Correction of non-imaging thermal measuring devices

Abstract

Additive manufacturing process comprising: Recording (900) temperature data describing temperature changes across a surface of a material (104; 256) during an additive manufacturing process in which the material (104; 256) is heated by a heat source (100; 250), the temperature data being generated using a non-imaging optical sensor (109, 111, 113, 122; 259) with a defined field of view (400, 500, 600, 600'); Estimating (906) the size of a hotspot (604) corresponding to the hottest region formed by the heat source (100; 250) on the surface, wherein the size of the hotspot (604) is estimated to be equal to the size of an energy beam (101; 254) emitted by the heat source (100; 250); Estimating (908) the size of a heated zone (505) corresponding to a locus of points within the field of view (400, 500, 600, 600') that contribute to the temperature data; and Correcting (912) the temperature data based on the estimated size of the hotspot (604) and the estimated size of the heated zone (505).

Description

BACKGROUND OF THE INVENTION

[0001] Additive manufacturing, or the sequential construction or creation of a part through a combination of material addition and energy input, can take many different forms. Additive manufacturing can be performed using any of several different processes that relate to the formation of a three-dimensional part in almost any shape. What the various processes have in common is the sintering, curing, or melting of a liquid, powder, or granular raw material layer by layer using ultraviolet light, a high-power laser, or an electron beam, respectively. Despite advances in additive manufacturing, there is a need for improved processes and systems related to additive manufacturing.

[0002] US 2017 / 0217104 A1 discloses controlling the heating of a surface by monitoring the temperature of a plurality of zones of the surface to output at least one temperature reading from each of the plurality of zones. The temperature readings are modulated in response to a pattern arranged over a portion of the plurality of zones. The energy supplied to each of the plurality of zones is controlled based on the modulated temperature readings to maintain a substantially homogeneous temperature distribution across the surface.

[0003] US 2017 / 0266762 A1 discloses a method for eliminating errors in additive manufacturing inherent in thermal measurement devices, allowing the presence of defects in a product manufacturing process to be identified and addressed with greater precision. Instead of monitoring a grid of discrete locations on the build plane with a temperature sensor, the intensity, duration, and in some cases, position of each scan are recorded to characterize one or more build processes.

[0004] US 2016 / 0185048 A1 discloses a multi-sensor quality inference system for additive manufacturing, as well as a quality system capable of detecting and addressing three quality issues: i) process anomalies or extreme unpredictable events that do not correlate with process inputs; ii) process variations or differences between desired process parameters and actual operating conditions; and iii) material structure and properties or the quality of the resulting material produced by the additive manufacturing process. BRIEF DESCRIPTION OF THE INVENTION

[0005] The invention relates to an additive manufacturing method according to claim 1 and an additive manufacturing system according to claim 8. The described embodiments relate to a large subset of additive manufacturing that uses an energy source in the form of a moving zone of intense thermal energy. Such energy sources are referred to below as heat sources. When the thermal energy results in a physical melting of the added material, these processes are generally referred to as welding processes. In welding processes, the material, which is added in a stepwise and sequential manner, is melted by the heat source, similar to a fusion weld.When the added material is in the form of powder layers, after each successive addition of an additional layer of powder material to the part under manufacture, the heat source melts the added additional powder into weld areas of the powder layer, creating a migrating molten zone, hereinafter referred to as a weld pool, so that after solidification, these become part of the previously successively added, melted and solidified layers below the new layer, which are part of the part under manufacture.

[0006] An additive manufacturing method is disclosed, comprising: recording temperature data describing temperature changes across a surface of a material during an additive manufacturing process in which the material is heated by a heat source, the temperature data being generated using a non-imaging optical sensor with a defined field of view, estimating the size of a hotspot corresponding to the hottest area formed at the surface by the heat source, estimating the size of a heated zone corresponding to a locus of points within the field of view that contribute to the temperature data, and correcting the temperature data based on the estimated size of the hotspot and the estimated size of the heated zone.

[0007] An additive manufacturing system is disclosed, comprising: a heat source, a non-imaging optical sensor configured to measure temperatures across a surface of a material during an additive manufacturing process in which the material is heated by the heat source, the optical sensor having a defined field of view, and at least one processor configured to receive temperature data from the optical sensor, the at least one processor further configured to: estimate the size of a hotspot corresponding to the hottest area formed by the heat source on the surface, estimate the size of a heated zone corresponding to a locus of points within the field of view that contribute to the temperature data, and correct the temperature data based on the estimated size of the hotspot and the estimated size of the heated zone. BRIEF DESCRIPTION OF THE CHARACTERS

[0008] The following detailed description, taken in conjunction with the accompanying figures, in which like reference numerals designate like structural elements, will assist in understanding the disclosure, wherein: Fig. 1 shows a schematic representation of a system with a strong heat source, in this specific case a laser beam, Fig. 2 shows a schematic representation of a system with a strong heat source, in this specific case an electron beam, Fig. 3 shows a representation of a typical field of view for a pyrometer, Fig. 4 shows an example of a field of view of a pyrometer in an additive manufacturing environment, Fig. Figure 5 shows a composite schematic representation of a field of view of an Eulerian pyrometer, including a trajectory of a thermally affected and molten material formed over time by means of a moving heat source. Fig. 6A shows a snapshot of a field of view of a Lagrange pyrometer and a thermally affected zone, Fig. Figure 6B shows another snapshot of a Lagrange pyrometer field of view observing a larger thermally affected zone, Fig. 7A - Fig. 7C show examples of scan patterns that can be traversed by a heat source, Fig. 8 shows a curve of a raw signal from an optical sensor, Fig. 9 shows a flowchart of a method for correcting the observed temperature based on an estimated size of a heated zone, Fig. 10 shows a process in which data recorded by an optical sensor, such as a non-imaging photodetector, can be processed to characterize an additive manufacturing build process, Fig. 11A - Fig. 11D show illustrative representations that illustrate how multiple scans can contribute to the energy introduced into individual raster areas, Fig. 12A - Fig. 12F illustrate how a grid can be dynamically created to characterize and control an additive manufacturing process, and Fig. 13 shows an example of a closed control loop having a feedback control loop for establishing and maintaining feedback control of an additive manufacturing process. DETAILED DESCRIPTION

[0009] The present disclosure relates to the correction of thermal measurements in additive manufacturing processes. The thermal measurements are performed during additive manufacturing, in particular, using a non-imaging optical sensor such as a pyrometer. Non-imaging sensors are typically limited to a single sensor element, e.g., a single photodiode. The sensor element can only output a signal corresponding to the temperature in an area toward which the sensor element is directed, but cannot itself generate an image of the measured area. The embodiments described herein can be applied to both Lagrangian sensors and Eulerian sensors. Eulerian sensors operate within an Eulerian reference frame by observing values associated with specific points in space at specific time intervals.An Eulerian sensor is typically mounted facing a surface to be measured. This is comparable to sitting on a riverbank and observing the river flowing by. In contrast, a Lagrangian reference frame considers physical quantities that are related to the transport process along streamlines in the flow field or in the medium itself, i.e., with the transport process "in motion." Therefore, a Lagrangian sensor can be attached to a moving heat source or receive light irradiation that follows the movement of the heat source. This is equivalent to sitting in a boat and observing the river while moving with the current.

[0010] Temperature and other data can be collected to make adjustments to an additive manufacturing process during the process or to analyze the process results afterward. Thermal measurement can be performed using pyrometers, charge-coupled device (CCD) arrays, thermal imaging cameras, or other optical devices. Non-imaging devices such as photodiodes and pyrometers are not as accurate as devices that produce a one-dimensional (1D) or two-dimensional (2D) image of the measurement area. An image is any pattern of light projected onto such a sensor array using a series of optics such as lenses, mirrors, gratings, etc. Non-imaging devices therefore often require calibration or temperature correction.Accordingly, embodiments of the present invention provide methods and systems for improving the accuracy of thermal measurements taken with non-imaging devices. This text has been copied by the. ADDITIVE MANUFACTURING SYSTEM WITH OPTICAL SENSORS

[0011] Fig. Figure 1 shows a schematic representation of an additive manufacturing system that includes both Lagrangian and Eulerian sensors. The system includes a heat source (100; 250), in this case a laser. The beam 101 emitted by the heat source (100; 250) originates from the laser head and passes through a partially reflective optic 102. The optic 102 is designed to be essentially fully transmissive at the specific wavelength at which the laser is operated and reflective at other wavelengths of light. Generally, the laser wavelength is in the infrared or near-infrared range, or typically has wavelengths of (100; 250) nm or greater. The laser may include a deflection head 103 consisting of x- and y-positioning galvanometers and a converging lens, such as an F-theta lens.Consequently, the beam 101 is focused and impinges on a workpiece 104 at a specified position 105, creating a melting zone on the workpiece 104. The heated zone causes light radiation 106 to be emitted isotropically and uniformly over a large solid angle. A portion of the light radiation 106 is returned by the deflection head 103 and reflected by the partially reflective optics 102.

[0012] The reflected light beam 107 then passes through a series of analytical Lagrangian instruments 109, 111, and 113. A beam splitter 108 sends a portion of the beam to a photodiode 109. The photodiode 109 may be configured to capture a frequency range at a sufficiently high speed and recording rate to detect possible anomalies occurring during a deposition process, i.e., sudden deviations from an average or mean intensity level. The remaining portion of the reflected light beam 107 is then directed to another beam splitter 110, and a portion of the beam is captured by a pyrometer 111. The pyrometer 111 may integrate the signal over time to assign a temperature to the captured light.The signal should be corrected for the various optical attenuations caused by beam splitting, as well as correction for the remote position of the melt zone 105 on the workpiece 104, which resulted in the light emission 106, a portion 107 of which was detected. Finally, the remaining portion of the reflected light beam 107 is directed by a mirror 112 into a high-speed optical image sensor 113, which may be a camera or other type of linear or area CCD array or other imaging array. The optical image sensor 113 captures a 1D or 2D image that correlates with the size of the melt zone. By using a relatively low-resolution sensor 113, the sensor 113 can be configured to record data at extremely high frame rates, allowing the sensor 113 to detect highly transient temperature profiles that occur during a build process.

[0013] In addition to the various sensors in the Lagrangian reference frame, the system can take measurements in an Eulerian reference frame using instruments 114 and 122. Instrument 114 is a stationary pyrometer 114 that takes an independent temperature measurement. The field of view (FOV) 115 of the stationary Eulerian pyrometer 114 is suitably selected to match the characteristic dimension of the melt zone 105 present on the workpiece 104, which is created by the focused laser beam 101 at the specific position where the deflection head 103 has directed and focused the beam 101. Instrument 122 is an Eulerian photodiode 122 with an FOV 124, which can be configured to detect temperature changes in virtually any area of the upper surface of the workpiece 104.In some embodiments, the pyrometer 114 may be configured to provide calibration information to the Euler photodiode 122, allowing the Euler photodiode 122 to accurately sense the temperature at any point on the top surface of the workpiece 104.

[0014] In addition to the sensors mentioned above, additional sensors may be added to enhance the measurements made by the additive manufacturing sensor system. The device 116 may be part of a mechanism that distributes powder layers over a top surface of the workpiece 104. The device 116 may include a contact sensor 118 configured to measure any disturbances in the powder deposition, such as vibrations or shocks that may result in uneven powder deposition. In some embodiments, detecting vibration of the device 116 may be used to accurately predict changes in the powder layer. The illustrated sensor system may also include an acoustic sensor 120.The acoustic sensor 120 may be disposed along one side of the build platform, whereby the acoustic sensor 120 may be configured to detect the formation of microcracks in the workpiece 104 during the build of the workpiece 104. The acoustic sensor 120 may be calibrated to determine various characteristics of microcracking in the workpiece 104. Microcracks can be caused by many factors, particularly improper cooling rates.

[0015] Fig. Figure 2 shows a schematic representation of an additive manufacturing system in which an electron beam forms the heat source. The system of Fig. 2 includes an electron gun 250 that generates an electron beam 251, which is focused by an electromagnetic focusing system 252 and then deflected by an electromagnetic deflection system 253, forming a finely focused and deflected electron beam 254 that creates a hot beam-material interaction zone 255 on a workpiece 256. Light energy 258 is emitted, which can be detected by an array of sensors 259, each equipped with its own FOV 260, which can be localized to the interaction zone 255 or encompass the entire build area of the workpiece 256. At least some of the sensors 259 can be Lagrange sensors with their own optical tracking and deflection system for tracking the movement of the electron beam 254 across the build area of the workpiece 256.Regardless of whether the sensors 259 have optical tracking or not, the sensors 259 can consist of pyrometers, photodiodes, spectrometers, and high-speed or low-speed cameras operating in the visible or infrared (IR) spectral range. The sensors 259 can also be sensors that combine a number of physical measurement types, such as a laser ultrasonic sensor that can actively excite or "ping" the deposition with a laser beam and subsequently measure the resulting ultrasonic waves, or "vibration," of the structure with a laser interferometer to measure or predict mechanical properties or the mechanical integrity of the deposition during buildup. Furthermore, contact sensors 262 can be located on a mechanical device 261 that distributes the powder. The contact sensors 262 can be accelerometers, vibration sensors, etc.Finally, other types of sensors 263 may also be present. These may include contact sensors such as thermocouples for measuring macrothermal fields, or acoustic emission sensors that can detect cracks and other metallurgical phenomena in the deposition during buildup. In some embodiments, one or more thermocouples may be used to calibrate the temperature data acquired by the sensors 259. The information provided in conjunction with the . Fig. 1 and Fig. The sensors described in section 2 can be used in the manner described to characterize the quality of any additive manufacturing process with sequential material build-up. THE HEATED ZONE AND THE HOTSPOT VIEWED FROM THE FIELD OF VIEW OF AN OPTICAL SENSOR

[0016] Fig. Figure 3 shows a representation of a typical FOV 300 of a pyrometer used to measure the temperature of a surface 301. The FOV 300 is kept as small as possible compared to the area of the surface 301. In general, a smaller FOV provides a better indication of the temperature at a specific location on the surface.

[0017] Fig. Figure 4 shows an example of a 400 FOV of a pyrometer in an additive manufacturing environment. In contrast to Fig. 3 is the FOV of Fig. 4 is significantly larger than the object being measured. In this case, the object is defined as a molten area, along with all other material within the FOV, that is hot enough to contribute to a signal, i.e., hot enough to radiate at a frequency or over a range of frequencies that can be detected by the pyrometer, and thus contribute to the temperature data generated by the pyrometer. If the temperature field within the FOV 400 is varied and both location-dependent and time-dependent, it can be assumed that each individual area element contributes to the averaged temperature, which is detected in proportion to the area's share of the total area of the pyrometer's FOV. Fig. Figure 4 shows a schematic representation of a small differential area. The FOV 400 can comprise a plurality of smaller radiating areas dA. Generally, these areas are distributed within the pyrometer's FOV at a specific radius R from the center of the FOV and a specific angular orientation θ. Furthermore, each individual area can have a different temperature-dependent emissivity.

[0018] Fig. Figure 5 shows a composite schematic of an FOV 500 of an Eulerian pyrometer while monitoring part of an additive manufacturing process. The FOV 500 of the pyrometer is depicted as a circular FOV with a radius R 502. Currently, a heat source is positioned so that its energy is directed into a zone 501. Because the energy is focused on zone 501, this forms the hottest area within the FOV 500 and can therefore be referred to as a "hotspot." Since the sensor is an Eulerian sensor, as the heat source is moved in direction 506 along a path 504, the hotspot can be observed to move over time. The heat source melts metal powder located along path 504, rapidly liquefying the metal powder, and the liquefied powder can then cool and solidify to become part of a workpiece to be manufactured. Fig. 5 also shows the heated zones 505, which are zones of the build plane that receive a sufficient amount of heat to contribute to the pyrometer's sensor readings for four consecutive time periods. The path 504 can vary in size, area, and location where it crosses the FOV 500 depending on the parameters of the heat source and the material properties of the metal powder. The heat source is not an instantaneous heat source, meaning it does not turn on immediately and does not immediately release a finite amount of heat. Rather, the heat source is a moving, continuous heat source. As the heat source is moved through the FOV while traversing the path 504, the temperature in various regions within the FOV continuously increases and decreases.Therefore, the observed temperature is to be interpreted as a time-integrated mean of the time-dependent thermal behavior of the hot and cold zones, each weighted by their area shares.

[0019] Fig. Figure 6A represents a snapshot of an FOV 600 of a pyrometer operating in a Lagrangian reference frame and shows a thermally affected zone 602 from which the pyrometer can detect radiated heat. Fig. Figure 6A depicts an instant when a heat source moves toward 606 and directs its energy into a zone 604. Zone 604 corresponds to a hot spot, and the energy delivered by the heat source to zone 604 liquefies the metal powder within and adjacent to zone 604, forming a molten pool.

[0020] Because the sensor is a Lagrangian sensor, there is no significant change in the position of hotspot 604 within the current FOV 600. Surrounding hotspot 604 are cooler regions that are still warm enough to be detected by the pyrometer, but are not necessarily melted. The locus of the points contributing to the signal generated by the pyrometer thus includes the hotspot and the cooler regions and defines what is referred to herein as the "heated zone" and depicted as the thermally affected zone 602. Beyond the heated zone are regions that are too cool to be detected by the pyrometer, including regions where the heated material is at ambient temperature and heated regions where the material is slightly above ambient temperature.

[0021] Fig. Figure 6B shows a snapshot of a 600' FOV for the pyrometer of Fig. 6A. The FOV 600' includes a hotspot 604', which is located at the same position as the hotspot 604, since the sensor's reference frame follows the movement of the heat source. Compared to Fig. 6A, the thermally affected zone 602' is significantly larger than the thermally affected zone 602. There are several reasons why the thermally affected region becomes larger. These reasons are described in more detail in the next section, but to summarize, there are factors that contribute to the size of the thermally affected region. These factors fall into two categories: one related to the operating parameters or properties of the heat source, and the second related to the properties of the material(s), specifically the material being heated. Some ways to increase the size of the thermally affected zone include increasing the power output or travel speed of the heat source, or switching to a material with a higher melting point.In addition to changing the size of the thermally affected zone, some factors can also affect the overall shape or heat distribution within the thermally affected area. For example, it is possible to maintain the same shape and position, even the same size, but obtain a different temperature distribution within the thermally affected area. Conventional temperature conversion methods do not account for these changes in the thermally affected zone. The resulting observed temperature values can therefore be extremely inaccurate.

[0022] Fig. 7A - Fig. 7C show examples of scan patterns along which a heat source can be moved. Fig. Figure 7A shows an exemplary scan pattern 706. As illustrated, the FOV 700 may cover a relatively small portion of the scan pattern 706, allowing only the heating and cooling within the FOV 700 to be accurately quantified. As previously mentioned, other scan patterns are also conceivable and may include narrower or wider scan patterns executed at different speeds and different output powers. The scan rate, power, and scan pattern each influence how much energy is dissipated during the additive manufacturing process.

[0023] Fig. Figure 7B shows how the scan pattern can be divided into a checkerboard pattern, with each square scanned sequentially from left to right and top to bottom. The numbers in each square indicate the order in which the squares of the checkerboard are scanned. Fig. Figure 7C shows the same checkerboard pattern, but now the scan order for each checkerboard square is randomized. Regardless of the specific scan pattern or scanning strategy used, the scanning process can involve many short, discrete scan lengths with a start and stop and a trajectory length.

[0024] If a stationary / Eulerian optical sensor is used to create a scanning pattern similar to that of the Fig. 7B or Fig. 7C, the data collected by the sensor may represent multiple scans. For example, the optical sensor may produce a time-varying raw signal, with multiple peaks in the signal, each peak representing a particular scan along a particular trajectory spanning the sensor FOV. The raw signal, which may be a voltage, varies in direct proportion to the intensity of the light emitted by the material and is a function of the trajectory length. At the beginning of each scan, the signal may be small or even zero because the heat source has just been turned on or has just been directed at a new or unheated area of the material. As the scan progresses, the material generally gets hotter and emits more light, so the signal increases.Of course, the raw signal exhibits a natural fluctuation and scatter as the light intensity changes throughout the process, due to the very chaotic nature of the interactions between heat source and powder, as well as the chaotic movement of the molten material and the changing geometric view factor from this small hotspot to the optical sensor. ESTIMATION OF THE HEATED ZONE AND THERMAL FIELD

[0025] Factors contributing to the size and shape of the heated zone include various operating parameters of the heat source and various physical properties of the material being heated. Fig. 6A, the heated zone 602 is shown as an oval. However, the actual shape of the heated zone can vary depending on the operating parameters of the heat source, such as scan pattern, traverse speed, and output power, as well as the physical properties of the material being heated. The size and shape of the heated zone changes according to temporal or spatial variations in the operating parameters. As a simple example of a temporal change, increasing the traverse speed will increase the size of the heated zone because the heat source performs more scans across the FOV, thereby reducing the amount of time certain sections of the material have before receiving additional energy from the heat source. Conversely, decreasing the traverse speed can reduce the size of the heated zone. Increasing the output power tends to increase the size of the heated zone.Increasing the output power can also reduce or compensate for the effects of changes in traverse speed. As an example of a spatial variation, narrower, i.e., denser scan lines can increase the overall size of the heated zone because the heat is more concentrated, whereas more widely spaced scan lines can decrease the size because the heating is more diffuse. It should be noted that many additive manufacturing processes can exhibit large spatial and / or temporal variations in order to correctly form features of different sizes and shapes. For this reason, a sensor that can accurately monitor energy input over a wide range of variation is crucial to successfully identify potential problems in an additive manufacturing process.

[0026] Regarding the effects of traversing speed, it is not always true that a lower speed results in a smaller heated zone or a higher speed results in a larger heated zone. The effects may depend, for example, on whether the material hardening mechanism is dominated by either capillary forces or inertial forces. To make this decision, a capillary time and a Rayleigh time can be calculated based on the physical properties of the heated material. The capillary time is the time required for an interface to return to its equilibrium shape after a disturbance and is given by: tCAP=ηLσ

[0027] Where η is the viscosity, L is the characteristic length, and σ is the surface tension. Similarly, the Rayleigh time is defined as the time required for the interfacial relationship to be satisfied under the influence of inertial and surface tension forces: tRAY=ρL3σ

[0028] Where ρ is the density, L is the characteristic length, and σ is the surface tension. If the capillary time is much shorter than the Rayleigh time, the solidification process after melting is dominated by inertial effects, counteracted by surface tension, as opposed to viscous effects. The jet interaction time can also be considered and can be calculated for different traversing speeds. The jet interaction time decreases with increasing traversing speed and is of the same order of magnitude as the Rayleigh time at low traversing speed values (e.g., 400 mm / s or less).

[0029] At higher traversing speeds (e.g., 600 mm / s or more), the beam interaction time becomes shorter than the Rayleigh time and much shorter than the heat conduction time. This reduces the energy density per unit length along the heat source path, and the average temperature of the melt pool decreases. Therefore, with increasing traversing speed, the average signal measured by the sensor may decrease at the point where the characteristic beam interaction time becomes shorter than the Rayleigh time. Therefore, higher speeds may initially result in larger heated zones, but increasing the speed beyond a certain point causes the heated zone to shrink. The speed at which this reversal occurs depends on the physical properties of the material and the operating parameters of the heat source. Higher power may increase the reversal point to, for example,800 mm / s and further reduce the rate at which the size of the heated zone decreases with further increasing speed.

[0030] Another operating parameter that can influence the size and shape of the heated zone is the angular orientation of the heat source relative to the surface of the material being heated. If the heat source is perpendicular to the surface, the hotspot can be assumed to be perfectly circular. If the heat source is at an angle to the surface, the hotspot will deform, for example, into an oval. When the hotspot deforms in this way, the heat is distributed over a larger surface area but is less concentrated per unit area, so the overall size of the heated zone tends to be smaller.

[0031] The size of the heated zone can be estimated based on the contributing factors, e.g., by inputting the operating parameters of the heat source along with other relevant information such as physical properties of the material to be heated (e.g., emissivity, powder density, thermal conductivity, etc.) into a thermal model. The thermal model can be a static model based, for example, on experimental observations of the various heated zones generated due to different combinations of operating parameters and materials. In some embodiments, the thermal model is updated based on results acquired during actual manufacturing. Therefore, the thermal model can be updated, for example, using a learning algorithm or a neural network that was first trained with experimental data.The heated zone can also be estimated using a reduced-order model that uses fewer inputs or coarser approximations than the thermal model, or using an analytical model that outputs values of the heated zone based on a solution to a system of equations describing the heated zone.

[0032] In some embodiments, the thermal model, the reduced-order model, or the analytical model is used to estimate a thermal field generated in the material and then calculate the size of the heated zone as a function of the thermal field. The thermal field describes the temperature distribution for the portion of the material that lies within the sensor's FOV. The model can estimate the thermal field using the same contributing factors, since temperatures within the thermal field change according to changes in the heated zone. Thus, the thermal field can be calculated based on factors such as the size of the heat source beam, the thermal constraints, other operating parameters of the heat source (scan pattern, traverse speed, output power, orientation, etc.), and the physical properties of the material (e.g.,Emissivity, density, or thermal conductivity of a powder layer on the surface of the workpiece) can be estimated. Some of these factors can change over time and be applied differently along the surface of the workpiece. Accordingly, the thermal field also changes and can be estimated as a function of both space (e.g., a two-dimensional coordinate system) and time.

[0033] The thermal field can be expressed as a two-dimensional matrix of temperature values or scalar quantities representative of the temperature, e.g., normalized temperature values. In some embodiments, the thermal field is estimated by entering the contributing factors into a lookup table, which is referenced in real time during additive manufacturing. An advantage of using a lookup table is that time-consuming calculations do not need to be performed. The table can be stored locally in the system. Alternatively, the table can be stored on an external computer or storage medium. If the thermal field estimation does not need to be performed in real time, a more time-consuming estimation method can be used, e.g., a more accurate thermal model running on a neural network.In some embodiments, the thermal field is estimated as a three-dimensional volume mapped to a two-dimensional matrix, where the matrix entries correspond to points located within a region defined by the intersection between the volume and the build plane of the workpiece. CONVERTING A RAW SIGNAL TO AN OBSERVED TEMPERATURE

[0034] As an example, a method is described in which a raw signal from an optical sensor is converted into temperature values, which are then corrected based on the emissivity and then again based on the size of the heated zone. The following briefly explains how a raw signal from an optical sensor can be converted into temperature values.

[0035] Fig. Figure 8 shows a trace of a raw temperature signal 800 generated by a pyrometer during an actual additive manufacturing process involving sintering of an Inconel alloy, specifically the nickel and chromium-containing alloy IN-718. As shown in Fig. 8, the raw data curve 800 is subjected to signal processing before temperature conversion to form a smoothed curve 801. Curves 800 and 801 represent the time course of voltage signals. The time scale is given in steps of 50 units, with each 50-unit step corresponding to one microsecond, which corresponds to a frequency of 50 kHz. The time scale in Fig. 8 was chosen to better represent a single peak. However, as already mentioned, the signal generated by an Eulerian sensor can contain multiple peaks, with each peak corresponding to a single scan across the FOV. The scan begins at about 0.5 microseconds with an initial value of about 3 volts and reaches a maximum value 803 of about 5.5 volts at about 2 microseconds. The heat source was located within the pyrometer's FOV from the beginning. As the scan continued, the heat source moved away from the pyrometer's FOV, allowing the material within the FOV to cool until the material was no longer hot enough to emit light that could be detected by the pyrometer. This limit was about 6 microseconds at a voltage value 805 of about 2.3 volts. The values 803 and 805 can define an upper and lower limit for the evaluation of the pyrometer signal.Similarly, thresholds can be used for signals from photodiodes and other types of optical sensors. For example, when estimating the size of a heated zone, a model or algorithm estimating the heated zone can consider the thresholds for the specific sensor that generated the signal and ignore temperatures associated with voltages outside the thresholds, such as temperatures associated with voltages above 5.5 volts and temperatures associated with voltages below 2.3 volts.

[0036] The raw signal can be converted to observed temperature values using a Stefan-Boltzmann relationship, such as the classical Stefan-Boltzmann law in equation (4) discussed below. The conversion can be based on an assumed emissivity of the heated material and can be performed by processing circuitry in the sensor or by a separate processor in the system. In some embodiments, the same processor that converts the raw signal to observed temperatures also corrects the observed temperatures. The following discusses ways to correct data acquired by an optical sensor with a FOV significantly larger than the heated zone to estimate the actual temperature from an observed temperature. In some embodiments, the observed temperature values are corrected based on an estimate of the size of the heated zone before further correction.Therefore, the raw signal may be corrected based on, for example, emissivity to produce a corrected set of observed temperature values. At least one of the corrected temperature values may then be further corrected based on an estimate of the heated zone. In some embodiments, the observed temperature values are corrected based on the estimate of the heated zone before further correction is made based on other factors, such as emissivity. These additional corrections are not always necessary. For example, in some embodiments, the effects of emissivity may be mitigated by using a dual wavelength sensor that collects data for light emitted at one of the two wavelengths.Since the emissivity is wavelength dependent, the sensor can calculate the observed temperature based on the average of the signals of both wavelengths without explicitly taking the emissivity into account. TEMPERATURE CORRECTION BASED ON THE SIZE OF THE HEATED ZONE

[0037] The temperature conversion method for an optical sensor is typically based on the sensor's FOV area, which, as previously explained, is much larger than the heated zone in additive manufacturing. Thus, the temperature conversion takes into account areas that do not contribute to the sensor's signal. As a result, the observed temperatures are usually significantly lower than the actual temperature of the heated zone. To correct this error, the observed temperatures can be inflated using a correction factor. One possible correction factor is based on the assumption that the observed temperature underestimates the actual temperature of the heated zone in proportion to the area of the field of view: φCORRECTION=AFOVAHEATED ZONE

[0038] Wherein A FOV indicates the area of the sensor's field of view and A AUFGEHEIZTE ZONEthe area of the heated zone. For this correction factor to be valid, two assumptions must be true: (1) all points within the heated zone contribute equally to the signal generated by the sensor and (2) the sensor can measure across all possible frequencies or wavelengths of emitted light. Neither assumption is actually true. The points within the heated zone do not contribute equally to the power radiated by the heated zone and the sensor is generally unable to detect radiation at all wavelengths. Using the above correction factor therefore leads to a strong overestimation of the actual temperature. As will be explained below in conjunction with the method of Fig. 9, a more accurate correction factor can be obtained based on the ratio of the size of the heated zone to the size of the hotspot.

[0039] Fig. Figure 9 shows a flowchart of a method for correcting temperature data acquired by a non-imaging sensor in an additive manufacturing system. The data could, for example, be acquired from one of the sensors previously described in connection with the Fig. 1 and Fig. 2. The method may be performed by at least one processor, for example, comprising a processor that executes instructions stored on a computer-readable medium. In some embodiments, the processor is hard-wired to perform at least a portion of the method. Thus, the method may be implemented in hardware, software, or a combination of hardware and software. The method may be performed during or after an additive manufacturing process in which the temperature data is collected. If the method is performed during additive manufacturing, corrected temperature values may be analyzed to make real-time adjustments to the additive manufacturing process, including, for example, adjustments to one or more operating parameters of the heat source.When applied after completion of an additive manufacturing process, the corrected temperature values can be stored for record keeping or later analysis.

[0040] At 900, raw data is acquired from an optical sensor. Whether Lagrangian or Eulerian, the sensor has a defined FOV. The size of the FOV is known or easily determined. For example, the FOV of a typical pyrometer used in additive manufacturing is on the order of 1 mm. The data can be acquired for a single scan of a laser or other heat source. Alternatively, the data can represent multiple scan passes across the sensor's FOV, possibly with different speeds, orientations, scan patterns, or output powers. The raw data can be obtained by sampling the raw signal generated by the optical sensor as driven by the emitted light.

[0041] At 902, the raw data is converted into temperature data based on assumed physical properties of the heated material. For example, the raw data may include a set of voltage values that the sensor converts into corresponding temperature values based on an assumed emissivity of the material. The assumed emissivity may be a preset value programmed or hardwired into the sensor or other device performing the conversion, or it may be a setting manually selected by an operator.

[0042] At 904, initial corrections are made to the temperature data. For example, the temperature data may be corrected based on differences between the assumed emissivity and a modeled or measured emissivity. In some embodiments, the initial correction may be made to the raw data, i.e., before temperature conversion. Thus, the initial correction may be made in a voltage or current domain rather than a temperature domain.

[0043] At 906, the size of the hotspot is estimated. To estimate the peak temperature, the area of the hotspot can be assumed to be equal to the area of the beam emitted by the heat source. For example, a typical laser beam diameter in an additive manufacturing system is on the order of (100; 250) micrometers (0.1 mm). Thus, the size of the hotspot can be estimated as corresponding to the area of the typical laser, e.g., π(0.05) 2 The beam size is usually a known or measurable quantity and can be obtained, for example, from the documentation of the heat source manufacturer.

[0044] At 908, the heat field generated on the build plane of the workpiece is estimated. As explained above, the heat field can be estimated based on various contributing factors, including operating parameters of the heat source and physical properties of the material, and can be expressed as a two-dimensional matrix of temperature values. A more accurate temperature correction method utilizing the estimated heat field is described below. First, the temperature values of the heat field are converted into radiant power based on a Stefan-Boltzmann relationship obtained by modifying the classical Stefan-Boltzmann law as follows.

[0045] The Stefan-Boltzmann law describes the relationship between the radiant power and the temperature of a surface of a blackbody radiator: E=ε⋅A⋅σ⋅T4

[0046] Where ε is the surface emissivity, A is the area of the region with temperature T, σ is the Stefan-Boltzmann constant, and T is the temperature in degrees Kelvin. The Stefan-Boltzmann law is related to the Planck equation: I(ω,T)=2hω3c2⋅1exp(hωkT−1)

[0047] Where I is the radiant power related to the area, the solid angle and the frequency, h is the Planck constant, ω is the frequency, k is the Boltzmann constant, c is the speed of light and T is the absolute temperature.

[0048] The Stefan-Boltzmann law is typically derived by integrating the Planck equation over all frequencies and all solid angles: PA=∫0∞2hω3c2⋅1exp(hωkT−1)⋅dω ∫ dΩ

[0049] Here, P / A is the power per unit area and the integration is carried out over all frequencies and all solid angles.

[0050] The assumption of uniform radiation across all frequencies is generally incorrect for optical sensors. For example, pyrometers are only sensitive within a specific frequency range, so the correct equation is: PA=∫ωLOWERωUPPER2hω3c2⋅1exp(hωkT−1)⋅dω ∫ dΩ

[0051] This integral has no analytical solution except over a finite frequency range, as it is a form of the Bose-Einstein integral. However, this leads to a law similar to the Stefan-Boltzmann law with one of four different exponents: E=ε⋅A⋅σ⋅Tn

[0052] The exponent n is a real number. For a narrow frequency range in the deep infrared, such as that of practical interest in pyrometers and other optical sensors used in additive manufacturing, the value of n can be as high as 6 or 7.

[0053] Applying the modified Stefan-Boltzmann law of Equation (8), the values of the thermal field can be converted into values representing the radiant power observed by a hypothetical, ideal optical sensor. Subsequently, at 910, the heated zone is estimated by identifying all points in the thermal field that are within a certain range of peak radiant power. The peak power occurs in the area directly irradiated by the heat source, i.e., the hotspot. The heated zone corresponds, for example, to all points that are within 50%, 25%, 10%, or 5% of the peak power. The zone can be determined depending on the operating parameters of the heat source and the physical properties of the material.For example, if the powder has a higher melting temperature, it must be heated to a higher temperature, creating a larger thermally affected area. Therefore, a 10% value may be used for a material with a melting temperature of 400°C, while a 35% value may be used for a material with a melting temperature of 200°C. In this way, thermal modeling estimates which parts of the FOV are hot enough to contribute to the sensor signal by identifying points where the radiated power is within a certain fraction of a peak radiated power of the thermal field.

[0054] After estimating the heated zone, at 912 each observed temperature value can be normalized with the following correction factor to obtain a corresponding peak temperature: φCORRECTION=AHEATED ZONEAHOTSPOT TMAX=φCORRECTION⋅TOBS

[0055] Wherein A HOTSPOT the area of the hotspot, T MAX the peak temperature and T OBS denote the observed temperature. INTEGRATION OF SIZE-BASED CORRECTION INTO AN AGGREGATION-BASED ANALYSIS SYSTEM

[0056] Fig. Figure 10 shows a flowchart in which data recorded by an optical sensor, such as a non-imaging photodetector, can be processed to characterize a build process in additive manufacturing. At (100; 250)2, raw sensor data is received, which may include both build plane intensity data and energy source drive signals that are correlated with each other. At (100; 250)4, by comparing the drive signal and the build plane intensity data, individual scans can be identified and located within the build plane. Generally, the energy source drive signal provides at least a start and end position from which the area over which the scan extends can be determined. At (100; 250)6, the raw sensor data associated with an intensity or power of each scan can be grouped into corresponding X and Y grid areas.In some embodiments, raw intensity or power data may be converted into energy units by correlating the dwell time of each scan within a particular grid area. Each grid area may have a size corresponding to one or more pixels of an optical sensor monitoring the build plane. It should be noted that various coordinate systems, such as polar coordinates, may be used to store grid coordinates, and that coordinate storage is not limited to Cartesian coordinates. In some embodiments, different scan types may be segregated so that analysis can be performed only on certain scan types. For example, an operator may want to focus on contour scans if these scan types are most likely to exhibit undesirable variations.At (100; 250), the energy input at each grid area can be added up, so that the total amount of energy received in each grid area can be determined using equation (11). Epdm=∑n=1Pixel−measurements in grid cellEpdn

[0057] This addition can be performed shortly before adding a new powder layer to the build plane, or alternatively, the addition can be delayed until a predetermined number of powder layers have been deposited. For example, the addition could only be performed after five or ten different powder layer regions have been deposited and melted during an additive manufacturing process. In some embodiments, a sintered powder layer can contribute approximately 40 micrometers to the thickness of a part; however, the thickness varies depending on the powder type used and the thickness of the powder layer.

[0058] At 1010, the standard deviation is determined for the measured values acquired and assigned to each raster area. This allows for better identification of raster areas where energy values vary to a lesser or greater extent. Fluctuations in the standard deviation may indicate problems with sensor performance and / or cases where one or more scans are missing or the performance level is well outside normal operating parameters. The standard deviation can be determined using equation (12). Epdsm=1#Measured value in position 1 n=1Measured values in pixels (En−E¯)2

[0059] At 1012, a total energy density received in each grid area may be determined by dividing the power values by the total area of the grid area. In some embodiments, a grid area may have a square geometry with a length of approximately 250 micrometers. The energy density for each grid area may be determined using Equation (13). ERaster position=∑n=1Measurement values−in−positionEpdnARaster position

[0060] At 1014, when more than one part is being built, different grid regions may be associated with different parts. In some embodiments, a system may include stored part boundaries that allow each grid region and its associated energy density to be quickly linked to its respective part using the coordinates of the grid region and the boundaries associated with each part.

[0061] At 1016, an area of each layer of a part can be determined. If a layer contains voids or contributes to the confines of internal voids, significant portions of the layer must not receive energy. For this reason, the affected area can be calculated by adding only grid regions determined to receive energy from the energy source. At 1018, the total amount of power received by the grid regions within the portion of the layer associated with the part can be added and divided by the affected area to determine the energy density for that layer of the part. Area and energy density can be calculated using equations (14) and (15). APart=∑n=1Part−Pixel1(Epdn>0) IPQMPartLayer=∑n=1Part−GridPositionsEpdnAPart

[0062] At 1020, the energy density of each layer can be summed together to obtain a metric that indicates the total amount of energy received by the part. The total energy density of the part can then be compared to the energy density of other similar parts at the build plane. At 1022, the total energy of each part is summed. This allows high-quality comparisons to be made between different built elements. Build comparisons can be helpful in identifying systematic differences such as powder variations and changes in overall output power. Finally, at 1024, the summed energy values can be compared to other layers, parts, or build planes to determine the quality of other layers, parts, or build planes.

[0063] There may be other metrics in the Fig. 10. For example, the maximum temperature may be one of these metrics, and the method described here for performing a temperature correction based on a calculated correction factor can be used to obtain a more accurate maximum temperature. The collected maximum temperature data can be used in a variety of ways. For example, the maximum temperature values can be averaged for each grid range or the highest maximum temperature can be linked to each grid range. Alternatively, the mode of the upper 5 or 10 percent of the maximum temperature values can be linked to the grid range to prevent a single occurrence of a greatly elevated maximum temperature from displacing an otherwise more nominal maximum temperature value.

[0064] It is pointed out that the Fig. 10 illustrate a specific method for characterizing an additive manufacturing build process according to another embodiment of the present invention. In alternative embodiments, other sequences of steps may be performed. For example, in alternative embodiments of the present invention, the steps described above may be performed in a different order. Furthermore, the steps in Fig. 10 may comprise several substeps that can be performed in various sequences depending on the respective step. Furthermore, additional steps may be added or removed depending on the application. Many variations, modifications, and alternatives will be apparent to those skilled in the art.

[0065] The Fig. 11A - Fig. 11D show illustrative illustrations showing how multiple scans can contribute to the energy input into the individual grid areas. Fig. 11A illustrates a grid pattern consisting of multiple grid regions 1102 distributed over a region of a part being built using an additive manufacturing system. Fig. 11A further shows a first pattern of energy scans 1104 extending diagonally across the grid areas 1102. The first pattern of energy scans 1104 may be generated by a laser or other strong source of thermal energy scanning across the grid 1102. Fig. Figure 11B shows how the energy deposited across the part is represented in each of the raster regions 1102 by a single grayscale color representative of an amount of energy received therein, with darker shades of gray corresponding to larger amounts of energy. Note that in some embodiments, the size of the raster regions 1102 may be reduced to obtain higher resolution data. Alternatively, the size of the raster regions 1102 may be increased to reduce memory and computational requirements.

[0066] Fig. 11C shows a second pattern of energy scans 1106 that overlaps with at least a portion of the energy scans of the first pattern of energy scans. As described in the text of Fig. As explained in Figure 8, the grid areas where the first and second patterns of energy scans overlap are shown in a darker shade to illustrate how the energy from both scans has increased the amount of energy received at the overlap of the scan patterns. Of course, the method is not limited to two overlapping scans and may involve many additional scans added together to fully represent the energy received in each grid area.

[0067] The Fig. 12A - Fig. 12F illustrate how a grid can be dynamically created to characterize and control an additive manufacturing process. Fig. 12A shows a top view of a cylindrical workpiece 1202 located on a portion of a build plane 1204. The workpiece 1202 is shown during the additive manufacturing process. Fig. Figure 12B shows how the cylindrical workpiece 1202 can be divided into multiple trajectories 1206, along which an energy source can melt powder distributed on a top surface of the cylindrical workpiece 1202. In some embodiments, the energy source can alternate directions 1206 as shown, while in other embodiments, the energy source can only move in one direction. Furthermore, the direction of the trajectories 1206 can vary from layer to layer to further randomize the orientation of the scans for building the workpiece 1202.

[0068] Fig. 12C shows an example of a scanning pattern for the energy source in forming a portion of the workpiece 1202. As indicated by arrow 1208, the direction of travel of an exemplary energy source across the workpiece 1202 is diagonal. Individual scans 1210 of the energy source may be oriented in a direction perpendicular to the direction of travel of the energy source along the path 1206 and may extend completely across the path 1206. The energy source may be briefly turned off between consecutive individual scans 1210. In some embodiments, the duty cycle of the energy source when traversing each path 1206 may be approximately 90%. By employing this type of scanning strategy, the energy source may cover a width of the path 1206 while traversing the workpiece 1202. In some embodiments, the stripe 1210 may have a width of approximately 5 mm.This can significantly reduce the number of traces required to build the workpiece 1202, since in some embodiments the width of a melt pool generated by the energy source can be on the order of about 80 micrometers.

[0069] The Fig. 12D - Fig. 12E illustrate how grid regions 1212 along each trajectory 1206 can be dynamically generated and sized to accommodate the width of each individual scan 1210. The precise position of subsequent scans can be predicted by the system by referencing energy source drive signals en route to the energy source. In some embodiments, the width of the grids 1212 can be equal to the length of each scan 1210 or within 10% or 20% of the length of each scan 1210. Again, the scan length of each scan 1210 can be predicted by referring to the energy source drive signals. In some embodiments, the grid regions 1212 can be square or rectangular. A thermal energy density can be determined for each of the grid regions 1212 as the energy source progresses along the trajectory 1206.In some embodiments, thermal energy density measurements within raster region 1212-1 can be used to adjust the output power of the energy source in the next raster region, in this case, raster region 1212-2. For example, if the thermal energy density values generated by individual scans 1210 within raster region 1212-1 are significantly higher than expected, the energy source output power can be reduced, the speed at which the energy source is scanned across individual scans 1210 can be increased, and / or the spacing between individual scans 1210 within raster region 1212-2 can be increased. These adjustments can be made in a closed-loop control manner. Although in . Fig. 12E clearly depicts only five individual scans 1210 within each grid area 1212, this is only an example, and the actual number of individual scans within a grid area 1212 may be significantly higher or, in some cases, lower. For example, if the melt zone created by the energy source is approximately 80 micrometers wide, approximately 60 individual scans 1210 are required for all of the powder within a 5 mm square grid area 1212 to fall into the melt zone.

[0070] Fig. 12F shows an edge region of the workpiece 1202 after the energy source has finished traversing the pattern of traces 1206. In some embodiments, the energy source may continue to supply energy to the workpiece 1202 after a majority of the powder has melted and resolidified. For example, the contour scans 1214 may be along the periphery 1216 of the workpiece 1202 to perform a surface treatment on the workpiece 1202. It should be noted that the contour scans 1214, as shown, are significantly shorter than the individual scans 1210. For this reason, the raster regions 1218 may be significantly narrower than the raster regions 1212. It should also be noted that the raster regions 1218 are not truly rectangular, as in this case, they follow the contour of the periphery of the workpiece 1202.Other cases that can lead to varying scan lengths include when a workpiece has walls of varying thickness. A wall of varying thickness can cause the scan length to vary within a single raster area. In such a case, the area of each raster area can be kept uniform by increasing the length of the raster area while decreasing the width to accommodate changes in the length of each scan.

[0071] Fig. 13 shows an example of a controller having a feedback control loop 1300 for establishing and maintaining closed-loop additive manufacturing. At block 1302, a baseline thermal energy density for the next grid area through which the energy source is to pass is input to the control loop. The baseline thermal energy density may be determined from modeling and simulation programs and / or from previously conducted experiments / test runs. In some embodiments, the baseline thermal energy density data may be adjusted by the energy density adjustment block 1304, which includes energy density values for various grid areas recorded in previous layers. The energy density adjustment block 1304 may include an adjustment for the energy density baseline block if previous layers received too much or too little energy.For example, if optical sensor readings indicate a thermal energy density below target values in an area of a workpiece, energy density adjustment values can increase the baseline energy density for grid areas that overlap grid areas with readings below the target thermal energy density. This allows the energy source to additionally melt powder that was not fully melted in the previous layer(s).

[0072] In addition to energy density, other metrics can be used in the Fig. 12A - Fig. 12F. One of these metrics may, for example, be the maximum temperature, and the method described herein for performing a temperature correction based on a calculated correction factor may be used to obtain a more accurate maximum temperature for a given grid area. The collected maximum temperature data could be used in a variety of ways. For example, the maximum temperature values for each grid area could be averaged or the highest maximum temperature could be associated with each grid area. Alternatively, the mode of the upper 5 or 10 percent of the maximum temperature values could be associated with the grid area to prevent a single occurrence of a greatly elevated maximum temperature from displacing an otherwise more nominal maximum temperature value.

[0073] Fig.13 further shows how the inputs of blocks 1302 and 1304 cooperate to generate an energy density control signal that is received by controller 1306. Controller 1306 is configured to receive the energy density control signal and generate thermal source input parameters configured to generate the desired thermal energy density within the current scan range. The input parameters may include power, scan speed, scan pitch, scan direction, and scan duration. The input parameters are then received by energy source 1308, and any changes to the input parameters are applied by energy source 1308 for the current scan range. Once optical sensors measure the scans of energy source 1308 that form the current scan range, the thermal energy density for the current scan range is calculated at block 1310 and compared to the energy density control signal.If the two values are equal, no change is made to the energy density control signal based on the optical sensor data. However, if the two values are different, the difference is added to or subtracted from the energy density control signal for scans in the next raster range.

[0074] The various aspects, embodiments, implementations, or features of the described embodiments may be used individually or in any combination. Various aspects of the described embodiments may be implemented by software, hardware, or a combination of hardware and software. The described embodiments may also be implemented as computer-readable code on a computer-readable medium for controlling manufacturing operations or as computer-readable code on a computer-readable medium for controlling a manufacturing line. The computer-readable medium may be any data storage device capable of storing data that can subsequently be read by a computer system. Examples of a computer-readable medium include read-only memory, random access memory, CD-ROMs, hard disks, DVDs, magnetic tape, and optical data storage devices.The computer-readable medium may also be distributed across network-coupled computer systems so that the computer-readable code is stored and executed in a distributed form.

Claims

[1] Additive manufacturing process comprising: Recording (900) temperature data describing temperature changes across a surface of a material (104; 256) during an additive manufacturing process in which the material (104; 256) is heated by a heat source (100; 250), the temperature data being generated using a non-imaging optical sensor (109, 111, 113, 122; 259) with a defined field of view (400, 500, 600, 600'); Estimating (906) the size of a hotspot (604) corresponding to the hottest region formed by the heat source (100; 250) on the surface, wherein the size of the hotspot (604) is estimated to be equal to the size of an energy beam (101; 254) emitted by the heat source (100; 250); Estimating (908) the size of a heated zone (505) corresponding to a locus of points within the field of view (400, 500, 600, 600') that contribute to the temperature data; and Correcting (912) the temperature data based on the estimated size of the hotspot (604) and the estimated size of the heated zone (505). [2] The additive manufacturing method of claim 1, further comprising: estimating a thermal field describing a spatial temperature distribution generated in the material (104; 256) by the heat source (100; 250). [3] The additive manufacturing method of claim 2, wherein the heated zone (505) is estimated by calculating the radiant power for various points within the thermal field and identifying points for which the radiant power is within a certain fraction of a peak radiant power of the thermal field. [4] Additive manufacturing method according to claim 3, wherein the thermal field is estimated as a function of space and time based on operating parameters of the heat source (100; 250) and physical properties of the material (104; 256). [5] Additive manufacturing method according to claim 4, wherein the operating parameters comprise the output power of the heat source (100; 250) and / or the travel speed of the heat source (100; 250) and / or the scan pattern (706) of the heat source (100; 250) and / or the spatial orientation of the heat source (100; 250) relative to the material (104; 256) and / or the size of an energy beam (101; 254) emitted by the heat source (100; 250) and wherein the physical properties comprise the density of the material (104; 256) and / or the thermal conductivity of the material (104; 256). [6] Additive manufacturing method according to any one of claims 1 to 5, wherein the temperature data are corrected by normalizing an observed temperature with a correction factor φCORRECTION=AHEATED ZONEAHOTSPOT, where A AUFGEHEIZTE ZONE is the estimated area of the heated zone (505) and A HOTSPOT the estimated area of the hotspot (604). [7] Additive manufacturing method according to one of claims 1 to 6, wherein the optical sensor (109, 111, 113, 122; 259) is a photodiode (109) or a pyrometer (111). [8] Additive manufacturing system comprising: a heat source (100; 250); a non-imaging optical sensor (109, 111, 113, 122; 259) configured to measure temperatures across a surface of a material (104; 256) during an additive manufacturing process in which the material (104; 256) is heated by the heat source (100; 250), wherein the optical sensor (109, 111, 113, 122; 259) has a defined field of view (400, 500, 600, 600'); and at least one processor configured to receive temperature data from the optical sensor (109, 111, 113, 122; 259), wherein the at least one processor is further configured to: Estimating (906) the size of a hotspot (604) corresponding to the hottest region formed by the heat source (100; 250) on the surface, wherein the at least one processor estimates the size of the hotspot (604) to be equal to the size of an energy beam (101; 254) emitted by the heat source (100; 250); Estimating (908) the size of a heated zone (505) corresponding to a locus of points within the field of view (400, 500, 600, 600') that contribute to the temperature data; and Correcting (912) the temperature data based on the estimated size of the hotspot (604) and the estimated size of the heated zone (505). [9] Additive manufacturing system according to claim 8, wherein the at least one processor is configured to estimate a thermal field describing a spatial temperature distribution generated in the material (104; 256) by the heat source (100; 250). [10] The additive manufacturing system of claim 9, wherein the at least one processor estimates the heated zone (505) by calculating the radiant power for various points within the thermal field and identifying points for which the radiant power is within a certain fraction of a peak radiant power of the thermal field. [11] Additive manufacturing system according to claim 9 or 10, wherein the at least one processor is configured to estimate the thermal field as a function of space and time based on operating parameters of the heat source (100; 250) and physical properties of the material (104; 256). [12] Additive manufacturing system according to claim 11, wherein the operating parameters comprise the output power of the heat source (100; 250) and / or the travel speed of the heat source (100; 250) and / or the scan pattern (706) of the heat source (100; 250) and / or the spatial orientation of the heat source (100; 250) relative to the material (104; 256) and / or the size of an energy beam (101; 254) emitted by the heat source (100; 250), and wherein the physical properties comprise the density of the material (104; 256) and / or the thermal conductivity of the material (104; 256). [13] Additive manufacturing system according to one of claims 8 to 12, wherein the at least one processor corrects the temperature data by normalizing a measured temperature with a correction factor φCORRECTION=AHEATED ZONEAHOTSPOT, where A AUFGEHEIZTE ZONE is the estimated area of the heated zone (505) and A HOTSPOT the estimated area of the hotspot (604). [14] Additive manufacturing system according to one of claims 8 to 13, wherein the optical sensor (109, 111, 113, 122; 259) is a photodiode (109) or a pyrometer (111).

Images