Thermal Monitoring in WAAM: Why Interlayer Temperature Drives Geometry

Run the same WAAM program from the first layer to the fortieth and the bead on layer 30 will not look like the bead on layer 3, even though current, travel speed and wire feed speed never changed. That is not drift in the machine and it is not an inconsistent operator. It is heat that has nowhere left to go as fast as it used to. Watched under a thermal camera, the difference shows up directly: a short, fast-cooling ridge early in the build stretches into a longer, slower-fading one later, because the wall underneath it is starting each new pass warmer than the one before. This article reads that signature the way it appears on video, explains why it happens, and points to the deeper guides on limits, metallurgy and control loops for the follow-up work.

Key takeaways

• The same programmed parameters do not guarantee the same bead, because the wall being built onto changes thermally even when the program does not.
• A thermal camera reads a calibrated temperature field, so the cooling tail behind a fresh bead and the interlayer temperature at the start of the next pass are both measurable, not just visible.
• A lengthening cooling tail from layer to layer is a direct visual sign that heat is accumulating faster than the build is shedding it.
• The wall’s own ability to shed heat drops as it grows, because more of a taller wall’s height sits away from the cool baseplate and surface-to-mass ratio falls.
• A program qualified early in a build encodes an assumption about starting temperature that later layers can silently violate.
• Bead widening from accumulated heat is a slow, many-layer version of the same heat-exceeding-capacity imbalance behind faster defects like burn-through, not the same event on a different clock.
• Monitoring interlayer temperature against layer number turns this into a leading indicator instead of a post-build surprise.

What the thermal video actually shows

In false colour, a fresh bead reads as a bright ridge with a trailing cooling gradient behind it. That tail is not decoration; its length is a direct readout of how much energy the surrounding material can currently absorb. Localised bright spots gather at corners and direction changes, where the torch dwells fractionally longer. Because the camera reports a real temperature value at every pixel, the footage can be compared quantitatively from one layer to the next, which a visible-light clip cannot do.

Watch two things across layers, not within one pass: the peak temperature just behind the arc, and the temperature of the surface the torch is about to deposit onto. That second number — the interlayer temperature — tends to climb through a build even when the first stays roughly constant, and it is what ties the visual signature to the geometry question below.

Why layer 3 and layer 30 are not the same problem

A WAAM wall does not present a fixed thermal environment to the torch. Early on, each new layer sits close to a large, relatively cool baseplate that conducts heat away efficiently, letting the deposit cool and solidify quickly. As the wall grows taller, that path lengthens: heat from a new layer has to travel through material that is itself already warm before reaching anything genuinely cool. Surface-to-volume ratio falls too, so there is proportionally less exposed area to radiate and convect heat away relative to how much thermal mass has built up.

The consequence is interlayer temperature — the previously deposited layer’s temperature at the moment the next pass starts — climbing through the build even when nothing in the program changes. A hotter starting surface keeps the next melt pool fluid for longer before it freezes, letting it spread further under its own weight before solidifying. Programmed current, travel speed and wire feed speed set how much energy goes in and how fast the torch moves; they say nothing about the temperature of the material receiving that energy, and that omission is exactly what separates layer 3 from layer 30.

The dwell, travel-speed and wire-feed trims that correct for this as the wall climbs are covered in WAAM layer-upon-layer deposition: what changes as the wall grows; this article stays on the thermal signature and what to do with the number once it is measured.

From visual signature to a build decision

A rising cooling-tail length or climbing interlayer temperature is only useful if it changes what happens next:

1. Confirm it is a trend, not one warm layer. A single hotter reading can come from a paused robot cycle. A trend climbing over several consecutive layers, independent of any one event, is what matters.
2. Compare it against the build’s qualified limit, not a generic figure — the limit is alloy- and geometry-specific and belongs in the procedure.
3. Act before the limit is reached, through a dwell period, active cooling, or a parameter change for the remaining layers, not after a layer has already deposited outside range.

Reading footage after the build is finished tells you what happened but not in time to change it. Reading the same signal live is what turns thermal imaging from a diagnostic into a control input — the subject of interlayer temperature control loop design.

The metallurgical side of the same signature

Geometry is what a camera sees directly, but the temperature history behind a widening bead also changes what the deposited metal becomes. A layer that stays warmer for longer cools more slowly through its transformation range, and each layer underneath gets reheated again by every subsequent pass without fully re-melting. On carbon and low-alloy steel builds, that repeated reheating can temper earlier microstructure and coarsen grain locally. The full mechanism, alloy by alloy, is in why carbon steel WAAM walls slump; the same rising-temperature trend a camera flags for geometry is what a metallurgist would want flagged for microstructure, from the same footage.

Where this fits against other defect signatures

Heat outrunning what the surrounding material can absorb is a recurring cause across weld quality problems, but timescale changes what catches it. A single pass losing root support develops in a fraction of a second and needs a high-speed camera resolving individual frames. A WAAM wall’s bead widening from accumulated heat develops over many minutes and many layers, so the useful signal is a temperature trend logged against layer number, not a frame-by-frame sequence. Lack-of-fusion imperfections — ISO 6520-1 reference 401 — are usually a too-little-heat problem: arc energy below the qualified window, travel speed too high, or slag between runs. Heat accumulation can reach the same defect differently: a pool spreading laterally on an over-warm substrate can run over the wall’s edges without wetting them, leaving an unfused margin under metal that looks fully melted from above.

From this video to a monitored build

Watching the footage above builds intuition for what heat accumulation looks like on real hardware, but a production cell needs the same signal running continuously, not reviewed after the fact. HeatCam IR-S is sized for WAAM and other embedded deposition cells, reading layer surface temperature live, pass by pass. HeatCore AI turns that stream into a per-layer record and a live trend line instead of a clip someone watches back. For camera placement and field-of-view choices, see WAAM monitoring camera setup, and for the broader case for continuous monitoring on additive cells, WAAM additive manufacturing monitoring.

FAQ

Why does layer 30 of a WAAM wall look different from layer 3 under the same programmed parameters? Because the parameters are not the only thing changing. Current, travel speed and wire feed speed can be held identical from the first layer to the last, but the wall they are deposited onto is not the same object each time: it carries more accumulated heat and has less relative surface area to shed it through. The bead that results is a function of programmed energy and existing thermal state together, and only one of those two stays fixed.

What does a WAAM thermal camera actually show that a visible-light camera does not? A radiometric temperature field rather than a picture of brightness. A visible-light view shows arc glare and a glowing pool that is hard to read quantitatively; a thermal camera reports a calibrated temperature value at every point, so the trailing cooling gradient behind the bead, the interlayer temperature at the start of the next pass, and the extent of the hot region at direction changes are all measurable, layer over layer, on the same scale.

Does the cooling tail visible in a WAAM thermal video change length as the wall gets taller? Typically yes, and that change is one of the more direct visual signals of heat accumulation. A short, fast-fading tail on an early layer means the fresh bead is losing heat quickly into cool surrounding material. A tail that stretches further behind the torch on a later layer, at the same travel speed, means the material ahead of it started warmer and stays above the freezing range for longer.

Can a WAAM program that runs correctly on layer 3 produce a defect by layer 30? Yes, and this is the practical reason interlayer temperature is monitored rather than assumed constant. A program qualified and proven early in a build encodes an implicit assumption about the temperature the material sits at when each pass starts. If nothing intervenes to slow that climb, the same instructions get executed against a hotter and hotter starting condition, and a wall that met tolerance early on can drift out of it well before completion.

Why does a WAAM wall lose the ability to shed heat as it grows, when the material and process are unchanged? Because the geometry of the heat sink changes even though the alloy does not. Early layers sit close to a large, cool baseplate that conducts heat away efficiently. As the wall grows taller, more of each new layer’s heat has to travel through material that is itself already warm, and the ratio of exposed surface to accumulated thermal mass drops. The physical process removing heat has less capacity to do its job, independent of anything the operator changed.

Is a widening bead late in a WAAM build the same signature as burn-through in a single weld pass? They share a root cause category, heat exceeding what the surrounding material can absorb, but not a timescale or a mechanism. Burn-through develops within one pass, over a fraction of a second, as a joint’s root loses support. Bead widening from WAAM heat accumulation develops over many passes, over minutes, as the whole build’s thermal baseline rises. Both are worth watching; neither view substitutes for the other.

What should a quality engineer actually do with a rising interlayer temperature trend on a WAAM build? Treat the trend as the leading indicator rather than waiting for a geometric or metallurgical result to confirm it. That means logging interlayer temperature against layer number so the climb is visible before it becomes a defect, and either building in a dwell or forced-cooling step, or adjusting the deposition parameters for later layers, once the trend crosses whatever limit the build’s procedure qualification set. What that limit should be, and how to hold to it automatically, is procedure- and alloy-specific and covered in the deeper guides below.

Therness solution: Welding camera — Thermal and high-speed imaging that watches the arc, the pool and the cooling seam while the weld is made.

Written by Luca Santoro , founder