THERMOCOUPLE PROCESS MONITORING FOR ADDITIVE MANUFACTURING BY SHANT KENDERIAN*†, TAIT MCLOUTH*, DHRUV PATEL*, AND JULIAN LOHSER* ABSTR ACT To understand the thermal history of parts manufactured in a laser powder bed fusion system, eight thermocouple sensors were imbedded at key locations with respect to the parts being built. The design comprised eight vertical cylinders 2.54 cm (1 in.) and 1.27 cm (0.5 in.) in diameter and four 2.54 cm (1 in.) horizontal cylinders. The temperature signature collected at the eight locations reveals the time intervals of depositing and melting each layer and the cooling trend associated with the stoppage required for filter cleaning. The temperature profile also reveals a fast rate of heat accumulation at the start of the process. As more layers are melted and the part becomes taller, the dissipation path for heat deposited by the laser increases prior to reaching the build plate. The heat accumulation, therefore, increases rapidly at first, then decreases, plateaus, and then drops slightly toward the end. Distortions due to residual stresses and resultant part separation from the build plate can be deduced from the thermal signature as detected by the thermocouple sensors. This allows the manufacturer to make adjustments or abort the process if necessary. Otherwise, these distortions that render the part a reject are discovered hours or days later upon completion of the additively manufactured part. KEYWORDS: additive manufacturing, process monitoring, thermocouples Introduction With the increased application and acceptance of additively manufactured parts in various industries, including the aero- space industry, a widespread effort has been dedicated toward nondestructive evaluation (NDE) applications for additively manufactured parts during the last decade. Different efforts of NDE techniques have been approached for additively manufactured parts worldwide. The additive manufacturing (AM) process itself went through phases as it was identified first as “3D printing” and “rapid prototyping” before it became known as “additive manufacturing.” However, the strong interest in NDE did not arise until late in 1999 (Griffith et al. 1999 Martinez-Anton et al. 1999). Additionally, although there may have been a number of conference presentations on NDE for AM in the past, the first entire conference session organized by the American Society for Nondestructive Testing (ASNT) on this topic was during the 2016 Annual Conference. Since 2016, there have been one or two sessions held during every ASNT Annual Conference and Research Symposium. In 2018, the European Conference on Nondestructive Testing (ECNDT) organized three sessions on NDE for AM. It is not always easy to determine what should be included in the AM NDE category. For example, it may be argued that post-process NDE on AM parts is nothing more than NDE, if it were not unique to AM. What follows are some observations regarding the NDE efforts that have been reported for AM: Most NDE techniques reported in literature were applied to the part after removal from the AM machine. Therefore, these applications are not used as AM process control, but rather as materials characterization tools to better under- stand the influence of the AM process on materials prop- erties of the product. Some examples to this are computed tomography (CT) (Slotwinski and Garboczi 2014 Smith et al. 2016 Beshears 2017) and neutron radiography (Brooks et al. 2017 Thiede 2018 Bilheux et al. 2016 Rodriguez et al. 2012 Watkins et al. 2013) applications. Some applications are not necessarily “nondestructive” or practical. The micro-CT technique, for example, requires cutting very small coupons from the AM part (Zhou et al. 2015). Some techniques were adapted so that they may potentially be used for AM, such as laser ultrasound (Everton et al. 2015 Cerniglia et al. 2015 Helvajian et al. 2016 Manzo et al. 2016 Klein and Sears 2004 Edwards et al. 2012 Kube 2016). * Space Materials Laboratory, The Aerospace Corp., El Segundo, CA 90245 USA † shant@aero.org Materials Evaluation 80 (4): 30–36 https://doi.org/10.32548/2022.me-04243 ©2022 American Society for Nondestructive Testing ME | TECHPAPER 30 M A T E R I A L S E V A L U A T I O N • A P R I L 2 0 2 2

Many references are published in the form of reports or conference proceedings. Concerning in situ monitoring of the AM part, a relatively small number have been published in refereed journals. A few papers have been published utilizing thermocouples to monitor the AM process. Abdul Aziz et al. (2013) investigated deformation due to residual stresses using strain gauges and thermocouples. Rodriguez et al. (2015) investigated the acqui- sition of absolute surface temperatures using in situ infrared imaging of the melted or solid surfaces layer by layer for an electron-beam AM fabrication system. To acquire absolute values from the infrared images, a calibration procedure was established to determine the solid material’s emissivity and reflected temperature or mean radiant temperature of the build chamber. Experimental validation of the model was per- formed using a thermocouple embedded during fabrication. A temperature difference of ~36 °C was observed between uncorrected infrared temperature data and corrected tem- perature data. Upon validation of the infrared parameters for a melted area, experimentation was conducted to also determine powder emissivity. Delinger et al. (2015) studied the effects of changes in path planning and dwell time between passes on distortion and residual stresses during the AM process. They used thermocouples to monitor the temperatures at the build plate. In this paper, a series of thermocouple sensors are placed under a metal build plate to monitor the in situ temperature received through the part being built. The initial motivation was to evaluate peak temperatures and their dwell times to determine if the environment is safe for attaching ultrasonic transducers at a later time and collect in situ measurements. It became evident, however, that the thermocouple measure- ments revealed several factors that are specific to the param- eters of the AM process and the geometry of the part. Most important is detecting the separation of the part from the build plate under the effect of high residual stresses. The separation disrupts the heat conductivity path and reduces the amount of heat received by the thermocouple element placed directly below the separated component. When it occurs, separation of the part from the build plate and the resultant distortion are undetected because the part is buried under the unused powder. It is typically revealed hours or days later upon the completion of the AM production schedule. The approach proposed in this paper can be used to identify this anomaly, abort the task early, and serve as a time- and cost-saving tool integral to the process. Experimental Design Inconel IN718 samples were prepared on a modified build plate made of 4130 tool steel in a commercially available powder bed selective laser melting (SLM) machine. Process parameters were held constant during the build: a laser power of 180 W scanning speed of 600 mm/s a hatch spacing of 105 μm and an island scanning strategy. These parameters have been used in previous studies and have been shown to produce 99 density (McLouth et al. 2018 Bean et al. 2018). Figure 1 provides a schematic of the machine and operating components. Once these parameters are entered and the part design is loaded, the build process begins when the build plate drops by 30 μm to allow a new layer of powder to fill the gap it leaves above it. The piston under the metal powder reservoir pushes the powder in the reservoir upward, and the re-coater arm spreads a new layer of powder above the bare build plate. Once a fresh layer of powder has been spread across the build plate, the laser melts the current layer in a pattern that corresponds to the horizontal cross section of the parts being built at that specific height. In this experiment, a number of cylinders are manufactured. Each cylinder’s entire cross section is fully melted before the laser moves to the next. The melt process pauses briefly when the laser com- pletes one layer and moves to the side, while the build plate is lowered 30 μm again to allow the re-coater arm to apply a new layer of powder. During this pause, the heat source is removed from the cylinders, allowing them to cool down slightly as heat flows down to the build plate, which acts as a heat sink. There is an additional pause that occurs once every hour associated with the stoppage required for shaking the hopper to clean the filter. To monitor the temperature of the build plate during the build process, Type-K thermocouple sensors are placed at eight locations and different depths in the build plate. A channel is milled on the underside of the plate to feed the thermocouple wires without interfering with the build process. The thermocouple readers are placed outside the enclosed volume of the AM machine. The schematic of the build plate in Figure 2 shows the eight positions of the sensors, the grooves milled for their placement, and channel for their wires. The distance of each sensor from the top surface of the build plate is designated as the “Z position” to differentiate it from the Laser Re-coater arm Metal powder inside hopper Build plate Powder overflow Thermocouple wires Thermocouple reader Build plate piston motor Hopper piston motor Figure 1. Basic operating principle of metal powder bed AM process: the powder supply pushes up, the re-coater arm spreads powder over the build plate, the laser melts a thin layer, and then the build plate drops down, ready for the next coat of fresh powder. A P R I L 2 0 2 2 • M A T E R I A L S E V A L U A T I O N 31