A channel-type induction furnace is electrically a short-circuited transformer: the primary coil on a laminated iron core induces a high current in a loop of molten metal that threads the same core, and that loop is the only path through which induced eddy currents can flow [S2][S3]. Remove the liquid metal and you remove the secondary winding, so the unit cannot be cold-started from solid charge like a coreless furnace can [S3][S4].
The practical consequence is that every channel furnace in service carries a permanent "heel" of liquid metal sized to keep the channel and at least part of the bath above the alloy's liquidus. A typical minimum operating heel is reported at 4 to 15 percent of the bath volume, with the lower end tied to minimum safe inductor coverage and the upper end to thermal mass during outages [S1].
Transformer action: the loop IS the secondary
Inside the inductor, a water-cooled (or air-cooled) primary winding is wound on a ring or shell-type laminated iron core, and the molten channel passes through that same core window [S2][S6]. The molten loop behaves as a single-turn secondary, and because it is short-circuited, the induced voltage drives very high current densities through the metal's own resistance, producing Joule heat right inside the channel [S2].
This is the reason the geometry is non-negotiable. A coreless induction furnace, in contrast, has the coil wrapped around the entire crucible and induces current directly into whatever charge sits inside, so it can be started cold, emptied 100 percent by back-tilt, and restarted on a different alloy with no heel [S4]. A channel furnace cannot replicate any of those moves because the magnetic circuit is closed through the metal, not through air [S3][S4].
Why a cold start is physically impossible
Solid metal in the channel is not a continuous conductor of the right geometry: the a/d ratio of chips, borings, or a frozen plug collapses coupling efficiency to near zero, so the primary cannot drive useful current into a discontinuous or non-liquid charge [S7]. Even a fully dense solid bar, if it could be placed in the channel, would not form the closed secondary loop the core expects.
Operationally this forces a defined startup ritual: a separate coreless or crucible furnace is used to melt a starter pool, that liquid is poured into the channel and the bath to form the heel, and only then is the channel coil energized [S3][S8]. The furnace is then committed to a single alloy for the entire campaign, because draining the heel to change chemistry is slow, expensive, and risks a freeze-up that damages the refractory throat [S3].
Thermal and electrical consequences of always having a heel
Because the channel must stay molten 24/7, the inductor itself runs hotter than the metal it delivers: a channel inductor pouring at 1400 °C typically sits at roughly 1500 °C to keep the loop above its freezing point, which shortens refractory life in the throat compared with a coreless coil that only sees bath temperature at the coil face [S4].
The heel also has a mass penalty. A channel furnace cannot fully empty, so a significant fraction of the melt is permanently immobilized: published figures for medium-tonnage iron channel units put the trapped heel on the order of two tonnes, the inventory that never sees a pour [S4]. For an operation running a single alloy nonstop, that locked-up capital is acceptable; for a job shop changing grade every shift, it is a deal-breaker, which is why channel furnaces are positioned as holding or duplexing units downstream of a primary melter rather than as flexible alloy melters [S3][S5].
Channel vs coreless: a criteria-based comparison
Selecting between the two architectures is a direct consequence of the heel requirement, so the decision criteria line up cleanly. The table below distills the engineering trade-off as it actually shows up on a foundry floor. [S4]
Startup state: coreless accepts a cold solid charge and melts it in place; channel requires a pre-melted liquid heel of typically 4 to 15 percent of bath volume and a dedicated melter to supply it [S1][S3]. Empty behavior: coreless tilts out 100 percent and can be grade-changed within a shift; channel cannot fully empty and is committed to one alloy per campaign [S3][S4]. Best fit: coreless for flexible melting of varied scrap and alloys, especially for ductile iron where siphons must be directly heated; channel for continuous holding, superheating, and low-melting-point alloys such as aluminum, zinc, and copper-base melts where the heel is a small fraction of the bath and the thermal mass is welcome [S3][S4][S5]. Energy profile: coreless wins on melting speed and power density, channel wins on holding efficiency because the buried loop sits inside an insulated refractory bath with low standing losses [S3][S4].
Failure modes tied to heel mismanagement
Letting the heel freeze is the catastrophic failure: solid metal in the channel blocks circulation, the inductor overheats, the refractory cracks, and the coil can fail. That is why channel furnaces are never de-energized for a weekend; they are kept on standby with the loop hot, ready for Monday's pour [S4].
Letting slag into the heel is the slow-burn failure. Channel furnaces rely on a clean metal surface for fluxes such as Redux EF40L to do their work, and any emulsified oxide or sulfide that drifts into the throat nucleates and grows on the channel refractory as a solid buildup layer, which in turn throttles the channel cross-section, raises local current density, and reduces electrical efficiency well before anyone notices a temperature problem [S1]. The combination of heel volume, fluxing practice, and channel temperature is therefore the single operating discipline that decides whether a channel furnace lasts two years or twenty.
Where the channel furnace still earns its place
For high-volume single-alloy production, the heel is not a liability, it is the feature. A channel furnace is more energy-efficient at holding than a coreless furnace because the buried loop loses less heat to the coil than a crucible surface does, the electromagnetic pumping in the channel delivers uniform bath temperature, and the unit can be sized into the hundreds of tonnes for cast-iron duplexing without the back-tilt mechanical complexity of an equivalent coreless [S3][S4][S5].
The trigger conditions for specifying a channel furnace in 2026 are therefore narrow and well defined: a continuous, single-alloy campaign; an existing primary melter that can supply the startup heel; an alloy chemistry and pour schedule that tolerates the heel mass; and an operation disciplined enough to keep the loop molten, the slag out, and the fluxing routine tight [S1][S3][S8]. Foundry groups thinking through the trade-off against flexible melting can compare the relevant equipment-side data points in this selection-oriented write-up of coreless vs channel induction designs from CIME and the operating-window discussion in this squeeze-pin firing window reference for HPDC cells; for the power and containment side of the same plant, the MCC section width vs starter bucket count sizing reference lays out the matching electrical-room constraints.
For component-level specifications, see induction furnace, and dry type transformer.