Transmission Line Power

The problem: moving 1 GW across 1000 km

The thermal generator burning coal in Wyoming or the wind farm in west Texas is rarely where the load is. Power has to be moved — sometimes a few kilometres to a city across the river, sometimes a thousand kilometres to a coastal load centre. The engineering question is straightforward in form and brutal in arithmetic: how do you push a gigawatt down a wire so long that even a small fractional loss costs a city's worth of electricity?

The answer that the industry settled on by the 1930s, and that has not changed in principle since, is to step the voltage up so the current is small. The numerical magic that makes this work is the quadratic in Joule's law.

Why high voltage beats fat wires

For a line carrying complex power S = V·I*, the magnitude of the current required to deliver real power P at line voltage V and unity power factor is I = P/V. The power dissipated as heat in the line's resistance R is

P_loss = I² R = (P/V)² R = P² R / V²

The loss scales as 1/V². Double the voltage, quarter the loss. Multiply the voltage by ten, divide the loss by a hundred. A 765 kV line delivers the same megawatts as a 76.5 kV line with one ten-thousandth the resistive loss — for the same conductor R. The economic case for transmission-class voltage is not that it lets you use thinner wires; it is that it lets you use the wires you have already built and waste vastly less energy as heat.

The cost of getting there is the transformers at each end. A 1 GW generator-step-up transformer is a refrigerator-sized device that weighs a hundred tonnes and costs millions of dollars, but it pays for itself in saved megawatt-hours within the first hours of operation. Substations at the receiving end step the voltage back down: 500 kV → 138 kV → 13.8 kV → 240 V at the meter.

Standard AC voltage classes

Class	Voltage (line-to-line)	Typical use	ROW width
Sub-transmission	69 kV, 115 kV, 138 kV	Regional industrial, urban interties	~20 m
High voltage (HV)	230 kV	Regional bulk transfer	~30 m
Extra-high voltage (EHV)	345 kV	Midwest, Northeast bulk	~45 m
EHV	500 kV	West, South, PJM workhorse	~60 m
Ultra-high voltage (UHV)	765 kV	Long AEP corridors	~75 m
HVDC	±250 to ±800 kV	Long links, submarine, async	~50 m typical

Higher class isn't always better. Insulation requirements scale roughly with voltage; tower height grows; right-of-way widths grow; corona losses grow if conductor design doesn't keep up. The class chosen for a given corridor is the one whose conductor + tower + ROW cost minimises the lifecycle cost for the expected MW-km of flow.

ACSR: the workhorse conductor

Almost all overhead transmission conductor in service is ACSR — Aluminum Conductor Steel Reinforced. The cross-section has a galvanised steel core (one, seven, or nineteen strands depending on diameter) surrounded by stranded aluminum, typically 1350-H19 alloy. The steel carries the mechanical tension; the aluminum carries the current.

Aluminum is the conductor of choice because it has 61% of copper's conductivity at one-third the weight and one-quarter the cost. You can build a line with copper, and the railways did exactly that in 1900, but you have to put towers every quarter mile to support the weight. Aluminum buys you longer spans and cheaper towers; the steel core gives it the tensile strength to span 300 m without unacceptable creep over a 50-year service life. Newer variants include ACSS (steel-supported, runs hotter), ACCC (composite core, even higher temperature and lower sag), and AAAC (all-aluminum alloy, used where corrosion of the steel is a concern).

Bundled conductors: the 230 kV threshold

Above about 230 kV, each phase is no longer a single conductor but a bundle of two, three, or four sub-conductors held in geometric position by metal spacers every few metres. The reasons are three interacting effects:

1. Reactance reduction. The geometric mean radius (GMR) of a bundle is much larger than the GMR of a single wire, so the inductance per kilometre L drops, the surge impedance Z_0 = √(L/C) drops, and the surge-impedance loading (and so the steady-state transfer capacity) rises by 20–80% depending on bundle size.
2. Skin effect. At 60 Hz in aluminum the skin depth is δ = √(2ρ/ωμ) ≈ 8.5 mm. A 30 mm-diameter conductor uses only its outer 8.5 mm shell for AC current; the inner aluminum is wasted. Bundling splits the phase into multiple thinner conductors with more outer surface for the same metal cost.
3. Corona discharge. The surface E-field on a thin conductor at 500 kV easily exceeds the 30 kV/cm dielectric strength of air (Peek's criterion). The air ionises, the conductor glows violet, you hear a faint hiss and a 50/60 Hz hum, and you bleed real power as audible noise, EMI, and chemical reactions (ozone, NOx). Bundling makes the conductor look electrically like a much larger diameter, dropping the surface field below the corona threshold.

The typical scheme: 230 kV runs single conductor; 345 kV often runs twin bundle; 500 kV almost always runs triple bundle; 765 kV runs quadruple. The Pacific Intertie HVDC, at ±500 kV, runs two bundled sub-conductors per pole.

Sag, tension, and the thermal limit

A flexible conductor hung between two towers under gravity assumes the catenary curve

y(x) = a · cosh(x / a)        where a = T_H / w

T_H is the horizontal component of tension and w is the conductor weight per unit length. The maximum sag at mid-span, for span L, is approximately wL²/(8T_H) when L ≪ a. Real engineering practice tracks the conductor temperature, and through it the conductor length:

ΔL = L₀ · α · ΔT        α_Al ≈ 23 × 10⁻⁶ /°C

A 300 m span of ACSR heated from 20 °C to 75 °C lengthens by about 38 cm; that extra length goes entirely into additional sag, roughly one extra metre at mid-span. The conductor temperature comes from a heat balance between Joule heating from current, solar gain, and convective cooling to wind:

I²R + α_s · Q_solar = h_c · A · (T_c − T_a) + ε σ A (T_c⁴ − T_sky⁴)

The line's ampacity — the rated current it can carry — is set not by insulation but by the highest current at which the steady-state sag stays clear of ground, vegetation, and nearby structures by the code-mandated margin. NERC TPL-001 and FAC-003 require that minimum clearance to vegetation not be violated at any operating condition. Dynamic line rating (DLR) systems read the actual sag in real time and let operators push more current on cool, windy days when the conductor runs cooler than nameplate.

Surge impedance and reactive power

A transmission line, viewed at the right level of abstraction, is an LC distributed network. Per unit length it has series inductance L (from the magnetic field surrounding the conductor) and shunt capacitance C (between conductors, and to ground). The characteristic impedance — the impedance the line "looks like" to a fast pulse — is

Z_0 = √(L / C)

For overhead lines this is in the range 290–400 Ω depending on bundle geometry. Underground cable, with its grounded coaxial structure, has a much higher C and a much lower Z_0, typically 30–60 Ω. The surge-impedance loading (SIL) of a line is the load at which the line's series inductance absorbs exactly as much reactive power as its shunt capacitance generates:

SIL = V² / Z_0

For a 500 kV line at Z_0 = 300 Ω, SIL is about 833 MW. At loads below SIL the capacitance dominates, the line generates reactive power, and the receiving-end voltage rises above the sending-end voltage (the Ferranti effect — important on lightly loaded long lines). Above SIL the inductance dominates, the line consumes reactive power, and voltage drops along the corridor. Real lines are operated near SIL when possible; deviation is corrected by shunt reactors (high load) or shunt capacitors (low load) at the substations.

Series compensation: cancelling the inductance

The steady-state transfer capability of a long AC line between two stiff buses is approximately P_max = V_s · V_r · sin(δ) / X, where X is the total series reactance. To push more megawatts through an existing corridor without building a new one, utilities install series capacitor banks that partially cancel X, effectively shortening the line electrically. The Pacific Intertie's 500 kV AC parallel paths carry series capacitors at intermediate substations; the canonical example is the Slatt 500 kV substation in Oregon, where TCSC (thyristor-controlled series capacitors) provide dynamic series compensation that can also damp inter-area oscillations.

Series compensation is not free. It introduces sub-synchronous resonance (SSR) risk between the LC circuit of the line and the torsional natural frequencies of nearby steam-turbine shafts; the 1970 Mohave generator shaft failure was traced to SSR. Modern installations use TCSCs and shaft-mode damping to avoid the same trap.

Insulators, towers, and right of way

• Insulators. Strings of porcelain or toughened-glass discs suspended below the crossarm support the conductor and isolate it from the grounded tower. A 500 kV line typically uses 22–25 disc units per insulator string. Modern installations increasingly use polymer (silicone-rubber) long-rod insulators, which are lighter, hydrophobic, and easier to install, at the cost of UV-degradation concerns.
• Tower types. Lattice steel towers — the iconic geometric trusses — are the cheapest at high voltage and the most common. Tubular steel monopoles are used in urban and suburban corridors where ROW is narrow and visual impact matters. H-frame wood or steel structures are used at sub-transmission voltages and in flat rural country where simple, low towers suffice.
• Right of way (ROW). The strip of land kept clear of buildings and tall vegetation along the line. Width scales with voltage: ~20 m at 138 kV, ~30 m at 230 kV, ~45 m at 345 kV, ~60 m at 500 kV, ~75 m at 765 kV. ROW acquisition and vegetation management are the dominant non-equipment costs of a new transmission corridor and the primary source of permitting delays.
• EMF. The 60 Hz electric and magnetic fields under a 500 kV line are typically 3–10 kV/m at ground level and a few microteslas. Decades of epidemiological work have found no consistent biological effect at these levels, but public perception of EMF is a recurring driver of ROW disputes.

HVDC: when the AC playbook stops working

For most of the grid, three-phase AC is the right answer — transformers are simple and lossless to first order, and AC switchgear is mature. But four cases push the choice to HVDC:

1. Very long overhead distance. Above 600–800 km the converter station capital expense is amortised by the elimination of AC reactive-power losses; HVDC becomes cheaper per MW-km. The Pacific DC Intertie (Celilo, OR to Sylmar, CA) is the canonical North American example: ±500 kV, 3.1 GW, 1362 km, in service since 1970.
2. Submarine cable. AC cables have such large capacitance that beyond about 80 km the charging current alone fills the conductor's thermal capacity, leaving nothing for real power. HVDC has no charging current. All long submarine power links — the NorNed (Norway-Netherlands), the IFA2000 (England-France), the Basslink (Tasmania-Victoria) — are HVDC.
3. Asynchronous interconnection. Grids running at different frequencies (60 Hz US vs. 50 Hz Europe) or merely out of phase can only be tied through a DC link. The Eastern and Western US Interconnections, both 60 Hz, are tied at a handful of DC back-to-back substations because synchronising them across the Rockies was deemed impractical.
4. Controllable power flow. HVDC links via voltage-source converters (VSC HVDC) control real and reactive power dispatch independently, on millisecond timescales. They are used to stabilise weak grids and to enforce contractual power flows on meshed AC networks.

Putting the losses in perspective

A well-designed transmission corridor at high voltage runs at about 2–3% loss over 500 km. The full generation-to-meter loss in the US averages 5%, with most of it in distribution (the low-voltage end). The numbers worth carrying:

Stage	Voltage	Typical loss
Generator step-up transformer	20 kV → 500 kV	~0.3%
500 kV transmission (500 km)	500 kV	~2–3%
Bulk substation step-down	500 kV → 138 kV	~0.3%
Sub-transmission	138 kV / 69 kV	~0.5%
Distribution substation	69 kV → 13.8 kV	~0.4%
Distribution feeders	13.8 kV	~2%
Service transformer	13.8 kV → 240 V	~1%
Total typical end-to-end	—	~5%

The grids of North America, Europe, India, China, and the rest together move roughly 5 TW continuously — about 24 PWh per year — almost all of it through this stack. Five percent of 24 PWh is more than a trillion kWh of waste heat per year. That is the cost of the convenience of an instantaneous power source at the wall; cutting it by even a fraction of a percent is worth tens of billions of dollars annually, which is why the engineering of transmission lines is one of the most carefully optimised industrial systems on the planet.

Common pitfalls

• Mistaking voltage class for power. A 765 kV line is not automatically a higher-power line than a 500 kV line. Power equals voltage times current; thermal ampacity sets current. Many 500 kV corridors carry more megawatts than under-loaded 765 kV ones.
• Treating line loss as constant percentage. Loss is quadratic in load (I²R for fixed R), not linear. A line at half its loading has one-quarter the loss as a fraction of the load. Aggregate annual loss therefore depends strongly on the load duration curve, not just on peak.
• Confusing surge impedance with line resistance. Z_0 = √(L/C) is a wave property and has nothing to do with the resistance that dissipates power. A lossless line still has surge impedance; it just doesn't lose anything.
• Forgetting reactive power. A line at full real power may be unable to hold voltage at the receiving end without reactive support. Voltage collapse — the 2003 Northeast blackout was partly such an event — is a reactive-power phenomenon, not a thermal one.
• Underestimating ROW and permitting. A transmission line's engineering can be finalised in two years; the right-of-way and environmental review can take ten. The grid is not bottlenecked on metallurgy; it is bottlenecked on land use.