Final Report Summary - DIRA-GREEN (Enable the powder metallurgy process to expand to new markets with more reliable parts and lower manufacturing costs through the inspection of green parts)
Executive Summary:
The powder metallurgy (PM) industry has long been feeling the need for a reliable method to inspect parts and detect local defects in the green state. The required test should be non-destructive, fully automatic, cost effective and fast enough to be integrated in a parts production line. Known inspection systems such as eddy current and magnetic bridge testing, magnetic particle inspection, acoustic resonance and ultrasonic testing can be applied to sintered parts, but fail when applied to green compacts. Optical inspection with cameras can only identify surface defects and cannot detect internal defects arising due to material porosity. X-ray computer tomography (CT) is a powerful tool even for green compacts, but due to the high cost of the equipment it is often commercially unattractive and it cannot be used in-line due to long processing times. Therefore a simpler, faster and less costly alternative was developed by DIRA-GREEN.
The goal of the DIRA-GREEN project was to build a digital radiography (DR) inspection system such that it responds in real time and can be incorporated into production lines for automatic defect detection in green state parts. The quality of the manufactured PM parts should then be assured by monitoring for compacted material porosity and identifying microscopic features such as cracks. The collected radiographic images for the PM components obtained using the DIRA-GREEN system should show the size and location of any flaws present. These images could be stored in a component-linked database and would facilitate an improvement cycle for the optimisation of powder metallurgy mould and die designs. Also, identifying defects before the parts are sintered would allow recycling of the powder material. The availability of a reliable and attainable inspection tool for detecting green part flaws would result in a substantial increase in the competitiveness of European small and medium PM enterprises (SMEs). It is expected that an increasing confidence in the supply of the final PM product can be achieved by improving the reliability of powder metallurgy parts and reducing costs associated with the production and control of the parts.
Project Context and Objectives:
The objective of this project was to develop a Non-Destructive Testing (NDT) technique, which enables online quality assurance of ‘green parts’, by monitoring compacted material porosity, and identifying microscopic cracks. This innovative tool uses digital radiography to create a density map for each component, indicating the size and location of defects.
The scientific objectives of the project were:
• To gain a greater understanding of digital radiography technology by using multi-dimensional, low frequency impedance measurement to assess porosity, cracks, and defects of P/M parts.
• To develop a deeper understanding of the type of defects present in ‘green parts’.
• A potential secondary scientific objective is to test DIRA-GREEN’s ability to inspect ceramics and P/M parts manufactured with high resistive materials.
The technological objectives of the project were:
• To design a unique inspection tool prototype for the assessment and quality control of internal defects in ‘green parts’ that is reliable, low cost, rapid, and easy to use (automated), thereby fulfilling the needs of the P/M SME industry, which currently employs inefficient/costly manual inspection techniques.
• To design the digital radiography subsystem including an electronic subsystem capable of generating and measuring the signals needed for operation of the digital radiography system.
• To implement a control system, along with the mechanical design, that ensures compatibility and ease of use in industry applications.
• To ensure the DIRA-GREEN inspection system is high flexible via the software control system, which allows reconfiguration within minutes.
• To implement a data processing and evaluation system.
• To create a software system that allows for the collection of data from defected parts, and the ability to store it in a database for use in ‘improvement cycles’.
• To perform the validation of DIRA-GREEN tool that is capable of identifying defects before sintering, thereby reducing scrap, and saving time and energy.
Project Results:
An integrated DR system, complete with an X-ray source controller, the external PC connection, power supply unit, interlocks and warning lamps, is shown in a diagrammatic representation in attachment Figure_1.jpg. During an inspection, a PM part is placed between an X-ray source and the digital detection media. The penetration of X-rays through the sample will depend on the X-ray energy as well as the thickness and density of the PM material under inspection. The penetrating X-ray quanta are detected using a digital X-ray detector. The detector is controlled, and images obtained, using a wired connection to a PC. As the X-rays pass through the component, any changes in bulk density, due to cracks, porosity, inclusions, or simply component thickness, attenuate the X-ray quanta in proportion to the material density or thickness. The digital detector pixel elements translate the received quanta to image pixel values. The detector software shows a dense region as dark, and less dense areas as light, and this can be used to image any defects.
The DIRA-GREEN radiography system was arranged as shown in attachment Figure_2a.jpg and Figure_2b.jpg. It mainly comprises a micro-focus X-ray source emitting X-rays vertically, a sample held by a mechanical manipulator placed in line of the X-rays, and an indirect flat panel digital detector. X-ray inspection starts with a sample being illuminated by an X-ray source which is held in a fixed position statically so as not to induce any mechanical vibration and project distortion onto the detector. PM components enter the enclosure on a conveyor belt system and are positioned on a purpose built manipulator in between the source and the detector. The sample holder and the detector are mounted on a vertical translator, allowing the source-to-detector distance (SDD) and the source-to-object distance (SOD) to be varied to alter the magnification of the resultant radiographic image. The manipulator can also change the orientation of the PM sample with respect to the X-ray beam. The sample holding pallets have been constructed using a homogeneous material such that it does not contribute a visible structure on the radiograph, potentially masking defects.
Essential components of the DIRA-GREEN prototype
A micro-focus X-ray source is an integrated solution encompassing the tube, high voltage power supply and control circuitry. It is controlled by PC via the source controller. The micro-focus X-ray source produces an ellipsoidal cone X-ray beam with beam angles of 120O/40O. This beam angle ensures that most parts to be imaged can be fully covered by the emitted X-rays. The degree of sharpness obtained from a radiographic image is determined by the focal spot size. The focal spot generated by the X-ray source of the DIRA-GREEN system is as small as 4 microns diameter. This is designed to reduce image blurring when compared to conventional mini-focus X-ray sources, and thus provides an enhanced flaw detection capability and sensitivity. The source size contributes to the radiographic quality, influences the geometry of the inspection, and sets radiographic image definition and resolution limits. With these factors in mind, as well as a detection accuracy of 0.5 mm required by the industry, an X-ray source with these specifications has been used in the DIRA-GREEN system.
The component generating the X-ray images is a flat panel detector. For a typical 40 x 40 mm sample provided by one of the consortium end users, and using 3x magnification, a detector field of view with a minimum 120 x 120 mm detection area is required. A survey was carried out in the X-ray detector supplier market to find a suitable flat panel detector. An indirect digital X-ray detector with a pixel resolution of 74.8 µm and a detection area of 145.4 x 114.9 mm resulting in an array of 1944 x 1536 pixels was chosen for the project. Attachment Figure_3.jpg shows the detector mounted on a frame. The detector comprises a scintillator material that converts X-rays into visible light photons, a fibre optic plate to guide the light and block the X-rays, and a CMOS sensor to convert the light into an electrical charge. With exposure times of 1 to 2 seconds the system is fast enough to be integrated in a parts production line.
The mechanical subsystem was designed for the manipulation of the inspected parts. The green part is placed on a pallet before it enters the lead chamber on a conveyor belt. A conveyor belt system driven by an electric motor has been constructed to accommodate specially designed pallets with interchangeable inserts to hold PM parts of varying shape and size. The conveyor feeds the parts into the X-ray chamber and presents the sample holding pallet to a customised mechanical manipulator which is located between the X-ray source and the detector, as depicted in attachment Figure_2a.jpg. Once inside the chamber, a manipulator takes the pallet from the conveyor and moves it into position for inspection. In order to detect non-planar defects, the manipulator allows orientation of the sample part between -45o and +45o to the X-ray beam, as shown in attachment Figure_2b.jpg. In addition the manipulator system allows the inspected part to be positioned such that its radiographic image can be magnified between 1.72x and 4.69x. These geometries were taken into account during the radiographic laboratory trials and optimisation. The best magnification depends on the size of the part and the size of defects to be detected. A magnification of 1.72x results in a maximum achievable image resolution of 35 microns and a magnification of 4.69x enhances the image resolution to 20 microns.
The pallet is made of a low density, homogeneous material such that it does not significantly attenuate the X-rays, or produce a pattern on the image which could be misinterpreted as, or mask, a defect. Several images are taken, for this purpose the manipulator can rotate the parts and change the aspect angle. Once the images are taken, the pallet is placed back onto the exit conveyor.
The DIRA-GREEN mechanical subsystem concept has been significantly improved compared to the originally proposed system. It became obvious that because of the wide range of part characteristics a universal gripping device is not applicable. Accordingly the task to develop a universal gripping tool is in excess of the DIRA-GREEN project scope. Therefore to make the DIRA-GREEN prototype simultaneously suitable for small and fragile MIM parts and more robust PM parts, a special pallet system had to be developed.
The factory personnel need to be protected from the ionising X-radiation produced by the X-ray source. For this reason a lead-steel enclosure was designed and constructed to surround all components of the system and prevent the radiation from escaping to the surrounding areas. The lead chamber for the X-ray protection has two sides, a top, bottom and back panels shielding the enclosure. On the front side there is a windowed door turning along the horizontal axis hence, enabling it to open in the upward direction. This door has a lead glass insertion which allows seeing inside the enclosure without the risk of exposure to X-radiation. There is also a double door at the front which locks the window door from opening while the X-ray source is in use.
The enclosure has been designed in accordance with the UK guidelines set out by the Health and Safety Executive (HSE) through their Approved Codes of Practice (ACOP), and by adherence to ‘The Ionising Radiations Regulations 1999 (IRR99)’ [2] made under the Health and Safety at Work Act 1974. A warning lamp tower is fitted ensuring information about the state of the system is communicated to personnel in the vicinity, and all openings are fully interlocked to guarantee that X-rays cannot be produced if a door is not fully closed. The enclosure, seen in attachment Figure_4.jpg has been tested and certified for safe operation by a designated Radiation Protection Adviser.
In order for the DIRA-GREEN system to be integrated in a production line, within a PM factory setting, a design as depicted in attachment Figure_5.jpg was chosen. Parts coming from the powder compaction press are fed by a conveyor system into the lead chamber. Inside the chamber is the radiation source, the manipulator and the flat panel detector. The inspected good parts leaving the lead chamber to the right are transferred to the sintering furnace.
DIRA-GREEN System Performance
The performance of the DIRA-GREEN system was first evaluated under laboratory conditions using standard methodologies. Trials on the radiographic equipment were also carried out on the stand-alone pieces to optimise the conditions and inspection parameters in the final integrated system.
The initial equipment testing was carried out in the existing general purpose laboratory radiography bay at TWI (Wales). At all times the necessary dark, gain and defect correction algorithms were processed to ensure clarity of the image. The performance of the X-ray equipment was evaluated by quantifying the focal spot size of the X-ray source and testing the actual pixel resolution of the detector. To perform these tests two different Image Quality Indicators (IQI) were used: a specialised high resolution line pair micro-chart, JIMA RT RC-02B, for the focal spot size measurement and a Duplex IQI (BS EN 462-5) [3] for the detector resolution.
JIMA RT RC-02B contains line pairs of varying distance ranging from 0.4 to 15 μm. The IQI was placed as close as possible to the source target to obtain the highest possible magnification in order to determine which of the micrometer-scale line pairs could still be distinguished as two lines. The radiographic images were obtained at X-ray source energies of 45kV. The 3 μm line pair was visible, as shown in Attachment Figure_6.jpg therefore 3 μm was inferred as the measure of the minimum focal spot size.
The pixel size of the digital X-ray detector as specified by the manufacturer is 74.8μm. The effective resolution of the detector was quantified using a duplex line-pair IQI (contains 13 line pairs ranging from 130 to 50 micron distance) by placing it on the surface of the detector at 1x magnification. In accordance with BS EN 462-5, the first unresolved line pair is then used as a measure of the basic spatial resolution. In Attachment Figure_7.jpg displays the radiographic image of the Duplex IQI including the intensity line profile drawn across the three smallest line pairs. It shows that 11D is the first unresolved line pair providing a measure of 80 micron for basic spatial resolution. This value is expected and proves that the pixel resolution of this particular detection is 74.8 μm.
The overall functionality and performance of the DR system was verified by inspecting PM parts with real defects. The green PM samples were inspected at a source-to-object distance that provided the greatest magnification whilst keeping the whole sample in the field of view of the detector. Experiments have been carried out using specific tube voltages and current, exposure times, PM sample orientation etc.
A radiograph of a green PM compact with intentionally produced defects is shown in Attachment Figure_8.jpg. The red circle indicates a greater thickness due to the sample holder. The blue circles show a crack at the top, and the green circles show loss of material on the left side edge and a crack on the right edge. The radiographic system has enabled inspection for differentiating local density changes of 2% in the PM parts as a target resolution.
Image processing system
For an automatic evaluation of part defects it is not sufficient just to acquire excellent images, but a suitable software evaluating the images is equally important. The task of the image processing module is to handle the images recorded by the digital radiography system and evaluate them with respect to defects. The X-ray detector produces the data in a matrix, allowing the measurement to be handled as images. Therefore, image-processing algorithms have been developed to identify discrepancies, voids and cracks within the internal structure of the material.
The evaluation software developed by DIRA-GREEN compares the recorded image of a compact to a reference (“golden image” - image of a perfect part). To be able to detect cracks with any orientation, several images are taken with different pallet rotation and tilting offset values. For each angle a separate golden image is required and each recorded image is compared to the corresponding golden image.
Attachment Figure_9.jpg shows a golden image of a MIM part and a recorded image of another MIM part from the same mould with a small area of reduced density. With the naked eye it is impossible to detect the defect in the second image. We will demonstrate how the image processing software is able to bring out this defect.
Image processing involves several steps. The first step is required to remove the background and noise from the input images and to effectively assign the area of comparison. This is done through threshold segmentation with noise elimination. Both the recorded image and the golden image undergo this process.
One of the main challenges of the image comparison in DIRA-GREEN was that the golden image and the recorded image had a slightly different position, orientation and scale due to mechanical tolerances of the manipulator hardware. To overcome this problem, contour images are created by eliminating all information of the part bodies (Attachment Figure_10.jpg). The two contour images are superimposed and aligned with a best fit (Attachment Figure_11.jpg). With the new contour coordinates a remapping of the recorded image is executed.
After remapping the recorded image, a distance image is created by computing the distance between each two corresponding pixels. High intensities in the distance image identify the possible defects, as shown in Attachment Figure_12.jpg. The purpose of generating the distance image is to create a matrix of data where each cell intensity represents a distance value from the ideal contour. A cell value is low/dark near the contour, and getting lighter as the cell is getting away from the ideal value. The generated distance image is measuring how far a transformed pixel is from the golden contour. With the contour distance image the user can directly measure the position errors. The creation of this image, however, is mainly for debugging purposes.
The final step of the algorithm is to automatically highlight defect areas in the distance image. As mentioned previously a defect is a high intensity area above a threshold value. Input parameters determine the actual threshold and size values. Again threshold and size segmentation is applied to screen out valid defect areas whose parameters can be fine tuned in the training mode. Right now these parameters have been adjusted to the latest images, but it is expected to need a fine tune update of these parameters during the test phase. The result is a generated output image with clearly marked defects which are marked by red colour. The total processing time of the evaluation is today about 2 ~ 3 seconds per image.
The DIRAGREEN prototype has been integrated and validation tests have been carried out in laboratory conditions. The evaluation clearly shows that digital radiography can adequately capture information on real PM samples with enough clarity to view the real defects. These results have been fed into the optimisation of the final developed DIRA-GREEN prototype, which is a self-contained, enclosed and radiation-safe digital X-ray radiography system for the automatic inspection of green PM parts. The fact that the inspection system is enclosed means that it is ready to be used, in-line, in a PM factory. The next stage is to thoroughly test the prototype during field trials in a real PM factory environment. The DIRA-GREEN system has already been shipped to the PM production site of a consortium end-user partner in order to undergo these trials. Further tests and fine-tuning are still necessary before commercialization of the system which is expected in 2016 by the consortium.
Potential Impact:
The main novelty of the developed DIRA-GREEN system is the use of digital radiography for the inspection of green parts. The use of digital radiography has become more and more widespread as it is a much faster way of getting images of the inside structure of object than the traditional x-ray methods were.
The DIRA-GREEN system proposed to apply the digital radiography technology for fast inspection of green parts which can benefit the PM industry in two ways. Firstly, by getting an “instant” feedback about the internal structure of a part the press that produces it can be set up correctly in case there is a fault. Secondly, by identifying that a part is faulty at a green state they can be scraped instantly (or recycled) and time and energy is saved by not sintering them.
The main result is the DIRA-GREEN system, a non-destructive evaluation system based on digital radiography technique, which provides a rapid, reliable and thorough quality assurance of green parts, by monitoring defects in compacted parts. From a commercial point of view some fine tuning and an expended database is required before the DIRA-GREEN device and the related support/maintenance services can be regarded as a product.
Some additional results were also the outcome of the project which cannot be used on their own but are valuable sub components of the whole system: i; A supporting defect database was set up that can be used to relate defect to part parameters, ii; A novel part manipulator system that uses a tray system to protect the green part from damaging during the inspection was combined with a mechanical and control system to allow taking images from a numerous angles, iii; The evaluation system that uses a golden image for reference. When a picture is taken of a sample image, it is orientated to be in line with the golden image so the two images can be subtracted from each other so any defects can be identified.
All this development resulted in a prototype that can be further improved to become a commercial product and have the following benefits over other solutions:
- Applied in the green state after the part comes out of the press prior to sintering
- The part can be inspected in 1 minute (for the prototype) which can be reduced for the commercial product stage
- Cracks in the region of 20 microns can be found
- Density map can be identified through the digital image
- The system can be reconfigured for different parts in a matter of seconds as long as the inspected part has a golden image and the settings are stored by the system
In order to the socio-economic impact and the wider societal implications of the results an end-user questionnaire was prepared and a survey was carried out at the Euro PM2014 conference in Salzburg after the training session. The potential end-users of the DIRA-GREEN system were given a detailed presentation about the system including an operation video and were asked to give their opinion about the system.
The responses were relatively positive considering the system is still in the prototype phase. Regarding its potential as a technology to improve tool design most responses agreed that it has a high potential but improvements are necessary. Similar responses were on the potential to improve the manufacturing process and the quality control. 85% of the responses also agreed that they would like to test the system on their own parts (some have already contacted the consortium).
It was quite apparent that the main benefit of the system is the fact it can carry out inspection before sintering. It was also highlighted that the system has many other benefits such as scrap reduction and also used for inspecting precious metal pieces.
Overall the feedback was very positive and useful. It is apparent that the PM industry needs such system and even though some improvement is necessary to the system before it can be used commercially the DIRA-GREEN prototype demonstrated that the digital radiography methodology is suitable for the inspection of green parts.
List of Websites:
www.diragreen.eu
The powder metallurgy (PM) industry has long been feeling the need for a reliable method to inspect parts and detect local defects in the green state. The required test should be non-destructive, fully automatic, cost effective and fast enough to be integrated in a parts production line. Known inspection systems such as eddy current and magnetic bridge testing, magnetic particle inspection, acoustic resonance and ultrasonic testing can be applied to sintered parts, but fail when applied to green compacts. Optical inspection with cameras can only identify surface defects and cannot detect internal defects arising due to material porosity. X-ray computer tomography (CT) is a powerful tool even for green compacts, but due to the high cost of the equipment it is often commercially unattractive and it cannot be used in-line due to long processing times. Therefore a simpler, faster and less costly alternative was developed by DIRA-GREEN.
The goal of the DIRA-GREEN project was to build a digital radiography (DR) inspection system such that it responds in real time and can be incorporated into production lines for automatic defect detection in green state parts. The quality of the manufactured PM parts should then be assured by monitoring for compacted material porosity and identifying microscopic features such as cracks. The collected radiographic images for the PM components obtained using the DIRA-GREEN system should show the size and location of any flaws present. These images could be stored in a component-linked database and would facilitate an improvement cycle for the optimisation of powder metallurgy mould and die designs. Also, identifying defects before the parts are sintered would allow recycling of the powder material. The availability of a reliable and attainable inspection tool for detecting green part flaws would result in a substantial increase in the competitiveness of European small and medium PM enterprises (SMEs). It is expected that an increasing confidence in the supply of the final PM product can be achieved by improving the reliability of powder metallurgy parts and reducing costs associated with the production and control of the parts.
Project Context and Objectives:
The objective of this project was to develop a Non-Destructive Testing (NDT) technique, which enables online quality assurance of ‘green parts’, by monitoring compacted material porosity, and identifying microscopic cracks. This innovative tool uses digital radiography to create a density map for each component, indicating the size and location of defects.
The scientific objectives of the project were:
• To gain a greater understanding of digital radiography technology by using multi-dimensional, low frequency impedance measurement to assess porosity, cracks, and defects of P/M parts.
• To develop a deeper understanding of the type of defects present in ‘green parts’.
• A potential secondary scientific objective is to test DIRA-GREEN’s ability to inspect ceramics and P/M parts manufactured with high resistive materials.
The technological objectives of the project were:
• To design a unique inspection tool prototype for the assessment and quality control of internal defects in ‘green parts’ that is reliable, low cost, rapid, and easy to use (automated), thereby fulfilling the needs of the P/M SME industry, which currently employs inefficient/costly manual inspection techniques.
• To design the digital radiography subsystem including an electronic subsystem capable of generating and measuring the signals needed for operation of the digital radiography system.
• To implement a control system, along with the mechanical design, that ensures compatibility and ease of use in industry applications.
• To ensure the DIRA-GREEN inspection system is high flexible via the software control system, which allows reconfiguration within minutes.
• To implement a data processing and evaluation system.
• To create a software system that allows for the collection of data from defected parts, and the ability to store it in a database for use in ‘improvement cycles’.
• To perform the validation of DIRA-GREEN tool that is capable of identifying defects before sintering, thereby reducing scrap, and saving time and energy.
Project Results:
An integrated DR system, complete with an X-ray source controller, the external PC connection, power supply unit, interlocks and warning lamps, is shown in a diagrammatic representation in attachment Figure_1.jpg. During an inspection, a PM part is placed between an X-ray source and the digital detection media. The penetration of X-rays through the sample will depend on the X-ray energy as well as the thickness and density of the PM material under inspection. The penetrating X-ray quanta are detected using a digital X-ray detector. The detector is controlled, and images obtained, using a wired connection to a PC. As the X-rays pass through the component, any changes in bulk density, due to cracks, porosity, inclusions, or simply component thickness, attenuate the X-ray quanta in proportion to the material density or thickness. The digital detector pixel elements translate the received quanta to image pixel values. The detector software shows a dense region as dark, and less dense areas as light, and this can be used to image any defects.
The DIRA-GREEN radiography system was arranged as shown in attachment Figure_2a.jpg and Figure_2b.jpg. It mainly comprises a micro-focus X-ray source emitting X-rays vertically, a sample held by a mechanical manipulator placed in line of the X-rays, and an indirect flat panel digital detector. X-ray inspection starts with a sample being illuminated by an X-ray source which is held in a fixed position statically so as not to induce any mechanical vibration and project distortion onto the detector. PM components enter the enclosure on a conveyor belt system and are positioned on a purpose built manipulator in between the source and the detector. The sample holder and the detector are mounted on a vertical translator, allowing the source-to-detector distance (SDD) and the source-to-object distance (SOD) to be varied to alter the magnification of the resultant radiographic image. The manipulator can also change the orientation of the PM sample with respect to the X-ray beam. The sample holding pallets have been constructed using a homogeneous material such that it does not contribute a visible structure on the radiograph, potentially masking defects.
Essential components of the DIRA-GREEN prototype
A micro-focus X-ray source is an integrated solution encompassing the tube, high voltage power supply and control circuitry. It is controlled by PC via the source controller. The micro-focus X-ray source produces an ellipsoidal cone X-ray beam with beam angles of 120O/40O. This beam angle ensures that most parts to be imaged can be fully covered by the emitted X-rays. The degree of sharpness obtained from a radiographic image is determined by the focal spot size. The focal spot generated by the X-ray source of the DIRA-GREEN system is as small as 4 microns diameter. This is designed to reduce image blurring when compared to conventional mini-focus X-ray sources, and thus provides an enhanced flaw detection capability and sensitivity. The source size contributes to the radiographic quality, influences the geometry of the inspection, and sets radiographic image definition and resolution limits. With these factors in mind, as well as a detection accuracy of 0.5 mm required by the industry, an X-ray source with these specifications has been used in the DIRA-GREEN system.
The component generating the X-ray images is a flat panel detector. For a typical 40 x 40 mm sample provided by one of the consortium end users, and using 3x magnification, a detector field of view with a minimum 120 x 120 mm detection area is required. A survey was carried out in the X-ray detector supplier market to find a suitable flat panel detector. An indirect digital X-ray detector with a pixel resolution of 74.8 µm and a detection area of 145.4 x 114.9 mm resulting in an array of 1944 x 1536 pixels was chosen for the project. Attachment Figure_3.jpg shows the detector mounted on a frame. The detector comprises a scintillator material that converts X-rays into visible light photons, a fibre optic plate to guide the light and block the X-rays, and a CMOS sensor to convert the light into an electrical charge. With exposure times of 1 to 2 seconds the system is fast enough to be integrated in a parts production line.
The mechanical subsystem was designed for the manipulation of the inspected parts. The green part is placed on a pallet before it enters the lead chamber on a conveyor belt. A conveyor belt system driven by an electric motor has been constructed to accommodate specially designed pallets with interchangeable inserts to hold PM parts of varying shape and size. The conveyor feeds the parts into the X-ray chamber and presents the sample holding pallet to a customised mechanical manipulator which is located between the X-ray source and the detector, as depicted in attachment Figure_2a.jpg. Once inside the chamber, a manipulator takes the pallet from the conveyor and moves it into position for inspection. In order to detect non-planar defects, the manipulator allows orientation of the sample part between -45o and +45o to the X-ray beam, as shown in attachment Figure_2b.jpg. In addition the manipulator system allows the inspected part to be positioned such that its radiographic image can be magnified between 1.72x and 4.69x. These geometries were taken into account during the radiographic laboratory trials and optimisation. The best magnification depends on the size of the part and the size of defects to be detected. A magnification of 1.72x results in a maximum achievable image resolution of 35 microns and a magnification of 4.69x enhances the image resolution to 20 microns.
The pallet is made of a low density, homogeneous material such that it does not significantly attenuate the X-rays, or produce a pattern on the image which could be misinterpreted as, or mask, a defect. Several images are taken, for this purpose the manipulator can rotate the parts and change the aspect angle. Once the images are taken, the pallet is placed back onto the exit conveyor.
The DIRA-GREEN mechanical subsystem concept has been significantly improved compared to the originally proposed system. It became obvious that because of the wide range of part characteristics a universal gripping device is not applicable. Accordingly the task to develop a universal gripping tool is in excess of the DIRA-GREEN project scope. Therefore to make the DIRA-GREEN prototype simultaneously suitable for small and fragile MIM parts and more robust PM parts, a special pallet system had to be developed.
The factory personnel need to be protected from the ionising X-radiation produced by the X-ray source. For this reason a lead-steel enclosure was designed and constructed to surround all components of the system and prevent the radiation from escaping to the surrounding areas. The lead chamber for the X-ray protection has two sides, a top, bottom and back panels shielding the enclosure. On the front side there is a windowed door turning along the horizontal axis hence, enabling it to open in the upward direction. This door has a lead glass insertion which allows seeing inside the enclosure without the risk of exposure to X-radiation. There is also a double door at the front which locks the window door from opening while the X-ray source is in use.
The enclosure has been designed in accordance with the UK guidelines set out by the Health and Safety Executive (HSE) through their Approved Codes of Practice (ACOP), and by adherence to ‘The Ionising Radiations Regulations 1999 (IRR99)’ [2] made under the Health and Safety at Work Act 1974. A warning lamp tower is fitted ensuring information about the state of the system is communicated to personnel in the vicinity, and all openings are fully interlocked to guarantee that X-rays cannot be produced if a door is not fully closed. The enclosure, seen in attachment Figure_4.jpg has been tested and certified for safe operation by a designated Radiation Protection Adviser.
In order for the DIRA-GREEN system to be integrated in a production line, within a PM factory setting, a design as depicted in attachment Figure_5.jpg was chosen. Parts coming from the powder compaction press are fed by a conveyor system into the lead chamber. Inside the chamber is the radiation source, the manipulator and the flat panel detector. The inspected good parts leaving the lead chamber to the right are transferred to the sintering furnace.
DIRA-GREEN System Performance
The performance of the DIRA-GREEN system was first evaluated under laboratory conditions using standard methodologies. Trials on the radiographic equipment were also carried out on the stand-alone pieces to optimise the conditions and inspection parameters in the final integrated system.
The initial equipment testing was carried out in the existing general purpose laboratory radiography bay at TWI (Wales). At all times the necessary dark, gain and defect correction algorithms were processed to ensure clarity of the image. The performance of the X-ray equipment was evaluated by quantifying the focal spot size of the X-ray source and testing the actual pixel resolution of the detector. To perform these tests two different Image Quality Indicators (IQI) were used: a specialised high resolution line pair micro-chart, JIMA RT RC-02B, for the focal spot size measurement and a Duplex IQI (BS EN 462-5) [3] for the detector resolution.
JIMA RT RC-02B contains line pairs of varying distance ranging from 0.4 to 15 μm. The IQI was placed as close as possible to the source target to obtain the highest possible magnification in order to determine which of the micrometer-scale line pairs could still be distinguished as two lines. The radiographic images were obtained at X-ray source energies of 45kV. The 3 μm line pair was visible, as shown in Attachment Figure_6.jpg therefore 3 μm was inferred as the measure of the minimum focal spot size.
The pixel size of the digital X-ray detector as specified by the manufacturer is 74.8μm. The effective resolution of the detector was quantified using a duplex line-pair IQI (contains 13 line pairs ranging from 130 to 50 micron distance) by placing it on the surface of the detector at 1x magnification. In accordance with BS EN 462-5, the first unresolved line pair is then used as a measure of the basic spatial resolution. In Attachment Figure_7.jpg displays the radiographic image of the Duplex IQI including the intensity line profile drawn across the three smallest line pairs. It shows that 11D is the first unresolved line pair providing a measure of 80 micron for basic spatial resolution. This value is expected and proves that the pixel resolution of this particular detection is 74.8 μm.
The overall functionality and performance of the DR system was verified by inspecting PM parts with real defects. The green PM samples were inspected at a source-to-object distance that provided the greatest magnification whilst keeping the whole sample in the field of view of the detector. Experiments have been carried out using specific tube voltages and current, exposure times, PM sample orientation etc.
A radiograph of a green PM compact with intentionally produced defects is shown in Attachment Figure_8.jpg. The red circle indicates a greater thickness due to the sample holder. The blue circles show a crack at the top, and the green circles show loss of material on the left side edge and a crack on the right edge. The radiographic system has enabled inspection for differentiating local density changes of 2% in the PM parts as a target resolution.
Image processing system
For an automatic evaluation of part defects it is not sufficient just to acquire excellent images, but a suitable software evaluating the images is equally important. The task of the image processing module is to handle the images recorded by the digital radiography system and evaluate them with respect to defects. The X-ray detector produces the data in a matrix, allowing the measurement to be handled as images. Therefore, image-processing algorithms have been developed to identify discrepancies, voids and cracks within the internal structure of the material.
The evaluation software developed by DIRA-GREEN compares the recorded image of a compact to a reference (“golden image” - image of a perfect part). To be able to detect cracks with any orientation, several images are taken with different pallet rotation and tilting offset values. For each angle a separate golden image is required and each recorded image is compared to the corresponding golden image.
Attachment Figure_9.jpg shows a golden image of a MIM part and a recorded image of another MIM part from the same mould with a small area of reduced density. With the naked eye it is impossible to detect the defect in the second image. We will demonstrate how the image processing software is able to bring out this defect.
Image processing involves several steps. The first step is required to remove the background and noise from the input images and to effectively assign the area of comparison. This is done through threshold segmentation with noise elimination. Both the recorded image and the golden image undergo this process.
One of the main challenges of the image comparison in DIRA-GREEN was that the golden image and the recorded image had a slightly different position, orientation and scale due to mechanical tolerances of the manipulator hardware. To overcome this problem, contour images are created by eliminating all information of the part bodies (Attachment Figure_10.jpg). The two contour images are superimposed and aligned with a best fit (Attachment Figure_11.jpg). With the new contour coordinates a remapping of the recorded image is executed.
After remapping the recorded image, a distance image is created by computing the distance between each two corresponding pixels. High intensities in the distance image identify the possible defects, as shown in Attachment Figure_12.jpg. The purpose of generating the distance image is to create a matrix of data where each cell intensity represents a distance value from the ideal contour. A cell value is low/dark near the contour, and getting lighter as the cell is getting away from the ideal value. The generated distance image is measuring how far a transformed pixel is from the golden contour. With the contour distance image the user can directly measure the position errors. The creation of this image, however, is mainly for debugging purposes.
The final step of the algorithm is to automatically highlight defect areas in the distance image. As mentioned previously a defect is a high intensity area above a threshold value. Input parameters determine the actual threshold and size values. Again threshold and size segmentation is applied to screen out valid defect areas whose parameters can be fine tuned in the training mode. Right now these parameters have been adjusted to the latest images, but it is expected to need a fine tune update of these parameters during the test phase. The result is a generated output image with clearly marked defects which are marked by red colour. The total processing time of the evaluation is today about 2 ~ 3 seconds per image.
The DIRAGREEN prototype has been integrated and validation tests have been carried out in laboratory conditions. The evaluation clearly shows that digital radiography can adequately capture information on real PM samples with enough clarity to view the real defects. These results have been fed into the optimisation of the final developed DIRA-GREEN prototype, which is a self-contained, enclosed and radiation-safe digital X-ray radiography system for the automatic inspection of green PM parts. The fact that the inspection system is enclosed means that it is ready to be used, in-line, in a PM factory. The next stage is to thoroughly test the prototype during field trials in a real PM factory environment. The DIRA-GREEN system has already been shipped to the PM production site of a consortium end-user partner in order to undergo these trials. Further tests and fine-tuning are still necessary before commercialization of the system which is expected in 2016 by the consortium.
Potential Impact:
The main novelty of the developed DIRA-GREEN system is the use of digital radiography for the inspection of green parts. The use of digital radiography has become more and more widespread as it is a much faster way of getting images of the inside structure of object than the traditional x-ray methods were.
The DIRA-GREEN system proposed to apply the digital radiography technology for fast inspection of green parts which can benefit the PM industry in two ways. Firstly, by getting an “instant” feedback about the internal structure of a part the press that produces it can be set up correctly in case there is a fault. Secondly, by identifying that a part is faulty at a green state they can be scraped instantly (or recycled) and time and energy is saved by not sintering them.
The main result is the DIRA-GREEN system, a non-destructive evaluation system based on digital radiography technique, which provides a rapid, reliable and thorough quality assurance of green parts, by monitoring defects in compacted parts. From a commercial point of view some fine tuning and an expended database is required before the DIRA-GREEN device and the related support/maintenance services can be regarded as a product.
Some additional results were also the outcome of the project which cannot be used on their own but are valuable sub components of the whole system: i; A supporting defect database was set up that can be used to relate defect to part parameters, ii; A novel part manipulator system that uses a tray system to protect the green part from damaging during the inspection was combined with a mechanical and control system to allow taking images from a numerous angles, iii; The evaluation system that uses a golden image for reference. When a picture is taken of a sample image, it is orientated to be in line with the golden image so the two images can be subtracted from each other so any defects can be identified.
All this development resulted in a prototype that can be further improved to become a commercial product and have the following benefits over other solutions:
- Applied in the green state after the part comes out of the press prior to sintering
- The part can be inspected in 1 minute (for the prototype) which can be reduced for the commercial product stage
- Cracks in the region of 20 microns can be found
- Density map can be identified through the digital image
- The system can be reconfigured for different parts in a matter of seconds as long as the inspected part has a golden image and the settings are stored by the system
In order to the socio-economic impact and the wider societal implications of the results an end-user questionnaire was prepared and a survey was carried out at the Euro PM2014 conference in Salzburg after the training session. The potential end-users of the DIRA-GREEN system were given a detailed presentation about the system including an operation video and were asked to give their opinion about the system.
The responses were relatively positive considering the system is still in the prototype phase. Regarding its potential as a technology to improve tool design most responses agreed that it has a high potential but improvements are necessary. Similar responses were on the potential to improve the manufacturing process and the quality control. 85% of the responses also agreed that they would like to test the system on their own parts (some have already contacted the consortium).
It was quite apparent that the main benefit of the system is the fact it can carry out inspection before sintering. It was also highlighted that the system has many other benefits such as scrap reduction and also used for inspecting precious metal pieces.
Overall the feedback was very positive and useful. It is apparent that the PM industry needs such system and even though some improvement is necessary to the system before it can be used commercially the DIRA-GREEN prototype demonstrated that the digital radiography methodology is suitable for the inspection of green parts.
List of Websites:
www.diragreen.eu
