An eddy current is a circulating electric current induced within an electrical conductor when the conductor experiences a changing magnetic field. Such changes may arise from a time-varying external field, movement through a spatially varying field, or movement that changes the magnetic flux linked by possible current paths. Unlike current deliberately routed through insulated wires, eddy currents form distributed paths within the conducting material itself. Their name refers to the resemblance between these circulating patterns and eddies in flowing water.
Eddy currents are an important manifestation of electromagnetic induction. They generate magnetic fields that react against the change producing them, exert mechanical forces, and convert electrical or mechanical energy into heat. These effects can be unwanted, as in the losses of transformers and electrical machines, or useful, as in induction heating, contactless braking, metal detection, and industrial inspection. Their magnitude and distribution depend jointly on the excitation, the conductor’s shape, its electrical properties, and the penetration of the magnetic field into its interior.
Physical origin and governing laws
The fundamental relation behind eddy currents is Faraday’s law of induction. For a stationary closed contour bounding a stationary surface , it states that
where is the electric field and is the magnetic flux density. A changing magnetic flux therefore produces an electric field with nonzero circulation. If conducting material occupies the region, that field drives current. No fabricated wire loop is necessary: a solid plate or block provides many possible conducting paths.
The minus sign expresses Lenz’s law: the induced response opposes the change responsible for it. If an applied field through a conducting plate increases, the induced currents produce a field opposing that increase. When the applied field decreases, their direction reverses, tending to maintain the original flux. This is not a general rule that induced fields always oppose applied fields; they oppose changes in linked flux. The distinction matters in alternating fields, where the relative phases of excitation and response depend on resistance and inductance.
Motion supplies another route to induction. Charges moving with material velocity through a magnetic field experience the Lorentz force. In a simple isotropic conductor, neglecting Hall and thermoelectric effects, the local constitutive relation is
where is current density and is electrical conductivity. The electric field includes fields produced by redistributed surface charges. Consequently, motion through a magnetic field does not automatically imply sustained eddy currents. A finite conductor translating uniformly through an ideal uniform, steady field can develop charge separation that cancels the magnetic driving term. Entering or leaving a localized field, rotating in an appropriate geometry, or encountering a nonuniform field can instead support circulating currents.
Ohm’s law and charge conservation determine how these currents close within the available material. In the quasistatic conditions typical of many applications, bulk charge accumulation is small and . At an electrically insulating boundary, the normal current vanishes. Holes, slots, interfaces, and changes in conductivity therefore redirect current paths even when the applied field remains unchanged.
The associated irreversible heating has local power density
for a stationary, isotropic conductor. In moving material, remains the resistive heating density, but electrical energy transfer and mechanical work must be distinguished. The induced field, material motion, and current distribution form a coupled system rather than a one-way response to an unaffected external source.
Magnetic diffusion and skin depth
Eddy currents alter the field that induces them. Combining Maxwell’s equations with a linear constitutive law gives a particularly useful description for a stationary, homogeneous conductor. If displacement current can be neglected and the magnetic permeability and conductivity are constant, the magnetic field obeys
This is a magnetic diffusion equation. Its mathematical form resembles that of heat conduction, but the diffusing quantity is magnetic field rather than temperature. The magnetic diffusivity is .
For a characteristic dimension , the field-adjustment time has the order of magnitude
The numerical coefficient depends on geometry and boundary conditions. High conductivity, high permeability, and large dimensions all tend to prolong the persistence of induced currents and delay penetration of changing fields. This explains why a thick conducting structure can continue to generate a secondary field after an external pulse has ended. Real structures often exhibit several decay times, corresponding to different spatial current patterns.
For a sinusoidal field applied to a thick, flat conductor, field amplitude decreases exponentially with depth. The characteristic penetration distance is the skin depth,
where is frequency and . At one skin depth the amplitude is , approximately 37 percent, of its surface value; it has not fallen to zero. The field also changes phase with depth. Thus skin depth characterizes both attenuation and a spatially distributed timing response.
For copper near room temperature, taking and , the skin depth is approximately 9.3 mm at 50 Hz, 2.1 mm at 1 kHz, and 0.21 mm at 100 kHz. These values illustrate why low-frequency fields penetrate substantial metal thicknesses, whereas radio-frequency heating and sensing are often concentrated near surfaces. In ferromagnetic metals, permeability can be much greater and can vary with field strength, frequency, and material condition, making a single penetration-depth value less representative.
The standard skin-depth expression assumes a linear, uniform medium and a locally planar surface. Thin sheets, sharp edges, curved parts, layered materials, and nearby conductors require more detailed treatment. At sufficiently high frequencies, displacement current and wave propagation also matter. In very pure metals at low temperature, electron transport can become nonlocal, producing an anomalous skin effect outside the assumptions of the ordinary conductivity model.
Current paths, heating, and geometric scaling
Eddy-current patterns are set by both the applied magnetic field and the conductor’s topology. A field perpendicular to a broad plate tends to induce currents circulating in the plane of the plate. A field parallel to a thin sheet drives loops whose cross sections are restricted by the sheet thickness. The distinction explains why the orientation of laminations is as important as their thickness: effective laminations interrupt the relevant loops without obstructing the intended magnetic flux path.
In a simple thin-sheet model, the dependence of loss on thickness and frequency can be calculated explicitly. Consider an electrically isolated sheet of thickness , with a spatially uniform sinusoidal magnetic flux density parallel to its faces,
If the sheet thickness is small compared with the skin depth, its dimensions along the face are large, and edge effects are neglected, the mean classical eddy-current loss per unit volume is
where is electrical resistivity. This formula describes the classical eddy-current contribution, not every source of loss in a magnetic material.
The quadratic thickness dependence makes subdivision especially effective. Under the assumptions of the formula, halving sheet thickness reduces this loss per unit volume by a factor of four. Doubling frequency or peak flux density increases it by a factor of four. These relations cannot be extrapolated indefinitely: strong skin effects, nonlinear magnetization, edge fields, and changing excitation conditions alter the current distribution.
Resistance alone does not determine the response. A current loop also has inductance, so its self-generated magnetic field limits rapid changes in current. An equivalent resistance–inductance model can approximate a dominant circulating path, but a solid conductor generally supports many interacting paths. Treating the entire object as one simple loop can miss local heating, different decay times, or the concentration of current around holes and corners.
Total heating follows from a volume integral,
Because the integrand depends on the square of current density, a small region of concentrated current can contribute disproportionately to heating. This is important near narrow bridges between slots, imperfect electrical contacts, and structural features exposed to strong leakage fields. The resulting temperature rise also depends on heat capacity, cooling, and thermal conductivity, not solely on electrical loss.
Sharp changes in applied waveforms introduce higher-frequency components. Consequently, a device supplied by a switched converter may have substantial eddy-current losses even when its fundamental operating frequency is relatively low. Reliable calculations must account for the waveform actually experienced by the material rather than replacing every excitation with an ideal sine wave.
Discovery and development of the theory
The experimental history of eddy currents began before electromagnetic induction had a general explanation. In 1824, François Arago reported that a rotating copper disk could influence a nearby magnetic needle. When the disk rotated, the needle tended to follow its motion despite the absence of the strong static attraction associated with iron. These “rotations of magnetism” showed that relative motion between a magnet and a nonferromagnetic conductor could produce mechanical interaction.
The observations were initially puzzling because copper does not retain a substantial permanent magnetization under ordinary conditions. The essential cause was not a permanent magnetic state of the disk but currents generated by its motion relative to the magnet’s nonuniform field. Their magnetic interaction transferred torque between disk and needle. Arago’s arrangement thus revealed an induction effect indirectly through motion before the relevant electrical mechanism was established.
Michael Faraday discovered electromagnetic induction in 1831. His experiments with coils, magnets, and conducting disks established that changing magnetic conditions could generate electrical effects. They supplied the conceptual basis for understanding Arago’s observations. Heinrich Lenz formulated the directional rule for induced currents in 1834, connecting their mechanical opposition to the process that generates them. This opposition became central to understanding both magnetic drag and the conversion of mechanical work into heat.
Léon Foucault’s 1855 experiments demonstrated pronounced braking and heating when a conducting disk rotated in a magnetic field. The alternative name “Foucault currents” reflects this association. The combination of resistance to rotation and rising disk temperature made the energy conversion particularly clear: sustained rotation required mechanical work, which was dissipated electrically within the metal rather than by rubbing surfaces.
The field description developed through the nineteenth century, especially with the electromagnetic theory of James Clerk Maxwell. Distributed induction could then be expressed through local relations among electric field, magnetic field, and current density. This framework explained why the response depended on shape, conductivity, frequency, and the magnetic field generated by the currents themselves, rather than only on a single circuit resistance.
Practical interest expanded with electrical machinery and alternating-current power systems. Designers had to suppress heating in iron cores while retaining useful magnetic flux. Thin insulated sheets became a central construction technique. In parallel, induction effects became tools for examining metals: David Edward Hughes demonstrated induction-balance methods in 1879 that could reveal differences between metallic specimens. Later developments in electronics, phase-sensitive measurement, and numerical field calculation transformed such observations into quantitative inspection techniques. The historical progression linked three uses of the same phenomenon: explaining magnetic drag, controlling energy loss, and extracting information about materials.
Losses in transformers and electrical machines
In a transformer, alternating magnetic flux links primary and secondary windings through a magnetic core. If the core were a solid, highly conducting block, the same changing flux would induce large circulating currents within it. Those currents would waste power and produce heating without contributing to the intended transfer of energy between windings. Eddy-current loss is therefore one component of core loss, distinct from magnetic hysteresis associated with cyclic changes in magnetization.
Power-frequency transformer cores commonly use thin sheets of electrical steel, electrically insulated from one another. The sheets are oriented so that the principal magnetic flux lies approximately in their planes. Insulation interrupts large current loops that would otherwise cross the stack, while small sheet thickness limits loops within each lamination. Silicon alloying raises steel’s resistivity, further reducing eddy-current loss. Lamination thicknesses of a few tenths of a millimetre are common, although the appropriate value depends on frequency, material, cost, and manufacturing requirements.
At higher frequencies, different core materials become advantageous. Ferrites have much higher electrical resistivity than metallic magnetic alloys, greatly restricting bulk circulating currents. Powder cores subdivide conducting magnetic material into particles separated by insulating boundaries. Amorphous and nanocrystalline metallic materials are often manufactured as thin ribbons, combining favourable magnetic properties with short eddy-current paths. None of these choices eliminates all loss: hysteresis, relaxation processes, winding resistance, and local field concentration remain relevant.
An electric motor contains several possible sources of eddy currents. Alternating flux in stator and rotor iron produces core loss, while spatial harmonics, slotting, and converter switching can expose conductive parts to additional time-dependent fields. Permanent magnets can also develop eddy currents, particularly when they encounter rapidly changing harmonic fields. Dividing a magnet into electrically isolated segments can reduce these currents, although the effectiveness depends on segment orientation and the three-dimensional field pattern.
Not all induced rotor currents are parasitic. An induction motor deliberately uses currents induced by relative motion between a rotating magnetic field and the rotor. In a squirrel-cage rotor, conducting bars and end rings define the principal current paths. Solid-rotor machines also exploit distributed induced currents. The distinction between useful induction and undesirable eddy-current loss is consequently functional rather than a difference in the underlying physical law.
Leakage flux can heat transformer tanks, clamps, bolts, shields, and other structural metalwork. Such losses require analysis beyond an ideal core-and-winding model. Countermeasures include magnetic shunts, conductive shields, nonmagnetic components, interrupted current paths, and altered geometry. In windings, skin and proximity effects likewise redistribute alternating current, increasing effective resistance. Transposed conductors and insulated fine-strand arrangements address these effects by controlling the paths available to circulating and redistributed currents.
Induction heating and material processing
Induction heating deliberately converts induced electrical currents into thermal energy. An alternating current in a coil creates a changing magnetic field, which induces currents in a nearby conducting workpiece. The workpiece behaves in part like a short-circuited transformer secondary, although actual current paths are distributed through its volume. The electromagnetic power supply and coil deliver energy without requiring an electrical contact to the heated object.
The spatial heating pattern depends strongly on frequency. Lower frequencies generally allow deeper field penetration; higher frequencies concentrate power nearer the surface. Coil shape controls where the field is strongest, while the spacing between coil and workpiece affects coupling. A cylindrical coil around a shaft, a flat spiral beneath a vessel, and a shaped coil around a gear tooth therefore produce different current distributions even when operating at the same frequency.
In ferromagnetic workpieces, heating may include both eddy-current dissipation and magnetic hysteresis. Conductivity and permeability change with temperature, so the electromagnetic conditions evolve as heating proceeds. Near and above the Curie temperature, ferromagnetic order disappears and permeability changes substantially. A heating calculation that assumes fixed room-temperature properties can therefore misrepresent penetration and power deposition during a thermal cycle.
Industrial applications include melting, forging preparation, brazing, soldering, shrink fitting, and surface hardening. Surface hardening uses rapid localized heating followed by controlled cooling to change the near-surface microstructure while limiting changes deeper in the part. In induction melting, currents can also generate electromagnetic forces that stir the liquid metal, affecting mixing and temperature uniformity. These forces are distinct from the resistive mechanism that supplies heat.
Domestic induction cooking applies the same principle on a smaller scale. A coil below an electrically insulating cooking surface couples magnetic energy into the vessel base. Conventional appliances are designed primarily for cookware with suitable magnetic and electrical properties; the vessel construction strongly affects coupling and efficiency. Although heating occurs directly in the vessel, the cooking surface can subsequently become hot through thermal contact with it.
Electromagnetic penetration depth is not identical to the final depth of a heated layer. Heat spreads by conduction after it is deposited, and moving workpieces transport energy through the process. A thin surface region may receive most of the instantaneous electrical power while a much larger volume warms over time. Predicting the outcome requires coupled electromagnetic and thermal analysis, sometimes including phase transformations, latent heat, and fluid motion. The resulting process depends on both where energy enters and how quickly it can redistribute.
Braking, damping, and magnetic forces
Eddy-current devices can exert force without mechanical contact. A conductor moving through a localized magnetic field develops induced currents whose interaction with that field produces a force density
The total force and torque follow by integrating over the conductor. In a passive braking arrangement, the component opposing motion removes kinetic energy, which is ultimately dissipated as Joule heat in the conductor and other resistive parts.
A familiar demonstration is a strong permanent magnet falling through a copper or aluminium tube. Although these metals are not ferromagnetic, the falling magnet changes the flux through successive sections of the tube. Induced currents create a force opposing its downward motion. At a suitable terminal speed, the upward magnetic braking force balances gravity. Cutting a longitudinal slit in the tube interrupts important circumferential current paths and substantially weakens the braking effect.
An eddy-current brake applies this principle to a rotating disk, drum, or moving rail. At sufficiently low speeds and under approximately fixed magnetic conditions, induced voltage and current often scale with speed, producing a braking force or torque that is approximately proportional to speed. At higher speeds, current redistribution, inductive effects, and magnetic shielding alter that relation. A simple linear extrapolation is therefore generally inadequate over the full operating range.
A central limitation is that a passive eddy-current brake does not ordinarily supply holding torque at rest. With no relevant relative motion and no changing excitation, the induced currents decay. Such brakes are useful for speed reduction, dynamometer loading, exercise equipment, and some transport braking systems, but stationary holding requires another mechanism. The absence of rubbing contact reduces wear at the electromagnetic interaction surface, yet the generated heat still has to be removed.
Eddy currents also provide damping in instruments and mechanisms. A conducting vane moving through a magnetic gap experiences a resisting force that suppresses oscillation. In a suitable small-motion regime, this resembles viscous damping: force depends approximately on velocity rather than displacement. The effect can reduce settling time without introducing the same contact friction and stiction as a rubbing damper.
The force need not be purely a drag force. Appropriate field geometry and phase relationships can create lift or repulsion as well as braking. Electrodynamic suspension and some magnetic levitation arrangements exploit these reactions. Eddy-current separators use rapidly changing magnetic fields to impart forces to nonferrous conducting fragments, helping separate them from less-conducting waste. In each case, the direction and magnitude of force must be derived from the induced-current pattern rather than inferred from the statement that magnetic fields “repel conductors.”
Nondestructive testing and sensing
Eddy-current testing examines conducting objects through their electromagnetic response without requiring the material to be cut apart. A typical probe contains an alternating-current coil placed near the surface. The coil induces currents in the specimen, and the secondary magnetic field produced by those currents changes the coil’s electrical impedance. Separate transmitting and receiving coils can also be used.
A crack, corrosion cavity, thickness change, or material boundary disturbs the current distribution. A surface-breaking crack is especially detectable when it interrupts currents that would otherwise cross it. Probe orientation therefore matters: a crack aligned with the local current can give a weaker signal than one cutting across it. Conductivity, permeability, and geometry all affect the measured response, so a signal change is not uniquely a defect indication.
Frequency selection determines the balance between penetration and spatial sensitivity. High frequencies emphasize near-surface regions and small surface features; lower frequencies can probe more deeply but often with reduced spatial resolution. Conventional sinusoidal inspection is limited by attenuation with depth. Pulsed eddy-current methods use transient excitation and analyse the time-dependent response, allowing different aspects of the conductor’s depth-dependent behaviour to be examined.
The distance between probe and specimen, often called lift-off, is a major influence. Paint, surface roughness, probe tilt, and uneven contact can change coupling even if the underlying metal is sound. Differential probes compare nearby regions to reduce sensitivity to slowly varying background conditions. Multifrequency techniques combine measurements at several frequencies to help distinguish overlapping influences, such as support structures, geometry changes, and flaws.
Applications include aircraft skin and fastener-hole inspection, heat-exchanger tubing, surface examination of metal components, and assessment of electrically conducting coatings or substrates. Eddy-current instruments can also sort alloys or estimate conductivity because composition, heat treatment, and microstructure affect electrical properties. In ferromagnetic materials, permeability variations can dominate the response; magnetic saturation is sometimes used in specialized inspection arrangements to reduce that complication.
Quantitative interpretation depends on calibration using representative reference standards, often containing artificial notches, drilled features, or known thickness variations. Such features do not necessarily reproduce every aspect of a natural flaw. Probe response also depends on flaw depth, opening, shape, orientation, and surrounding structure. Reporting a numerical defect size therefore requires an appropriate model and an assessment of measurement uncertainty, rather than a direct conversion from signal amplitude alone.
Field control, scientific instruments, and numerical modelling
Eddy currents are important wherever magnetic fields must change rapidly and predictably. In magnetic resonance imaging, switched gradient coils can induce currents in nearby conducting structures. Their secondary fields persist after the commanded gradient has changed, causing spatial and temporal errors that can distort images or alter signal phase. These effects arise principally in scanner hardware, although changing fields can also induce electric fields within the subject.
Actively shielded gradient coils use additional windings to reduce field changes outside the intended region, thereby limiting excitation of surrounding conductive structures. Electronic pre-emphasis modifies the commanded waveform to compensate for measured delayed responses. A common representation expresses the residual field as a sum of decaying components,
where each amplitude and time constant describes part of the system response. Real field errors may also vary spatially and couple between gradient directions, so correction can require more than one scalar decay curve.
In particle accelerators, changing magnet fields induce currents in metallic vacuum chambers and other nearby structures. These currents can delay field penetration, alter the field seen by the beam, and generate heat. Chamber material, wall thickness, segmentation, and magnet ramp rate consequently influence achievable performance. Thin or relatively resistive walls can reduce some effects, but mechanical integrity, vacuum requirements, and other electromagnetic constraints must also be satisfied.
Conducting enclosures can shield alternating magnetic fields through their induced response. Such shielding is frequency-dependent: a normal conducting shell generally provides much less protection against very slowly varying magnetic fields than against sufficiently rapid changes. Static-field shielding instead relies on other mechanisms, such as flux diversion through high-permeability material. Perfect-conductor idealizations also differ from superconductivity; the exclusion of magnetic flux in the superconducting Meissner state is not explained simply by ordinary resistive eddy currents.
Engineering calculations commonly use the finite element method to solve electromagnetic field equations in realistic geometries. Relevant inputs include conductivity, magnetic constitutive behaviour, source waveforms, material motion, and boundary conditions. When skin depth is small, the mesh must resolve steep near-surface gradients, or suitable surface-impedance approximations must replace explicit modelling of the entire conductor interior.
A coupled model may transfer calculated losses into a thermal solution and feed the resulting temperature-dependent properties back into the electromagnetic calculation. Motion can require an additional mechanical solution. Validation commonly compares predicted coil impedance, force, field decay, or temperature with measurements. Agreement in total power alone does not establish that local current concentration and peak temperature have been represented correctly, particularly near joints, edges, and narrow conducting paths.