Electronics Guide

From 33 kV to 66 kV Collection

For the first two decades of commercial offshore wind, collection networks operated at 33 or 34 kilovolts, a voltage inherited from onshore distribution practice. As turbine ratings climbed from 3 megawatts toward 15 megawatts and beyond, that voltage became a bottleneck: a 33 kV string could accommodate only a handful of large machines before the cable reached its current limit, which forced short strings, more cable circuits, and more switchgear bays on the platform. Hornsea Project One, at 1.2 gigawatts one of the largest wind farms in operation, belongs to that generation: it collects at 34 kV.

The Carbon Trust's Offshore Wind Accelerator, a collaborative research program funded by a consortium of developers, identified the shortage of commercially mature 66 kV array cable systems as the barrier and ran a competition to close it, funding the testing and qualification of designs from the manufacturers JDR, Nexans, and Prysmian; commercial products followed from around 2015. Prysmian, announcing its qualified system, put the potential saving at up to 15 percent of wind farm cost.

The arithmetic behind such claims is straightforward. Doubling the voltage doubles the power a given conductor can carry at the same current, which either halves the number of circuits or doubles the number of turbines per string, and fewer circuits mean less cable, fewer terminations, fewer switchgear bays, and less vessel time. A 66 kV three-core cable with a large conductor carries on the order of 800 amperes where the seabed allows it to shed heat, or roughly 90 megavolt-amperes: six 15 megawatt turbines on one string, against three for the same conductor at 33 kV.

Sixty-six kilovolts is now the default for new fixed-bottom projects, and it is the collection voltage assumed in standardized offshore connection designs. TenneT's 2 GW platforms take 66 kV alternating current directly from the wind farms they serve, which removes a whole asset from the chain: the developer no longer builds an intermediate collector substation, and no 155 kV link is needed between it and the transmission operator's converter platform. Attention has since turned to whether a further step, to 132 kV collection, would pay for itself in very large farms with very large turbines; the answer depends on whether the savings in cable count outweigh the cost of higher-voltage turbine transformers, switchgear, and terminations.

String Topology and Redundancy

The simplest and most common array topology is the radial string: turbines connected in series, with the cable section nearest the substation carrying the full string output and each subsequent section carrying less. Radial strings are cheap and simple to protect, but they are single-fault-intolerant. A cable failure in the first section disconnects the entire string, and the lost energy accumulates until a repair vessel can be mobilized.

Ring and looped topologies address this by providing a second path back to the substation, either by closing the far end of a string onto another string or by running a dedicated return cable. A ring converts a single cable fault from a whole-string outage into a partial outage that the protection scheme can isolate, at the cost of additional cable, additional switchgear, and a more complex protection philosophy that must handle bidirectional fault current. The choice is an economic one, made by comparing the cost of the redundant path against the expected value of the energy it would recover, and it is sensitive to water depth, distance from a repair port, and the assumed failure rate of the cable itself. Many projects adopt a hybrid approach, applying redundancy only to the strings whose loss would be most costly.

Cable Sizing Against Thermal and Cost Limits

Array cable sizing is a constrained optimization rather than a single calculation. The binding constraint is thermal: current heats the conductor, and the insulation system, typically cross-linked polyethylene, has a maximum continuous conductor temperature of 90 degrees Celsius. How much current produces that temperature depends on how effectively heat escapes into the surrounding seabed, which in turn depends on burial depth and on the thermal resistivity of the sediment. A cable buried deeply in dry sand runs hotter than the same cable buried shallowly in saturated silt, so the rating of an array cable is a property of the route as much as of the cable.

Against that limit the designer weighs cost. Conductor cross sections are drawn from a discrete ladder—95, 150, 240, 300, 400, 630, and 800 square millimeters are common—and cost rises steeply with size, as does bending stiffness, which complicates handling and laying. A well-designed string therefore steps down in size along its length, using a large cable for the heavily loaded section nearest the substation and progressively smaller cables toward the far end. Each step adds a joint or a change of cable type, so the number of steps is itself a trade-off between conductor cost and installation complexity.

Two further considerations bound the result. Electrical losses in the array network are proportional to the square of the current, so a larger conductor buys lower losses over the life of the plant; the optimum size is often one step above the thermally adequate size once lost energy is valued. And because wind farm output follows a wind distribution rather than a constant load, cables spend most of their life well below rated current. Dynamic rating schemes, which use the fiber in the cable to measure the actual conductor temperature along its length, allow operators to exploit the resulting thermal headroom instead of designing for a worst case that rarely occurs.

Dynamic Cables for Floating Platforms

A floating turbine cannot be served by a cable resting on the seabed. The platform moves continuously in surge, sway, heave, roll, pitch, and yaw, and the cable connecting it must accommodate that motion for the entire design life without fatiguing. The solution, borrowed directly from offshore oil and gas riser practice, is the dynamic cable: a suspended section shaped into a configuration such as a lazy wave, in which distributed buoyancy modules lift a portion of the cable into an arch that decouples platform motion from the touchdown point on the seabed.

Dynamic cables differ from static ones in construction as well as in geometry. They require armor designed for repeated bending rather than for a single lay operation, water-blocking that survives cyclic loading, bend stiffeners at the hang-off where the cable leaves the platform, and careful control of torsional balance so that the cable does not unwind under tension. Qualification is correspondingly more demanding. CIGRE Technical Brochure 862, published in January 2022 by Working Group B1.63, sets out recommendations for the mechanical testing of submarine cables for dynamic applications. It extends the static-cable test recommendations of Technical Brochure 623 with a defined full-scale fatigue test, and it draws on offshore oil and gas riser practice, notably ISO 13628-5 and the recommended practice DNV-RP-F401, which the working group took as its starting point. Design analysis couples a global motion model of the moored platform to a local model of the cable cross section, so that the fatigue damage accumulated in the armor wires can be estimated over the wave and current scatter diagram of the site.

Dynamic cables are more expensive per meter than static cables, they are longer than the water depth because of their curved path, and they represent a component with no long service history at wind farm scale. They are, consequently, one of the principal technical risks in floating wind, and one of the areas where the industry expects the sharpest cost reductions as designs mature and volumes rise.

Transformers

The step-up transformers are the heaviest single items on the topside and the principal driver of its structural design. A typical fixed-bottom platform carries two or three transformers, each of several hundred megavolt-amperes, connecting the 66 kV collection busbar to an export voltage of 132, 155, 220, or 275 kV. Splitting the duty across multiple units provides partial redundancy: the loss of one transformer curtails the plant rather than stopping it, and the remaining units can often be operated above nameplate rating for a period by exploiting forced cooling and the thermal inertia of the oil.

Offshore transformer design departs from onshore practice in several ways. Weight is minimized aggressively because topside mass propagates into foundation cost. Oil containment is mandatory, with bunds and drainage arranged so that a leak cannot reach the sea. Cooling must reject heat in an enclosed, salt-laden environment, which usually means a closed forced-oil circuit exchanging heat with air or seawater. And because transportation and replacement are extraordinarily difficult, the units are specified and tested with a conservatism that would be unusual ashore.

Switchgear and Protection

Medium-voltage switchgear on the platform terminates each array string in its own bay, providing circuit breaker, current and voltage transformers, and protection relays. At the export voltage, gas-insulated switchgear is standard, because its compact footprint suits a platform where volume is expensive and because a sealed enclosure protects live parts from a corrosive atmosphere. Historically that gas has been sulfur hexafluoride, a potent greenhouse gas; the industry is actively migrating to alternatives based on fluoronitrile mixtures or on vacuum interruption with clean air insulation, driven both by European Union restrictions on fluorinated gases and by developers' own emissions commitments.

The protection scheme must cope with a network that behaves unlike a conventional distribution system. Fault current is supplied predominantly by converter-interfaced turbines, which limit their contribution to little more than rated current and whose fault current phase angle is determined by control software rather than by machine impedance. Overcurrent protection graded on the assumption of a large, inductive fault infeed is therefore unreliable, and offshore collection networks lean instead on differential protection, distance protection with careful settings, and directional elements. Cable circuits are also predominantly capacitive, which complicates earth-fault detection and makes the choice of neutral earthing—resistance earthing being the common answer—a defining decision for the whole scheme.

Reactive Compensation and Auxiliary Systems

Beyond transformers and switchgear, the topside houses shunt reactors for reactive compensation, harmonic filters where studies show they are needed, an earthing transformer to establish a neutral reference, low-voltage distribution for platform services, uninterruptible supplies for protection and control, and standby diesel generation. The auxiliary supply deserves particular attention: when the wind farm is not generating, the platform and every turbine still require power for control systems, heating, dehumidification, lighting, and navigation aids, and that power must flow backward down the export circuit from shore. When the export circuit is unavailable, the diesel generator carries the load, and its fuel capacity sets how long the plant can survive a transmission outage without a vessel visit.

HVAC Export and Shunt Reactive Compensation

A submarine cable is a long coaxial capacitor. Its conductor and its metallic sheath are separated by a thin, high-permittivity insulation layer, so its capacitance per kilometer is roughly an order of magnitude greater than that of an overhead line. Under alternating voltage that capacitance draws a charging current that flows whether or not the wind farm is generating, and the reactive power associated with it grows with the square of the voltage and linearly with length. For a three-core cable at 150 kV, the charging reactive power is on the order of 1.5 megavars per kilometer, and it is substantially greater at 220 kV.

Charging current has two consequences. It consumes conductor ampacity that would otherwise carry real power, so the useful transmission capacity of an AC cable falls as it lengthens, reaching zero at a theoretical critical length. And it raises the voltage along the cable, which can push the system outside the limits the transmission operator permits. Shunt reactors, connected at one or both ends and sometimes at an intermediate point, absorb the surplus reactive power and restore the voltage profile. Hornsea Project One in the United Kingdom illustrates the extreme case. Its 1.2 gigawatt output is gathered at 34 kV, stepped up to 220 kV at three offshore substations, and carried to shore over three export circuits totaling roughly 467 kilometers of submarine cable, the longest AC offshore export cable system built at the time of its completion. Compensation at the ends alone could not hold the voltage profile, so the project added the first purpose-built offshore reactive compensation station, installed in 2018 roughly 60 kilometers from shore, near the midpoint of the longest circuit and carrying one 220 kV shunt reactor for each export cable.

HVAC export is simpler, cheaper, and better understood than the alternative, and it remains the default for shorter connections. Its practical range is extended by compensation, but each compensation asset adds cost, and an offshore reactive compensation station is itself a platform requiring foundation, fabrication, installation, and maintenance.

HVDC Export Beyond the AC Break-Even Distance

Direct current eliminates the problem rather than compensating for it. A DC cable has no steady-state charging current, so its transmission capacity does not decay with length and no reactive compensation is required along the route. The cost is a converter station at each end: a voltage-source converter platform offshore and a corresponding station onshore, together representing a large fixed investment that AC transmission does not incur.

The two cost curves cross. Below the break-even distance, HVAC is cheaper because it avoids the converters; above it, HVDC is cheaper because the AC solution needs more cable capacity, more compensation, and eventually more circuits. For offshore wind the break-even distance is commonly cited as falling between roughly 80 and 100 kilometers, though the figure is a range rather than a threshold: it moves with transmitted power, cable voltage, water depth, the strength of the receiving grid, and the price assumed for losses. Projects near the crossover are decided by study rather than by rule of thumb, and some have been built either way at comparable distances.

Offshore wind connections use voltage-source converters exclusively rather than the line-commutated technology of older point-to-point links, because a voltage-source converter can energize a passive network, control real and reactive power independently, and black-start the offshore collection system. Germany's North Sea connections, beginning with BorWin1 in 2009, established the pattern; the practice is examined in more detail on the companion page for HVDC transmission systems.

The 2 GW HVDC Standard

Each early offshore HVDC connection was engineered individually, at ratings between roughly 400 and 900 megawatts, with the delay and cost that bespoke design implies. The transmission operator TenneT responded with its 2GW Program, a deliberately standardized grid connection system intended to be replicated many times across the Dutch and German North Sea. The standard specifies a single platform design converting 2 gigawatts from 66 kV alternating current to a bipolar direct-current system at plus and minus 525 kilovolts, with a cable bundle comprising a positive pole, a negative pole, a metallic return, and a fiber-optic element. The metallic return is what makes the bipole useful in service as well as on paper: one pole can be taken out for maintenance while the other continues to carry roughly half the rated power, drawing from wind farms connected to both poles. The platforms are also specified as multi-terminal ready, with space reserved for the equipment that a future connection between converter stations—potentially across a national border—would require, so that the step to a meshed offshore grid does not mean rebuilding them.

The rationale is industrial as much as technical. Standardization allows a supply chain to be contracted in volume—TenneT tendered platforms and cable systems for many connections at once rather than project by project—compresses engineering effort, and makes spare parts and operating procedures common. Because each system carries more than twice the capacity of the connections then in operation, fewer than half as many are needed for a given target: the program was scoped at fourteen systems for 28 gigawatts of connection capacity, with a corresponding reduction in the number of platforms placed in the sea. The 2 gigawatt, 525 kV configuration has since become a de facto reference for other North Sea transmission operators, and it has pulled the extruded DC cable and converter supply chains toward a common specification.

Route Survey and Laying

Every cable route begins with survey work: geophysical surveying to map the seabed and the sediment beneath it, geotechnical sampling to characterize soil strength and thermal resistivity, and a search for unexploded ordnance, wrecks, boulders, and existing cables and pipelines. The route is then engineered to avoid hazards, to cross third-party infrastructure at acceptable angles under crossing agreements, and to keep the cable within its minimum bending radius throughout.

Installation is performed from cable-lay vessels carrying the cable in large rotating carousels or turntables. The cable is paid out under controlled tension while the vessel maintains position, and the laid catenary must be managed so that the cable neither goes slack, which risks a loop or kink, nor exceeds its tension limit. Pull-in at each turbine and platform brings the cable end up through a J-tube or an internal conduit, where a bend restrictor and a hang-off clamp secure it before termination. Weather windows govern the whole operation, and vessel day rates make schedule slippage expensive.

Burial and Protection

Burial is the primary defense against the dominant external hazard, which is contact with fishing gear and vessel anchors. Target burial depth is typically one to three meters, set by a burial risk assessment that weighs the local anchor and trawl threat against what the seabed will allow. Jetting tools fluidize sandy sediments and let the cable settle; mechanical ploughs cut a furrow in stiffer soils; cutting or trenching machines handle rock and hard clay. Where burial is impossible—on rock outcrops, at crossings, and in the scour zone immediately around a foundation—the cable is protected instead with rock berms, concrete mattresses, or articulated cast-iron shells.

The interface at the foundation deserves particular attention because it combines several stresses. The cable emerges from the seabed, passes through a scour-prone zone, and enters a rigid structure. Cable protection systems bridging that transition have themselves been a notable source of failures, and the Offshore Wind Accelerator has published guidance on their design and qualification. Standards specific to this domain include DNV-ST-0359, Subsea power cables for wind power plants, which sets requirements across every phase of a cable project, and the recommended practice DNV-RP-0360, Subsea power cables in shallow water, which gives detailed guidance for statically installed cables from concept through decommissioning.

Cable Failure and Its Consequences

Subsea cables are, by a wide margin, the largest source of financial loss in offshore wind. The insurer GCube, reviewing a decade of offshore wind claims from 2010 to 2020 in its report Uncharted Waters, found that subsea cables accounted for roughly 30 percent of claims by count but more than half of total claims expenditure, making them both the most frequent and the most expensive category. Over the same decade the average claim nearly doubled, from about 1.7 to about 3.1 million pounds.

The imbalance between frequency and cost is the essential point: a cable fault is not merely a repair but a marine campaign. Locating the fault, mobilizing a repair vessel, deburying the cable, cutting out the damaged section, installing joints, and reburying can take months, and the lost generation over that period often exceeds the physical repair cost. Root causes cluster around installation damage, inadequate or lost burial, failures of cable protection systems at the foundation interface, and manufacturing or jointing defects—that is, around handling and protection rather than around the electrical design of the cable itself. GCube attributed 55 percent of claims by frequency and 83 percent of claims expenditure across the whole market to contractor error and component defect, and 44 percent of cable claims expenditure specifically to contractor error during transit and laying.

Operators respond on several fronts. Distributed temperature and strain sensing over the integral fiber gives early warning of overheating and of movement. Periodic surveys verify that burial depth has not been lost to seabed mobility. Partial discharge monitoring detects insulation degradation before breakdown. And condition-based approaches feed back into design, since the evidence from claims consistently points to the interfaces and the installation phase as the places where reliability is won or lost.