How Maglev Trains Actually Work: The Physics (2026)

A maglev train glides a few millimeters above its track, hits 500 km/h with no rolling friction, and looks like cheating. It is not. Understanding how maglev trains actually work comes down to three coupled subsystems — levitation, propulsion, and guidance — and two competing engineering schools that solve them in very different ways. In 2026 the Chuo Shinkansen is still trickling toward partial revenue service, China has commissioned a 600 km/h demonstrator, and yet conventional high-speed rail keeps winning the cost-per-passenger fight. This post unpacks the physics behind the levitation gap, the linear synchronous motor, and the guidance loop, then asks the harder question: if maglev is so good, why is the world map of working lines still so short?

What this post covers: subsystem decomposition, EMS vs EDS physics, linear motor propulsion, guidance and lateral stability, the 2026 SCMaglev status, and the honest economics of why maglev has not displaced HSR.

What “maglev” really means

Answer-first summary: A maglev (magnetic levitation) train is any rail vehicle that suspends itself off the track using magnetic forces rather than wheels, and is propelled by a linear motor rather than a rotary drive. Strip away the marketing and the defining feature is contactless force transfer for both lift and thrust.

The everyday word “train” implies steel wheels on steel rails. Maglev breaks both halves of that picture. The vehicle never touches the running surface above its takeoff speed, and the motor is wrapped around the guideway instead of bolted under the car. That single architectural switch removes the rolling-contact wear envelope that has bounded conventional rail since the 1830s.

It also adds new constraints. Without wheels you cannot use the rail head to steer the car, so guidance becomes an independent magnetic problem. Without a rotary motor on the bogie you cannot brake by reversing the traction system in a simple way, so braking has to combine regenerative linear motor braking with eddy-current brakes and (at low speed) friction shoes.

The term covers a wide design space. Three operating systems have carried passengers in revenue service: the Shanghai Transrapid Pudong shuttle (EMS, German design, commissioned 2003), the Linimo line in Aichi, Japan (EMS, low-speed urban, 2005), and the South Korean Incheon Airport Maglev (EMS, also low-speed). The Japanese SCMaglev (EDS, superconducting) is the most ambitious system and is still pre-revenue at the time of writing, with the Chuo Shinkansen project running well behind its original 2027 Tokyo–Nagoya target. China has run a 600 km/h Qingdao-built EMS prototype on test loops but has not yet committed to a long inter-city revenue line.

If you want background on how precision timing underlies the train control systems on these networks, our explainer on how GPS atomic clocks work and precision timekeeping is a useful companion — modern train control depends on time discipline at a level most riders never see.

The three subsystems every maglev needs

Answer-first summary: Every maglev, no matter the brand, decomposes into three subsystems that must work simultaneously: levitation (vertical force against gravity), propulsion (horizontal force along the track), and guidance (lateral force keeping the car on the centerline). Each subsystem is its own control loop, and they are mechanically and electrically coupled.

The diagram in Figure 1 is the mental model to carry through the rest of the post. The vehicle carries magnets (electromagnets, permanent magnets, or superconducting coils, depending on the design). The guideway carries reaction surfaces (steel rails, sidewall coils, or stator windings). Each subsystem owns a piece of the magnetic interaction between the two.

Levitation generates a vertical force equal to the vehicle weight, held within an air gap that ranges from about 8 mm in EMS systems to roughly 100 mm in SCMaglev EDS systems. Propulsion is supplied by a linear synchronous motor whose stator is along the guideway and whose rotor is the magnet array on the car. Guidance produces a restoring lateral force whenever the car drifts off centerline.

The three loops have very different bandwidths. The levitation loop has to be fast — typically a kilohertz-rate current control loop in EMS, since the open-loop physics is unstable. Propulsion is set by the trackside power electronics and runs at the frequency of the traveling wave, in the tens of Hertz at cruise. Guidance can be slower in EDS designs because it is passively stable, but in EMS it shares the same control hardware as levitation.

The interaction matters. A bump that pushes the car upward changes the levitation gap, which changes the propulsion air-gap coupling, which changes the thrust. Real maglev controllers do not treat the three loops as independent; they linearize a multi-input, multi-output model and design a coupled state-feedback controller. Vendors do not publish the exact gain matrices, but the academic literature in IEEE Transactions on Magnetics and IEEE Transactions on Transportation Electrification documents the structure clearly.

Electromagnetic suspension (EMS) — the German Transrapid school

Answer-first summary: EMS uses ordinary DC electromagnets on the vehicle that are attracted upward to a ferromagnetic rail under the guideway. The system is intrinsically unstable, so a high-bandwidth current controller modulates coil current in real time to hold an air gap of roughly 8–10 mm. The Shanghai Transrapid and Linimo lines are the production references.

The physics

Two facts drive EMS design. First, the force between a current-carrying iron-core electromagnet and a ferromagnetic plate is attractive and varies roughly as the square of the current and inversely as the square of the gap. Second, Earnshaw’s theorem from 1842 proves that a configuration of static magnets cannot levitate a body in stable equilibrium using only inverse-square forces. EMS sidesteps Earnshaw by making the coil current actively controlled rather than static.

The control loop in Figure 2 is the heart of the system. An inductive gap sensor reads the air gap. A real-time controller compares the gap to a setpoint (typically 8–10 mm for the Transrapid TR-09) and commands a DC chopper to adjust coil current. Increase current to pull harder when the gap grows; decrease current when the gap shrinks. The loop has to run faster than the mechanical resonance of the car suspension, which puts the controller in the multi-kilohertz range.

Why EMS is the easier engineering choice

EMS magnets are conventional copper coils on iron cores. No cryogenics. No superconductors. The on-board power requirement is modest because the magnets are operating in a small gap where attractive force per ampere is high. The Transrapid TR-09 manages this with on-board batteries plus inductive power pickup from the guideway, so the vehicle stays levitated even when external power is interrupted briefly.

The trade is the small air gap. Eight millimeters is roughly the thickness of a phone charger cable. Track survey tolerances, thermal expansion of the guideway, and snow or ice accumulation all eat into that budget. EMS guideways have to be built and maintained to civil-engineering tolerances that are tighter than conventional HSR, which contributes to the line cost.

Where EMS sits in 2026

EMS still has the only continuously running revenue maglev lines in 2026: Shanghai Pudong (commissioned 2003, ~30 km, top speed 430 km/h), Linimo near Nagoya (2005, low-speed urban), and Incheon Airport Maglev (2016, also low-speed). The 600 km/h Qingdao prototype is also an EMS design, scaled up. No new EMS revenue lines have entered service since Incheon.

Electrodynamic suspension (EDS) — the Japanese SCMaglev school

Answer-first summary: EDS levitation uses powerful superconducting magnets on the vehicle that induce eddy currents in passive figure-eight “null-flux” coils set into the guideway sidewalls. Lift only appears once the car moves fast enough — typically above ~150 km/h — so the vehicle uses retractable rubber wheels at low speed. The Japanese SCMaglev Chuo Shinkansen is the canonical example.

The physics

EDS works by Lenz’s law. A strong, slowly varying magnetic field on the vehicle moves past a passive coil on the track. The changing flux through the coil induces an EMF, the coil’s own resistance and inductance let an eddy current flow, and that current creates an opposing magnetic field. The opposition manifests as repulsive lift on the moving magnet.

For EDS to make useful lift, the induced current has to be large, which means the source field has to be strong and the motion has to be fast. The strength comes from superconducting coils that carry hundreds of kiloamperes of effective current at zero ohmic loss. In SCMaglev these coils sit inside a cryostat at roughly 4 K cooled by liquid helium, though Central Japan Railway has publicly discussed transitioning to high-temperature superconductor (HTS) coils that could run at 20 K and be cooled by closed-cycle cryocoolers — a switch that would simplify on-board cryogenics significantly.

The speed dependence is what forces the rubber-wheel arrangement. Below roughly 150 km/h the induced currents are too small to lift the 60-tonne car. Above that, lift grows and the wheels retract. EDS is therefore not a pure maglev at standstill — but it does not need active control to stay aloft once it is moving.

Null-flux geometry — why the coils are figure-eight

The clever bit of the SCMaglev sidewall coil is the figure-eight winding. When the on-board magnet is exactly centered on the coil, the flux through the upper loop equals the flux through the lower loop but with opposite sign, so the net EMF is zero and no current flows. As soon as the car drops or rises, the flux balance breaks and a current flows that pushes the car back to center. The same coils therefore provide both levitation and vertical guidance in one passive structure. Sumiyoshi Miyamoto and others at the former JNR research lab introduced this geometry in the 1970s and it remains the technical signature of the Japanese system.

Why EDS is harder but more tolerant

EDS sounds harder than EMS — and it is, in cryogenic and electrical engineering terms. But the system has two structural advantages.

First, the levitation gap is on the order of 100 mm rather than 8–10 mm. That is more than ten times the EMS budget. Survey tolerances, thermal expansion, and weather effects shrink to a much smaller fraction of the gap. The Chuo Shinkansen runs through long, deep mountain tunnels where guideway temperature swings are not trivial, and the wide gap absorbs them comfortably.

Second, EDS is passively stable in the vertical direction. The null-flux geometry restores any deviation without active control. The on-board magnets are essentially “set and forget” once they are persistent-current-trained. The active control is on propulsion, not on lift.

Our explainer on how atomic clocks work and quantum resonance applications in GPS shares some of the underlying low-temperature physics, and our quantum sensing explainer covering MRI and SQUID magnetometers explores the same family of superconducting circuits used in SCMaglev coils, just at very different field strengths.

Linear synchronous motor propulsion

Answer-first summary: Maglev propulsion almost always uses a long-stator linear synchronous motor (LSM). The stator winding sits in the guideway, the field magnets sit on the vehicle, and trackside inverters create a traveling three-phase magnetic wave that the vehicle’s magnets lock onto. Trackside power means no high-power equipment on the car.

Unrolling a rotary motor

A standard three-phase synchronous motor has stator windings around a cylindrical rotor. Energize the windings with three currents 120 degrees apart and a rotating magnetic field appears in the air gap; the rotor’s field magnets lock to it. Unroll that geometry into a flat strip and you have a linear synchronous motor: the stator is now a series of coils laid along the guideway, the rotor is the vehicle’s magnet array, and the rotating field becomes a traveling field that moves down the track at a speed set by the inverter frequency.

The thrust per unit length is set by the stator current and the on-board magnet strength. For SCMaglev the on-board magnets are the same superconducting coils that handle levitation, so the propulsion engagement is very strong. For Transrapid the on-board “long stator” magnets are conventional but well-matched to the stator coils embedded in the guideway.

Segmenting the stator

A maglev guideway can be hundreds of kilometers long. Energizing every meter of stator winding all the time would be wasteful, so the stator is segmented. Only the section currently carrying a train is energized; the rest is idle. Substations along the route each feed one or two segments through high-power inverters. Segment hand-off — passing a train from one inverter’s segment to the next — is a fast control problem solved with overlap zones and synchronized firing.

Power per segment is significant. SCMaglev sections during the Yamanashi test runs drew tens of megawatts at peak. That cost is paid by the operator at trackside substations, not by an on-board pantograph — which is why maglev can run faster than conventional HSR, where the pantograph-catenary current pickup becomes a physical bottleneck around 350–400 km/h.

Braking

Linear motors brake by inverting their phase relationship: the traveling wave is shifted to oppose the car’s motion, and the kinetic energy is fed back into the grid through the inverters. Below the speed where regenerative LSM braking is effective, EDS systems use eddy-current brakes (drag the on-board magnets against passive aluminum strips), and below that the rubber wheels engage friction brakes. The Transrapid TR-09 likewise has multi-mode braking with a friction backup.

Guidance, lateral stability, and yaw

Answer-first summary: Guidance is the third subsystem and the one most often glossed over. In EMS the same electromagnets that lift the car also have lateral pole faces that pull it back to centerline, controlled by additional gap sensors. In EDS the figure-eight null-flux sidewall coils handle both lift and guidance passively. Yaw and roll damping are typically active in both systems.

The maglev train guidance system is what keeps the car aligned with the guideway, especially through curves and crosswinds. Conventional rail uses the conical profile of the wheel tread plus the flange of the inner wheel as a self-correcting mechanism — and that mechanism is wear-limited above 350 km/h. Without it, maglev has to make guidance an explicit magnetic problem.

EMS guidance

In the Transrapid design, the U-shaped guideway wraps under the car. The vehicle’s lifting electromagnets are oriented to pull upward, and a second set of magnets on either side pull inward against the vertical webs of the guideway. Lateral gap sensors feed a separate control loop that modulates side-magnet current to hold the car centered. The bandwidth is similar to the lift loop, and the controllers are mechanically packaged together in the same bogie module.

EDS guidance

In SCMaglev the figure-eight null-flux coils on each sidewall handle lateral as well as vertical restoring force. If the car shifts toward the right wall, the flux balance on the right-side coils breaks and the induced current pushes the car away from that wall. The left-side coils respond symmetrically. No active controller is needed for the lateral DOF below the resonant frequency of the suspension.

Yaw and roll

Pure null-flux guidance damps lateral and vertical translation but not the yaw and roll rotations. Both EMS and EDS systems add active dampers — sometimes magnetorheological, sometimes electromechanical — between the bogie and the cabin to damp those modes. Without them, ride quality at 500 km/h would be unacceptable, and motion sickness on curves would be a real problem.

Chuo Shinkansen — the 2026 status

Answer-first summary: The Central Japan Railway Chuo Shinkansen, which will use SCMaglev between Tokyo and Nagoya, is the largest active maglev project in 2026. Tunneling delays through Shizuoka Prefecture have pushed the original 2027 opening target well beyond 2030. Test operations at Yamanashi continue, and the system has held the 603 km/h piloted rail speed record since 2015.

The Chuo Shinkansen is the maglev project most often cited in the press, and the one most worth understanding in detail. The line will eventually run from Shinagawa in Tokyo to Nagoya (~286 km) and ultimately to Shin-Osaka. Roughly 80–90% of the route is in tunnel because the route runs through the Akaishi Mountains. JR Central is the operator and the system uses SCMaglev cars derived from the L0 series tested at the Yamanashi test line in Tsuru.

In April 2015 a manned L0 set the world rail speed record at 603 km/h on the Yamanashi line. That figure has not been surpassed by any rolling-stock system. JR Central has separately stated that revenue operating speed will be 505 km/h, leaving headroom for safety and energy considerations.

The construction story since 2015 has been less triumphal. The Shizuoka Prefecture section, which crosses the upper Oi River basin, became a multi-year political fight over how tunneling might affect groundwater flows and the river volume reaching downstream tea farms. The prefecture withheld construction approval for years. JR Central has publicly stated that the original 2027 Tokyo–Nagoya opening cannot be met, and credible recent reporting puts the partial opening into the early 2030s. Total project cost was last reported in the ¥9 trillion range (roughly USD 60 billion at typical 2026 exchange rates), but cost figures should be treated as estimates pending JR Central’s next official update.

For readers who want a parallel example of an enormous capital project that delivered transformational results, our explainer on cryo-EM at 1.2-ångström resolution shows what sustained investment in advanced technology can produce in a different domain.

Why maglev is still niche — cost, integration, energy

Answer-first summary: Maglev is technically excellent and economically punishing. The dominant problem is dedicated infrastructure: a maglev line cannot share track with any other rail traffic, so the entire right-of-way has to be built new. Add the gap-tolerance and power-distribution requirements and per-kilometer line cost lands well above conventional high-speed rail. Operating efficiency at very high speeds favors maglev marginally, but not enough to recover the capital gap on most routes.

Dedicated guideway

This is the headline reason. A conventional high-speed line, even when newly built, plugs into the national rail network. A TGV can roll off the LGV onto classic track and serve city-center stations that already exist. A maglev cannot. Every meter of guideway has to be new construction, and every station has to be a purpose-built maglev terminal. The Shanghai Transrapid is a 30 km airport shuttle precisely because that was the only place a short, isolated maglev made economic sense.

Capital cost

Recent published figures for maglev line construction land in the range of USD 100–200 million per kilometer for SCMaglev-class projects when tunneling is involved, versus roughly USD 30–60 million per kilometer for new HSR in similar terrain. These numbers vary wildly by source, route, and labor cost, so treat them as orders of magnitude rather than precise quotes. The Chuo Shinkansen total cost would buy several new HSR lines elsewhere.

Energy and aerodynamics

At 500 km/h, aerodynamic drag dominates everything else. Maglev removes rolling resistance and the pantograph drag, but it does not change the cross-section of the train hitting the air. Energy per passenger-kilometer at 500 km/h is therefore higher than HSR at 320 km/h, even though it is lower than what a wheeled train would consume at the same speed. The “free lift” of EDS is partially offset by the magnetic drag of the same eddy currents that produce lift — a drag that scales with speed in a non-trivial way.

Integration with the existing network

Most high-volume rail corridors in Europe, China, and Japan are already served by HSR that runs on the legacy network at its ends and dedicated track in the middle. Replacing a 1000-km HSR corridor with a maglev would require terminating both ends in greenfield stations and rebuilding the urban access. The political and land-acquisition cost of that is what has historically killed maglev expansion proposals in Germany, the UK, and the US.

Trade-offs and gotchas

Maglev’s central trade-off is the wear-versus-cost curve. Removing rolling contact removes the wear envelope that limits conventional rail above ~350 km/h, but the price of doing so is paid up front in concrete and copper, not over decades in steel-wheel replacement.

The cryogenic system on SCMaglev is the second-order risk. A loss of cryogenic cooling does not crash the train — the on-board magnets stay persistent for hours — but it does take the system out of revenue service until cooling is restored. JR Central has invested heavily in cryostat reliability and in the planned HTS transition specifically because revenue operations cannot tolerate prolonged cooling outages.

Power infrastructure is the third one. A maglev line is essentially a giant, segmented motor; the trackside substations are large, expensive, and have to be located along the entire route. Conventional HSR concentrates power infrastructure at fewer points because the train carries its own motor. The capital and land cost of trackside substations is often missing from headline maglev cost comparisons.

The fourth gotcha is electromagnetic interference. Both EMS and EDS systems generate significant stray fields, and the SCMaglev magnets are particularly strong. JR Central publishes shielding studies showing that in-cabin fields are within ICNIRP limits, but the trackside exclusion zone is wider than for conventional rail, which constrains how close the line can run to homes and farms.

Practical recommendations for transit planners

For transit planners and procurement teams evaluating a maglev pitch, the questions worth asking are blunt.

• Is the route so isolated from the existing network that a non-interoperable mode is acceptable? Airport shuttles and brand-new corridors yes; legacy upgrades almost never.
• Is the demand projection sufficient to amortize a 100–200 million USD per kilometer guideway? Below ~30 million annual passengers, the answer is usually no.
• Does the corridor exceed 600–800 km point-to-point, where the 500 km/h speed actually saves enough time to pull mode share from short-haul aviation?
• Can the right-of-way be acquired greenfield? Brownfield maglev is essentially impossible because of gap-tolerance and station-redesign requirements.
• Has the bidder included realistic substation, cryogenic, and EMI-zone costs in the line item, or are those buried as “balance of system”?

The checklist version:

1. Confirm corridor isolation from legacy rail.
2. Stress-test ridership projections at HSR fare levels, not maglev premium fares.
3. Demand a substations-and-cryogenics line item separately from guideway.
4. Compare against a credible HSR alternative on the same route at the same delivery year.
5. Reserve at least a 30% capex contingency; SCMaglev experience shows maglev cost overruns are systematic, not exceptional.

FAQ

How fast does a maglev train go?

The operating limit depends on the system. The Shanghai Transrapid runs revenue services at 430 km/h, the Linimo and Incheon Airport lines stay under 110 km/h, and the planned SCMaglev Chuo Shinkansen will operate at 505 km/h. The world rail speed record stands at 603 km/h, set by an SCMaglev L0 series in 2015 on the Yamanashi test line. China’s 2021 Qingdao demonstrator targets 600 km/h but has not entered revenue service.

Why does an SCMaglev use rubber wheels?

Electrodynamic suspension only produces lift when the on-board magnets move fast enough to induce strong eddy currents in the sidewall coils — typically above ~150 km/h. Below that speed, the lift force is too small to support the vehicle, so the SCMaglev rides on retractable rubber tires built into the bogie. The wheels stow once the car is levitating and deploy again before each stop. EMS systems like the Transrapid do not need wheels because their electromagnets pull from standstill.

Is maglev more energy-efficient than high-speed rail?

At the same speed, yes; at typical operating speeds, no. Removing rolling contact eliminates a real loss, but aerodynamic drag dominates above 250 km/h and rises with the square of speed. A 500 km/h maglev therefore consumes more energy per passenger-kilometer than a 320 km/h HSR train, even though it consumes less than a wheeled train would at 500 km/h. The honest framing is that maglev buys higher speed for higher energy use, not lower energy use at the same speed.

What is the difference between EMS and EDS maglev?

EMS (electromagnetic suspension) uses conventional DC electromagnets on the car attracted upward to a ferromagnetic rail under the guideway, with a kilohertz-rate control loop holding an 8–10 mm gap. EDS (electrodynamic suspension) uses superconducting magnets on the car repelled from passive figure-eight coils in the guideway sidewalls, with a ~100 mm gap. EMS is mature and runs the Shanghai and Linimo lines; EDS is more tolerant of disturbances but requires cryogenics and minimum-speed wheels.

Why is the Chuo Shinkansen so delayed?

The Tokyo–Nagoya section of the SCMaglev Chuo Shinkansen has been delayed primarily by the Shizuoka Prefecture tunnel section, where local concerns about groundwater impact on the Oi River basin held up construction approval for years. JR Central has stated the original 2027 opening cannot be met. Credible recent reporting points to a partial-service start in the early 2030s, but the operator has not committed to a fixed revised date. Technical risk has been managed; political and environmental approval has been the binding constraint.

Could maglev replace high-speed rail on existing corridors?

In almost no case. Maglev cannot share track with conventional trains, so the existing right-of-way and stations cannot be reused. Replacing a corridor means greenfield construction at 100–200 million USD per kilometer plus new urban terminals. That capital cost is higher than building a parallel new HSR line and far higher than upgrading existing HSR. The realistic role for maglev in 2026 is new high-density corridors above 600 km where short-haul air mode share is the prize, not as a drop-in HSR replacement.

Further reading

• Internal: How GPS atomic clocks work — precision timekeeping (2026)
• Internal: How atomic clocks work — quantum resonance and GPS (2026)
• Internal: Quantum sensing — MRI, SQUID magnetometers explained (2026)
• Internal: Cryo-EM at 1.2-ångström resolution — milestone explained (2026)
• External: Central Japan Railway technical briefs on SCMaglev and the Chuo Shinkansen project page at JR Central.
• External: IEEE Transactions on Magnetics and IEEE Transactions on Transportation Electrification — review papers on linear synchronous motors and maglev guidance control.
• External: Transrapid operator and TR-09 vehicle documentation for EMS reference p