⚡ electrical engineering

Influence of Emissivity Calibration on Thermal History and Process–Property Relationships in Arc-Based Directed Energy Deposition

This study demonstrates that implementing in-situ emissivity calibration for arc-based directed energy deposition significantly improves temperature measurement accuracy, enabling the precise correlation of interpass thermal histories with mechanical properties and revealing that excessive heat accumulation from short interpass times degrades the ultimate tensile strength of ER70S-6 carbon steel walls.

Original authors: Joel Sá, Carlos Zhu, Kaike Monteiro, Fernando Lopes, Cândida Malça, David Andrade

Published 2026-10-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Joel Sá, Carlos Zhu, Kaike Monteiro, Fernando Lopes, Cândida Malça, David Andrade

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Building large metal parts layer by layer is a process that has transformed how engineers think about manufacturing. Known as arc-based directed energy deposition, this technique uses a robotic arm to melt a wire with a welding torch, depositing it bead by bead to form a solid structure. It is a powerful method for creating medium to large components quickly and cheaply, but it carries a hidden risk: heat. As each new layer is added, the metal below retains warmth from the previous pass. If this heat is not managed carefully, it accumulates, changing how the metal cools and solidifies. This thermal history dictates the material's internal structure, which in turn determines whether the final part will be strong enough to hold a bridge or weak enough to fail under stress. To control this, manufacturers need to know the exact temperature of the metal as it cools. They often use infrared cameras, which see heat as light, to watch the process in real time. However, these cameras rely on a setting called emissivity—a measure of how well a surface radiates heat—which can be tricky to pin down on hot, rough metal. If this setting is guessed incorrectly, the temperature reading can be wildly off, leading to poor decisions about how to build the part.

A team of researchers at the University of Coimbra and the Polytechnic Institute of Coimbra set out to solve this problem of guessing. They wanted to know how much a wrong guess about emissivity would distort the picture of the metal's thermal history and, ultimately, how that distortion would affect the understanding of the part's strength. To find out, they built three identical walls made of carbon steel using a robotic welding system. They then applied three different rules for how long to wait between adding layers. Two of the walls were built with a strict temperature limit: the robot would pause and wait until the metal cooled down to either 150 degrees Celsius or 250 degrees Celsius before adding the next layer. The third wall was built with a strict time limit: the robot would wait exactly 30 seconds between layers, regardless of how hot the metal actually was.

The researchers watched these walls grow using an infrared camera, but they also embedded tiny temperature sensors directly into the metal to get the true ground truth. By comparing the camera's view with the sensors' readings, they discovered that the metal's ability to radiate heat was not a fixed number. Instead, it changed as the metal got hotter or cooler. They found that using a single, constant guess for this setting could lead to massive errors in temperature measurement, especially when the metal was very hot. By creating a model that adjusted the setting based on the current temperature, they reduced the error in their measurements to less than four percent. This calibration allowed them to see the true thermal story of each wall.

The results revealed a stark difference between the strategies. The walls built with temperature limits cooled down in a steady, predictable rhythm. The metal had enough time to shed its heat, resulting in a fine, uniform internal structure. These walls showed a specific cooling time, measured as the duration it took to drop from 800 degrees to 500 degrees, of about 22 to 27 seconds. In contrast, the wall built with the 30-second timer suffered from severe heat accumulation. Because the robot did not wait for the metal to cool, the temperature kept rising, eventually stabilizing at a scorching 730 degrees Celsius. The metal never had a chance to cool down properly, and the researchers could not even measure the standard cooling time for the upper layers because the metal never dropped below 500 degrees.

This difference in thermal history had a direct and measurable impact on the metal's strength. The two walls built with temperature control turned out to be remarkably similar and strong, with a maximum pulling strength of about 515 megapascals. Their internal grains were small and tightly packed. The wall built with the 30-second timer, however, was significantly weaker, with a maximum strength of only 470 megapascals. Its internal structure had become coarse and uneven, with large grains that formed because the metal stayed hot for too long. While this weaker wall was slightly more stretchable, it lacked the robust strength of the others. The study concluded that the 30-second timer was a poor strategy for this material, as it led to a loss of quality that a simple time-based rule could not prevent.

Perhaps the most critical finding was about the danger of bad data. The researchers showed that if an engineer had used a standard, uncalibrated guess for the heat-radiating setting, they would have calculated a completely different thermal history for the weak wall. They might have thought the metal was cooling at a normal rate when it was actually sitting at a dangerous, high temperature. This error would have led them to believe the metal's properties were normal, potentially causing them to approve a part that was structurally unsound. The work demonstrates that for this type of manufacturing to be reliable, the infrared camera must be carefully tuned to the changing nature of the hot metal. Without this precise calibration, the link between the heat the machine sees and the strength the part possesses remains broken, leaving engineers guessing about the safety of the structures they build.

Technical Summary: Influence of Emissivity Calibration on Thermal History and Process–Property Relationships in Arc-Based Directed Energy Deposition

Problem Statement
Directed Energy Deposition using arc-based heat sources (DED-Arc), or Wire Arc Additive Manufacturing (WAAM), is a viable route for producing large-scale metallic components. However, the process is highly sensitive to thermal conditions, particularly interpass temperature, which governs cooling rates, microstructural evolution, and final mechanical properties. While infrared thermography is increasingly used for non-contact thermal monitoring, its quantitative accuracy relies heavily on the surface emissivity of the material. In steel DED-Arc, emissivity is often assumed to be a constant value, despite evidence that it varies with temperature, oxidation state, and surface roughness. This study addresses the knowledge gap regarding how such emissivity assumptions propagate errors into critical thermal descriptors (such as $\Delta t_{8/5}$ cooling times and interpass temperatures) and subsequently bias the interpretation of process–property relationships.

Methodology
The research utilized a robotic DED-Arc system to deposit ER70S-6 carbon steel walls onto S460ML substrates. Three distinct interpass control strategies were investigated:

W150: Interpass temperature controlled at a maximum of 150 °C.
W250: Interpass temperature controlled at a maximum of 250 °C.
W30s: A fixed interpass dwell time of 30 seconds, regardless of thermal state.

Thermal monitoring was conducted using a combination of non-contact infrared thermography (FLIR A655sc) and contact-based embedded Type-K thermocouples. The thermocouples served as ground truth for an in-situ emissivity calibration. The authors compared thermographic data against thermocouple readings to identify optimal constant emissivity values and to model the temperature dependence of emissivity. A second-order polynomial regression was developed to describe emissivity as a function of temperature.

The reconstructed thermal histories were used to calculate cooling parameters, specifically the $\Delta t_{8/5}$ interval (time to cool from 800 °C to 500 °C). Following deposition, the walls underwent microstructural characterization (optical microscopy and grain size analysis), microhardness testing, and uniaxial tensile testing (with Digital Image Correlation for strain measurement) to establish correlations between thermal history and mechanical performance.

Key Results

Emissivity Calibration: The study confirmed that emissivity is not constant but varies with temperature. A temperature-dependent model ( $\varepsilon(T)$ ) reduced the mean absolute temperature measurement error to 3.73%, outperforming constant emissivity assumptions (which yielded errors between 4.66% and 5.18%). Sensitivity analysis revealed that using uncalibrated, low emissivity values (e.g., $\varepsilon = 0.4$ ) could lead to massive overestimations of thermal parameters, with errors reaching up to 131% for $\Delta t_{8/5}$ calculations.
Thermal History: The temperature-controlled strategies (W150 and W250) achieved stable, quasi-steady thermal cycles with average $\Delta t_{8/5}$ values of 22.1 s and 26.6 s, respectively. In contrast, the fixed-time strategy (W30s) resulted in severe heat accumulation, driving interpass temperatures to approximately 730 °C. This accumulation prevented the material from cooling below 500 °C after the initial layers, rendering $\Delta t_{8/5}$ calculation impossible for the majority of the build.
Microstructure and Hardness: The stable thermal cycles of W150 and W250 produced fine, homogeneous ferritic microstructures with mean grain sizes of 7.8–8.4 and microhardness values of ~156–159 HV0.5. The W30s condition, characterized by prolonged high-temperature exposure, resulted in significant grain coarsening (mean size ~12.6) and reduced microhardness (148 HV0.5).
Mechanical Properties: The superior thermal control of W150 and W250 yielded higher mechanical performance, with ultimate tensile strengths (UTS) of approximately 515 MPa and yield strengths ( $f_y$ ) around 400 MPa. The W30s wall exhibited a marked reduction in strength (UTS ~470 MPa, $f_y$ ~334 MPa) but slightly higher ductility. All conditions met the ductility requirements for structural steel (EN 1993-1-1), though the W30s condition showed greater scatter in yield strength due to thermal heterogeneity.

Significance and Claims
The paper asserts that accurate emissivity calibration is not merely a measurement refinement but a prerequisite for establishing reliable process–property relationships in DED-Arc. The authors demonstrate that uncalibrated thermography can artificially distort thermal histories, leading to incorrect interpretations of why specific mechanical properties were achieved. For instance, an incorrect emissivity assumption could falsely suggest that a component failed to meet structural standards (e.g., S355 yield strength requirements) due to thermal mismanagement, when the actual thermal history was different.

The study concludes that temperature-based interpass control is a more robust strategy than fixed-time deposition for ensuring consistent thermal conditions and mechanical performance in carbon steel WAAM. Furthermore, it validates the use of infrared thermography for quantitative process monitoring in arc-based additive manufacturing, provided that dynamic, temperature-dependent emissivity calibration is employed to minimize measurement uncertainty and ensure the fidelity of thermal descriptors.

More like this