAr do steady-state work.
| Equipment | Speed | Relative cost | Holds up in a voltage dip | Pick it when |
|---|---|---|---|---|
| Switched capacitor bank | Seconds, stepwise | 1x per MVAr | No, output falls with voltage squared | Steady-state reactive support. Always the base layer, everywhere. |
| Shunt reactor | Seconds, stepwise | 1x per MVAr | n/a | Long lightly loaded lines and any EHV cable circuit, where the line's own capacitance pushes voltage up. |
| SVC | One to two cycles | 5–7x | Partly | Continuous regulation at lower cost than a STATCOM, on a system that stays reasonably strong. |
| STATCOM | 2–5 ms | 10x | Yes, holds rated current | Fast dynamic support in a weak grid or next to a large inverter plant. Buy only the dynamic increment. |
| Synchronous condenser | About 100 ms | 8–12x, less if converted | Yes | You need inertia and short-circuit current, not just MVArs. Converting a retiring generator is the cheap route. |
| Gas-insulated switchgear | n/a | 2–4x air-insulated | n/a | Land is unavailable or the air is corrosive. Specify an SF6-free option unless there is a reason not to. |
Holding up a system with few synchronous machines
As thermal plant retires, three things go with it: inertia, short-circuit strength, and voltage control. They are separate services and no single device supplies all three well.
| Option | Inertia | Fault current | Speed | Pick it when |
|---|---|---|---|---|
| Synchronous condenser | Real, 2–6 MW-s per MVA | 5–6x rated | About 100 ms | Protection is misoperating and converters are unstable because the system is weak. The only device that fixes short-circuit strength. |
| Grid-forming inverter | Synthetic, very fast | 1.1–1.5x rated | Tens of ms | New batteries and inverters are being procured anyway. Cheap at procurement, expensive to retrofit. |
| STATCOM | None | 1.1–1.5x rated | 2–5 ms | The problem is voltage, not frequency. Pairs well with a synchronous condenser rather than replacing it. |
| Storage on fast frequency response | Effective, not physical | Low | Under 1 s full output | Frequency response is the shortfall and a market exists to pay for it. |
| Demand response | None | None | Seconds to hours | The shortfall is capacity for a few hundred hours a year. Cheapest option on this sheet, with a baseline you have to police. |