When a machine builds a metal part layer by layer, a small pool of molten metal changes shape as the heat changes. This study finds that one simple dimension of that pool can reveal temperature well enough to help control the build. Laser-and-wire metal printing can make precise components quickly, but the interaction between the laser beam and metal is highly temperature-sensitive.

Heat input affects the finished part’s shape and overall integrity. The key idea is like judging a campfire by the shape of its flame: the system watches the molten pool's visible shape. Rather than relying on off-axis pyrometry alone, the approach uses a coaxial vision camera integrated into the laser head to monitor the process.

In particular, the study identifies the measured melt pool length as a useful indicator of temperature variations during the deposition process. A predictive model links changes in melt pool length with temperature, while a camera built into the laser head replaces the separate, direction-dependent temperature sensor.

Direct temperature sensors can be noisy, prone to errors, dependent on viewing direction, and difficult to mount. Their readings can also depend on the metal surface itself. So the study uses a compact camera integrated into the laser head. It measures melt pool length with less dependence on deposition direction and less risk that the view will be blocked.

The study then connects the pool’s shape with the temperature where the molten metal begins to solidify, under continuous, multilayer building and realistic heat accumulation. To measure the molten pool, camera images are first cropped to remove irrelevant information.

The remaining images are grouped and averaged so the useful pattern is easier to separate from the background. The images are then compared and simplified using a brightness cutoff. The result is a black-and-white image that captures the pool’s shape for measurement.

The recorded temperature contains three pronounced spikes caused by step changes in laser power. Each spike is followed by an abrupt decline before the signal begins growing again. As more layers are added, both the melt pool length and measured temperature tend to increase.

Reducing laser power significantly affects both quantities. That makes laser power a tool for controlling and optimizing the changing melt pool and temperature. Both melt pool area and length can correlate with temperature, but length consistently has a higher correlation with temperature than area.

The result identifies melt pool length as a reliable indicator of temperature variations, with a strong and consistent relationship between length and the corresponding temperature. The melt pool’s length tracks temperature more reliably than its area across the tested image cutoffs, reaching a strong relationship over a broad range.

That makes length the more useful visual signal for monitoring heat without relying only on a temperature sensor. The study builds a dynamic model of the relationship between melt pool length and temperature for use in control strategies. The model accounts for the delay between a length change and the resulting temperature change.

Validation data produced a fit of eighty-five point one percent. This suggests that the model can predict temperature without a separate temperature sensor, using machine vision instead. The study focuses on a single alloy and explores only a limited range of shapes with sharp angles, so the findings do not yet cover every material or part design.

Future work could test different alloys, more complex geometries, and improved automatic sensing and control to further refine understanding and optimization. The melt pool’s length tracks temperature more reliably than its area, and a predictive model can use that relationship without a separate temperature sensor.

That could make industrial metal printing easier to monitor and control.