METHOD AND APPARATUS FOR DETERMINING METAL POWDER CONDITION

Abstract

A method of determining the condition of a metal powder for use in an additive manufacturing process, involving processing an image of the powder to measure a surface property of the powder, such as colour, texture or particle shape. The proportion of powder whose measured surface property falls outside a pre-determined range is determined and can be used to decide whether or not the powder is suitable for re-use. The proportion is determined by identifying individual particles in the image which are identified as statistical outliers amongst all of the particles shown in the image when considering a measured surface property. The relevant proportion may be determined statistically.

Claims

1. A method of determining the condition of a metal powder for use in an additive manufacturing process, the method comprising the step of processing an image of the powder.

2. A method as claimed in claim 1 comprising the step of taking the image of the powder.

3. A method as claimed in claim 1 wherein the image is an optical image.

4. A method as claimed in claim 1 wherein the image is processed to measure a surface property of the powder, such as colour or texture.

5. A method as claimed in claim 1 comprising the step of cropping the image.

6. A method as claimed in claim 1 comprising the step of checking the quality of the image.

7. A method as claimed in claim 1 wherein the image is formed by, or divided into, a plurality of image elements.

8. A method as claimed in claim 7, wherein the ratio of image elements in the image to the number of particles in the imaged powder is at least 100:1.

9. A method as claimed in claim 7 wherein image elements having a luminance below a threshold are identified and excluded from further processing.

10. A method as claimed in claim 9 comprising the step of calculating the ratio of image elements above and below the threshold.

11. A method as claimed in claim 1 wherein the image is processed to estimate the total number of particles of powder it shows.

12. A method as claimed in claim 1 where the image is processed to identify image elements which represent a surface property of imaged powder which falls outside a chosen range.

13. A method as claimed in claim 12 wherein the image elements are identified by selecting those which represent a predetermined outlying proportion of the distribution of represented surface property of all image elements considered.

14. A method as claimed in claim 13 wherein the outlying proportion is less than 5% or less than 1%.

15. A method as claimed in claim 12 wherein groups of connected identified image elements above a predetermined size are identified thereby to identify particles of powder with a surface property that falls outside the chosen range.

16. A method as claimed in claim 15 comprising the step of storing, for each identified particle, data relating to each image element representing the particle and a unique identifier.

17. A method as claimed in claim 1 comprising the step of determining the number of particles in the image with a surface property which falls outside a chosen range.

18. A method as claimed in claim 12 comprising the step of indicating that the powder is not suitable for re-use when the proportion of measured powder whose measured surface property falls outside the chosen range exceeds a predetermined value.

19. A method as claimed in claim 12 comprising determining the average measured surface property of the measured powder particles whose measured surface property falls within the chosen range.

20. A method as claimed in claim 19 comprising the step of indicating that the tested powder is not suitable for re-use when average measured surface property of the powder whose measured surface property falls within the chosen range is greater or less than a predetermined threshold.

21. A method as claimed in claim 1 comprising the step of processing the image to resolve individual particles and measuring a surface property of each individual particle.

22. A method as claimed in claim 1 comprising the step of placing the powder into an enclosure and illuminating the powder in the enclosure.

23. Apparatus for determining the condition of a metal powder for use in an additive manufacturing process, the apparatus comprising a processor arranged to process an image of the powder.

24.-34. (canceled)

Description

DETAILED DESCRIPTION OF THE INVENTION

[0058] In order that the invention may be more clearly understood embodiments thereof will now be described, by way of example only, with reference to the accompanying drawings, of which:

[0059] FIG. 1 is a schematic view of an embodiment of apparatus for analysing powder condition;

[0060] FIG. 2 is a schematic view of another embodiment of apparatus for analysing powder condition;

[0061] FIG. 3 is a schematic view of another embodiment of apparatus for analysing powder condition;

[0062] FIG. 4 is a flowchart showing steps involved in processing an image of powder;

[0063] FIG. 5 is a graph showing number of particles against colour; and

[0064] FIG. 6 is a plot of image elements on axes representing the a and b parameters in CIELAB colour space.

[0065] Referring to the drawings, FIG. 1 shows a first apparatus for analysing metal powder. It comprises an openable substantially light tight enclosure 1. The enclosure houses a container 2 for powder 3 which may take the form of a dish or slide, or any other suitable form. The container is open to the top and has a substantially square opening with a side of about 6 mm, giving it a surface area of about 36 mm.sup.2. It has a depth of at least 2 mm. The container may be removed from the enclosure. The enclosure also houses a microscope 4 which is mounted to a digital camera 5, and lamps 6. The camera 4 comprises a substantially square sensor, such as a CCD sensor, with approximately four mega pixels and is connected to a computer 7 which comprises a keyboard and mouse or other user interface and is connect to a display 8 and/or other output device. The lamps are arranged to provide a diffuse light. They are shown as dome or flat dome lamps. In an alternative arrangement (and in other embodiments) a ring light could be used.

[0066] In use, a sample of powder 3 taken from a batch of powder to be analysed is introduced into the container 2, either with the container in or out of the enclosure 1. The powder is introduced in sufficient quantity to form a depth of powder which entirely obscures the bottom of the container 2 when viewed from above. So the depth of powder typically comprises at least two, and preferably more than two, layers of particles. The powder is levelled in the container so that it has a substantially flat upper, planar surface. If powder has been introduced into the container whilst outside the enclosure the container is then positioned in the enclosure beneath the microscope and the enclosure closed.

[0067] The lamps 6 are then activated. The lamps may be controlled by the computer 7. The lamps are arranged to illuminate the powder 3 in the container 2. Illuminating the powder with lamps in a substantially light tight enclosure enables powder to be analysed in controllable and repeatable light conditions.

[0068] The camera 5 is then caused to take an image of the upper surface of the powder in the container and to transmit it to the computer 7. The camera and microscope are arranged to take an image of substantially all of the surface of the powder in the container. The field of view of the camera and microscope thus images an area of about 36 mm.sup.2. Metal powders used in AM processes typically have an average diameter of the order of tens of microns. As such, the number of particles visible to the surface of the powder imaged by the camera will be of the order of thousands and so about three orders of magnitude less than the number of pixels of the sensor. The camera is thus able to produce a digital image of the surface of powder in which there about 1000 times as many pixels than the number of particles of powder shown in the image.

[0069] The image taken by the camera is then transmitted to the computer 7 for processing.

[0070] FIG. 2 shows a second apparatus for analysing metal powder, in this example built in to powder transport pipes 10 leading into and out of a powder sieve, and into a connected a powder blending device. Each pipe includes a transparent window 11 over which is fitted an enclosure 12 which houses a digital camera 5 fitted with an appropriate lens 4 for taking an image of powder in the pipe 10 through the window 11. A lamp or lamps 6 is/are provided in the enclosure around the lens 4 to illuminate the powder through the window 11. The camera outputs its image to a computer 7 without output device 8.

[0071] As with the apparatus shown in FIG. 1 the digital camera 5 has a sensor with about 1000 times the number pixels than the expected number of powder particles visible in the area of the window 11 imaged by the camera when the pipe is full of powder to be analysed.

[0072] A window 11 and associated enclosure 12 with camera 5 is provided in both of the pipes 10 leading to and from the sieve enabling the condition of powder to be analysed before and after sieving. Cameras also enable the condition of powder entering both the powder blending device.

[0073] Windows could be provided into powder transport conduits or powder storage containers of other types of apparatus, such as for example an additive manufacturing machine.

[0074] FIG. 3 shows a third apparatus for analysing metal powder, which in this example is built into an additive manufacturing machine 13. The machine comprises an enclosure 14 which is or can be made substantially light tight. The enclosure houses a powder delivery container with a powder delivery piston 15, a build container with a build platform 16, and a wiper blade 17 mounted to a moveable support for transferring powder from the powder delivery container to the build container. The enclosure also houses output optics 18 of a laser for selectively melting powder on the build platform 16. These features are all common to known selective layer melting additive manufacturing machines.

[0075] The enclosure 14 additionally houses two cameras 5 with appropriate lenses 4 for taking an image of an area of the top surface of powder in the powder delivery and build containers, and lamps 6 disposed around each camera. An imaging sensor 19 and lamp 20 is also mounted to the moveable support for the wiper blade and arranged to scan an image of the surface of powder in the powder supply or build containers as the wiper blade travels to and fro across the containers.

[0076] The cameras 5 and sensor 19 are arranged to output an image to a connected computer 7 with output device 8. As with the apparatus shown in FIGS. 1 and 2 the digital camera 5 and sensor 19 are arranged to produce an image of powder with about 1000 times the number pixels than the number of particles shown in the imaged area of powder.

[0077] It will be appreciated that the machine shown in FIG. 3 need only have a single imaging device.

[0078] In use each embodiment of the apparatus produces a digital image of the surface of powder in the apparatus. The digital image comprises a set of data defining properties of image elements. The ratio of image elements to the number of particles of powder shown in the image is about 1000. The image data is transmitted to the computer where it is stored in a manner where the colour and luminance of each element of the image is defined in the CIELAB colour space by variables L, a and b.

[0079] The computer is arranged to process the image data in order to determine information relating to the condition of the powder shown in the image by performing at least some of the steps shown by FIG. 4.

[0080] As a first optional step 21 the image may be cropped to a predetermined size, excluding elements outside a boundary (or some other chosen region) of the original image. This optional step allows distorted areas of an image to be excluded as well as enabling images taken by different cameras or sensors to be reduced to represent the same area and/or to have the same number of pixels.

[0081] The remaining image data, or remaining image data, may then be tested 22 to ensure that it is of sufficient quality for further processing. If not, it is rejected at 23 and a new image is obtained.

[0082] If the image data is of sufficient quality, the computer then identifies elements with a luminance below a predetermined threshold and removes these from the image data at 24, with the aim of removing elements which represent space between particles of powder (or other background material) in the image. The actual threshold will depend upon characteristics of the particular apparatus being used and type of powder being tested. With the elements of the image removed which lie outside the threshold the remaining image elements are taken to represent particles of powder in the foreground of the image.

[0083] The data for the remaining image elements may then be processed at 25 to estimate the number of particles they represent using a suitable technique, such as watershed segmentation. The total number of particles represented can also be estimated in other ways. For a given powder and apparatus the number of particles expected to be visible in an area of the surface of the powder corresponding to that represented by the image data can be calculated with a knowledge of the expected particle size and expected packing density of the powder.

[0084] The data for the remaining image elements is then statistically analysed at 26 to determine how the colour of each image element is distributed about the mean colour of all remaining elements to detect outlier elements with a colour that places them outside a threshold proportion of the entire population of elements. This may be performed using a Chi-squared test for outlier detection. The relevant proportion of the population may be selected according to the type of powder being analysed, but a typical proportion is 0.1%, that is to say that the elements of interest, the outlier population, make up 0.1% of the entire population of elements.

[0085] Visual representations of this step are shown in FIGS. 5 and 6. FIG. 5 plots the number of image elements on the vertical axis against a measure of colour on the horizontal axis. This shows a generally bell-shaped curve of colour distribution about a mean value at 27, and lower 28 and upper 29 thresholds which identify the outlying 0.1% of the population represented by the area under the curve outside the thresholds. FIG. 5 shows the same data where the a and b values for each image element in CIELAB colour space are plotted respectively on the horizontal and vertical axes and the threshold is the circle (or ellipse) 30. The elements falling within the circle 30 have been removed leaving the remaining 0.1% outlier population.

[0086] The outlier elements are then subjected to a connected component filter at 31 to determine if they are spatially connected in the image they define. Any group of connected image elements which exceeds a predetermined number of elements is considered to represent a single particle. The data representing each such identified group is associated with a unique particle identifier with the first identifier identifying the largest group of connected elements, the second identifier identifying the next largest group of connected elements, and so on.

[0087] At this stage the computer has produced sets of image data which define the size and colour of individual particles within the images powder with a colour that causes them to represent statistical outliers within the powder.

[0088] This data is then analysed to extract useful data relating to the condition of the analysed powder, including:

[0089] The number and thus proportion of particles having a colour which lies outside a predetermined range.

[0090] The mean colour of image elements lying within the predetermined range.

[0091] The mean colour of all imager elements.

[0092] It has been found that the colour of metal particles changes as the particles degrade. In particular it changes as particles oxidise and/or are exposed to heat. The more a particle is oxidised or the higher temperature a particle is exposed to the greater its colour changes. So, the amount of colour change is related to the degree of degradation a particle has suffered and thus also its suitability for re-use.

[0093] It has further been found that, notwithstanding the average condition of a batch of powder, the presence of highly degraded particles can render the batch unsuitable for re-use. This is because inclusion of even a single highly degraded particle in a build can significantly affect properties of the build. Where a highly degraded particle or particles become(s) incorporated into an article this could render the article unsafe, especially if the particle(s) is/are incorporated into the article at a location where there will be a stress concentration in use.

[0094] The first measure above will, assuming that the batch of powder from which the sample is taken is well mixed or the imaged area of a powder is representative of the constitution of the powder as a whole, generally mirror the proportion of significantly degraded particles throughout the sample and throughout the, or batch of, powder tested. Multiple samples may be taken from a given batch and analysed separately, or multiple tests performed on a batch of powder in order to improve accuracy such as by taking multiple images of a surface of the powder. And/or a particular sample could be analysed, mixed, and then reanalysed. An appropriate colour range and threshold minimum proportion outside that range can be determined for a given powder and build and where the proportion of particles outside the threshold exceeds the chosen limit the batch of powder is deemed unsuitable for re-use, at least for the build in question.

[0095] Thus, this measure enables powder condition to be determined independent of a bulk quantity.

[0096] The second measure provides an indication of the average degradation of the remaining powder when the particles with colours lying outside the threshold have been discounted. Such a measure is more akin to the result of a conventional bulk oxygen measurement, but obtained in a more convenient and non-destructive way, save that it excludes the influence of significantly degraded particles (or any internal oxygen). Powder may be deemed unsuitable for re-use where the average colour of the remaining particles, when the particles with colours lying outside the predetermined range have been discounted, lies outside another predetermined range of colours.

[0097] The third measure is similar to the second measure, but takes account of the significantly degraded particles. Powder may be deemed unsuitable for re-use where the average colour of particles lies outside another predetermined range of colours.

[0098] A decision whether or not to re-use powder can be based on one or more of the three measures described above. Typically a powder would not be re-used if any measure determines that the powder should not be re-used. In one embodiment the first and second measures are calculated and powder deemed unsuitable for re-use if either measure indicates this.

[0099] It is useful to analyse new powder before it is used and to subsequently analyse it after it has been used in a build process and any further build processes. Analysis of the new powder provides useful control data with which to compare that of subsequent analysis.

[0100] Other information about a powder may be determined from an image of the powder. Non-powder artefacts may be detected in the powder. Anomalous powder particles may also be detected, for example particles made of different material to that intended, where the anomalous particles may be identified by an observable property. An estimate of total incident energy received by the powder may be made. Where multiple images of a sample or batch of powder are made and analysed it is possible to determined how well blended the powder is by comparing results between images. Images taken form powder processed by different machines, such as AM machines, may be used to compare machine performance and/or determine machine health. Images taken a different time periods and/or different positions in apparatus can help to track transit of powder through the apparatus.

[0101] Where apparatus is incorporated into powder handling apparatus or an additive manufacturing machine the output of test results may automatically cause the apparatus to perform a certain function. For example, powder may be rejected for further use, or combined with other powder to refresh it before further use. An additive manufacturing machine may be stopped or a wiper blade caused to remove a layer of powder and replace it before any powder is fused.

[0102] Data relating to analysis of powder may be stored so as to validate a build using the powder. In particular, analysis of at least part of the surface of layers of powder deposited during a build process may be stored to provide evidence of the consistency or otherwise of the powder used throughout a build process. Also, time stamped data can be compared from multiple images of powder taken at different points throughout a powder transport system to audit the performance of that system, e.g to show how effectively oxidised or contaminated powder moves through it.

[0103] In each example, the computer is provided with suitable software to cause the camera to take an image, to process the image to determine colour distribution amongst image elements, to enable a user to input ranges, proportions or other values, to calculate one or more of the three measures, to determine if a particular sample may or may not be re-used having regard to the range(s) and proportion specified by a user and to output this result to a user via the display 8 or otherwise.

[0104] In one example a sample of used Ti64 alloy powder was analysed using the described apparatus. The image of the powder produced showed the vast majority of the powder to have a silver/grey colour generally similar to the colour of virgin powder, and a very small proportion to have a green, brown, blue, purple or black colour, these colours being indicative of increased oxidation or other degradation. A pre-determined range of colour was therefore chosen to encompass silver/grey particles and exclude the other colours representing 0.1% of the overall population of image elements. This range effectively encompassed particles which had suffered no or limited oxidation and been subjected only to low temperatures. A proportion of particles having a colour outside this range, effectively a proportion of particles that have suffered significant oxidation or degradation through exposure to high temperatures, could then be chosen as the limit beyond which the powder should not be re-used. When the colour of the observed particles was plotted in the manner shown in FIG. 5 it could be seen that the particles with colours other than silver/grey brown effectively represented outliers to the general distribution of particles.

[0105] The above embodiments are described by way of example only. Many variations are possible without departing from the scope of the invention as defined in the appended claims.