Japan Foundry Society, Inc. Application of Recent X-ray CT Technology to Investment Casting field Kouichi Inagaki ICC / IHI Corporation 13 th WORLD CONFERENCE ON INVESTMENT CASTING Paper: T3 Copyright reserved: Neither the Japan Foundry Society, Inc. nor its officers accept legal responsibility for information, advice given or opinions expressed.
Application of Recent X-ray CT technology to Investment casting field K. Inagaki 1 Y. Saitou 2 H. Tohyama 3, H. Honda 3 1Materials Technology Department Aero-Engine & Space Operations IHI Corporation 2 Materials Department Corporate Research & Development IHI Corporation 3 Enginnering Department IHI castings Corporation Abstract X-ray computed tomography (CT) technology has been applied in medical field from early 1970s and in industrial field from 80s. Primary application in industrial field was for inspecting or observing the inner structure and defects. Recently, it has also been applied to acquire actual 3D CAD data and for dimensional measurement (1). This paper describes the utilization of actual 3D CAD data and the application of dimensional measurement for investment casting. 1. Introduction These days, X-ray CT has been applied for dimensional metrology. X-ray CT is the only method able to measure and acquire not only outer structure but also inner structure without need to destroy it. The dimensional metrology in investment casting field, especially for comparably absorbing materials of X-ray such as Ti and Ni alloy castings and for thicker casting parts, calls for higher penetrating power and higher resolution CT system. Recent development achieved to produce higher penetrating power and higher resolution X-ray CT system. This system will expand the utilization for investment casting field. For example, utilization for casting simulation, shape comparison (between CAD model and actual 3D volume data by X-ray CT), dimension measurement and so on are assumed. In this paper, utilization for casting simulation and dimension measurement are discussed. Casting simulation is widely used to establish a manufacturing process of casting. In the simulation,castingmoldis generalyuseduniformcadmodel.so,themodelcan t realize the actual casting mold (e.g. some voids and non-uniform thickness). The difference between uniform and actual (non-uniform) model would affect partially the accuracy of the simulation. To propose the improvement of casting simulation accuracy, actual CAD model is acquired by X-ray CT and applied for the simulation, since X-ray CT system can produce CAD model of actual casting mold which has outer structure, inner structure, inner voids and non-uniform thickness. In dimension measurement of wall thickness in investment casting, for example hollow turbine blade, UT is generally used. However, UT is affected by human factor and secondary crystallographic orientation. This paper proposes to use X-ray CT for wall thickness measurement. Since X-ray CT data can be acquired automatically and its data is digital, CT would be less affected by human factor than UT. And CT scan is generally not affected by secondary crystallographic orientation.
2. X-ray CT system The X-ray system used in this study consists of X-ray source, rotating table, X-ray detector, reconstruction unit and measurement unit. The X-ray source has higher power 450kV acceleration voltage. Two types of detectors are used. One is curved line detector array (CLDA) consisting of 1D array pixels. The other is flat panel detectors consisting of 2D array of pixels. The detector is selected according to the purpose of the study. Since the flat panel size is about 400mmx400mm area, large area can be detected at once. Since the object for the study of casting simulation is large, flat panel detector is used. CLDA is consisted with solid elements which can detect higher energy X-ray photons effectively and it realizes higher measurement accuracy in higher energy. For the study of dimensional measurement, higher energy X-ray is needed, therefore CLDA is used. However, CLDA is line array detector so inspection area is restricted compared with flat panel detector. In the scanning of this study, the X-ray source and the detector are fixed, while the object is put on the turntable and rotated 360deg between X-ray source and the detector. 1000 or more X-ray perspective images of different angles are obtained by detector. Prior to scanning, inspection parameters such as acceleration voltage, current, projection number and so on, are fixed. After inspection, acquired perspective images are reconstructed into 2D cross-sectional image or 3D volume data. There are also parameters to be fixed in the reconstruction process such as resolution level and so on. After the reconstruction, the 2D cross-sectional image or 3D volume data is displayed on the PC monitor by the specific 3D software. To fix scanning and/or reconstruction parameters which affect dimension accuracy and spatial resolution, standard cylinder test piece for dimension accuracy and slit test piece for spatial resolution are scanned prior to part scanning. Fig.1 X-Ray CT data (left: Cylinder test piece, right: slit test piece) 3. As application for quantitative evaluation, improvement of Casting simulation accuracy by actual 3D volume image acquired by X-Ray CT In conventional casting simulation, 3D CAD model of uniform mold is used. However, actual casting mold is not uniform: some voids are included and shell thickness is not constant. If this difference can be compensated, casting simulation accuracy will be improved. In this paper, to compensate for this difference, the actual casting mold model
of 3D volume data by X-ray CT is applied for casting simulation. The evaluation object is the cast mold for simple test piece (for example, tensile or HCF test piece. Fig.2 left). After scanning and reconstruction of the test piece, 3D volume data which includes inner voids and presents non-uniform shell (figured in Fig.2 right) was acquired. Fig.2 Figure of casting mold (Left) and actual casting mold model acquired by CT (Right) 10M Acquired CT volume STL format 360MB 6M CAD B Mesh data Fig.3 Sample result of casting mold (2) Since the data includes inner voids and presents non-uniform shell, the data volume was 36GB. The data was too large to be imported to casting simulation software. The data needed to be reduced its data volume. And, since the data of the actual casting mold model of 3D volume data by X-ray CT was STL format, the data needed to be translated into mesh data.
Point 1 Point 1 Point 4 Point 2 Point 4 Point 2 Point 3 Point 3 Left: Simulation from CT volume. Right: Simulation from CAD data Fig. 4 Comparison with simulation from CT volume and CAD data These days, reverse engineering software has achieved remarkable development and some software are released. For this study, most appropriate reverse engineering software Simpleware was searched and selected. This software can translate 3D volume data by X-ray CT (e.g. STL format, polygon format and so on) into 3D mesh data and reduce data volume. Fig.3 shows the result of translations conducted by Japanese agent JSOL Corporation. The acquired STL data of cast mold was translated into PATRAN format mesh model by Simpleware.Andthen,themeshmodelwas importedtocastingsimulationsoftware. Fig.4 left shows the simulation result of casting mold model acquired by 3D volume data by X-ray CT. Fig.4 right shows the simulation result of uniform CAD model. Fig.4 shows that shrinkage positions anticipated by X-ray CT data and uniform CAD model is consistent each other, however, differences are obtained in shrinkage occurrence risk and cooling rate. This result indicates that using actual model improve the casting simulation accuracy, however actual verification is needed. 4. The utilization of X-Ray CT for the dimensional measurement using comparative evaluation method To apply CT for dimensional measurement, there are still some problems (e.g. accuracy, reliability) to be solved. One of problems is how to define the correct boundaries of a part
in 3D volume data acquired by CT. This paper proposes to use comparative evaluation method to solve the problem for the specific parts. 4.1 Test piece Five test pieces assumed Ni hollow turbine blade of single crystal were manufactured. Each test piece had eight different wall thickness areas to be measured. Since comparative evaluation was applied in this study, No1 (Fig.5) test piece was fixed as reference standard. No3, No4 and No5 test pieces had same shape as No1 but the wall thicknesses were different from each other. Detail thickness differences are listed on Table1. The differences are nominal value. Actual dimension and its difference were measured by CMM. No2 test piece had the same shape and wall thickness as No1 but its secondary crystallographic orientation was different direction from No1, No3, No4 and No5 had same secondary crystallographic orientation. Fig.5 Test piece simulated hollow turbine blade No.1 RV. No.2 Same as RV. Table1. Wall thicknesses Wall thickness No.3 Wall thickness for each area is RV + 0.05mm. No.4 Wall thickness for each area is RV + 0.10mm. No.5 Wall thickness for each area is RV + 0.15mm. Recommend Secondary crystallographic orientation is different from others. RV: Reference value 4.2 CT data Fig.5 shows the 2D cross-sectional image of No1 standard test piece. The 2D image is the distribution of gray value which expresses the X-ray attenuation. Fig.7 shows the gray value profile on the a-b line written in Fig.6. As Fig. shows, the boundary between the air and the object changes with a gradient. In this study, the boundary was defined as center of the object and the air gray value (Iso-50%) in the histogram plotting the number of voxels versus the gray value (See Fig.8). 4.3 Measurement The measurement accuracy was calibrated by the standard cylinder test piece and the slit test piece. But non-symmetric shape of the object and some CT noise affected the measurement accuracy. So, the error was normalized by the No1 standard test piece.
Measurement positions of wall thickness are illustrated infig.6. Table2 shows the deviation of the wall thickness measured in CT data from CMM. In this study, dimensionalacuracyachieved±0.03mm or less. This result showed CT measurement accuracy is about same as that of ultrasonic inspection. Maximum deviation for No2 test piece which had different secondary crystallographic orientation from No1 was less than ±0.03mm. In the measurement of No2 test piece, crystal orientation correction was not applied. This result indicated that unlike ultrasonic inspection CT scanning was able to measure the dimension without being affected by crystal orientation. area-c area-d a area-b area-f area-a area-e b area-g area-h Measurement position Fig.6 CT result of the simulated hollow turbine blade test piece b a Fig. 7 Gray value profile of the line in Fig.4 Material Iso-50% Background Fig. 8 Histogram plotting the number of voxels versus the gray value (No.1 Sample)
Table2. Measurement result of deviation from CMM result Unit:mm area No.1 *1 No.2 *2 No.3 No.4 No.5 A 0.00-0.01 0.01-0.01-0.01 B 0.00 0.01-0.01-0.01-0.02 C 0.00 0.01 0.00 0.01 0.01 D 0.00 0.01 0.01 0.00 0.00 E 0.00 0.00-0.03 0.01-0.01 F 0.00 0.01-0.01 0.01-0.03 G 0.00 0.00 0.01-0.02 0.02 H 0.00 0.00-0.01 0.00 0.00 *1: No.1: RV *2: No.2: Secondary crystallographic orientation is different from others. 5. Summary (1) Simulation results from uniform CAD model and from 3D volume data by X-ray CT showed some differences. This result indicates that using actual model improve the casting simulation accuracy. (2) Dimensional measurement by CT indicated the enough accuracy to be replaced with ultrasonic measurement. References (1) J.P. Kruth et al, 2011, Computed tomography for dimensional metrology, Manufacturing Technology 60 p821-842. (2) InformationonSimpleware wi lbeadiversionarticlefromawebsiteofjapan agency corporation JSOL (http://simpleware.jsol.co.jp/index.html).