Home

fulltext - DiVA Portal

Contents

1. Nondestructive measurements of fracture aperture in crystalline rock cores using x ray computed tomogra phy Journal of Geophysical Research 98 98 B2 1889 1900 Ketcham and Carlson 2001 Ketcham R A and Carlson W D 2001 Acquisition optimization and interpretation of x ray com 67 puted tomographic imagery applications to the geosciences Com put Geosci 21 4 381 400 Masad et al 2005 Masad E Tashman L Little D and Zbib H 2005 Viscoplastic modeling of asphalt mixes with the effects of anisotropy damage and aggregate characteristics Mechanics of Materials 37 12 1242 1256 McNitt Gray 2006 McNitt Gray M 2006 Mo a valb 01 Trade offs in image quality and radiation dose for ct Medical Physics 33 6 2154 2155 Schlangen and van Mier 1992 Schlangen E and van Mier J 1992 Simple lattice model for numerical simulation of fracture of concrete materials and structures Materials and Structures 25 534 542 Wang 2010 Wang L 2010 Mechanics of Asphalt Microstructure and Micromechanics McGraw Hill Wang et al 1999 Wang Z M Kwan A K H and Chan H C 1999 Mesoscopic study of concrete i generation of random aggre gate structure and finite element mesh Computers and Structures 70 5 533 544 Wikipedia Wikipedia Mass attenuation coefficient Wittmann et al 1985 Wittmann F H Roelfstra P E and Sadouki H 1985 Simulation
2. c the emission of x ray Check if the full image is formed on detector and if not you have to make necessary alignment changes These changes can be made using the control gear The process of re alignment of the detector X ray source and the scanning platform can be monitored using the CCTV monitor outside the scanning chamber There are four cameras installed in the scanning chamber Changing the Scan settings The energy intensity at which the scan is to be performed can be changed on the FXE Control application while the setting of the amount of details to capture can be changed on the CT rotate 225 application a b Settings on the FXE Control application The scanning voltage and current can be adjusted by moving the voltage slider and current slider The sliders for the voltage and the current are independent of each other The voltage corresponds to the energy intensity of the X ray The adjustment should be made until one gets a good scan image on the computer screen A too dark image implies a bad scan and a too light image indicates that some details might have been lost during the scan This process is very important in the scan process and requires the judgment of the person carrying out the scan The linear attenuation coefficient of the materials to be scanned can be used to get an idea of the appropriate energy level required for the scan The changes made to the energy intensities are immediate
3. during a single 360 degree of rotation The balls of the calibration tool must be at least 1 ball diameter from the edge of the image It is also beneficial but not necessary to have the calibration tool reach from the bottom of the detector to the top If the tool does not span from the top to the bottom place the tool so that one portion of it reaches the top or bottom This allows the creation of larger ellipses during rotation and better data for the software to build the volume off of 6 35 mm f p jb 1 59 mm _ 15 mm 40 mm 5mm t 1 mm Small Medium Large CT calibration procedure a Place the calibration rod on the scanning platform There are three sizes of calibration tools to choose from large medium and small Tool size used is determined by the magnification used during acquisition and detector size b On CT Project window click on CT Calib and select the 15mm large tool if you are using the large tool for the calibration else choose what corresponds to your choice and then click OK Then the calibration process begins TSC MT 13 002 www kth se
4. separating the particles is calculated using the fastwatershed command 36 in avizo This principle of the fastwatershed is called immersion and it is based on a simulation of the rise of water from a set of markers The level p is flooded at a uniform speed from the fronts coming from level p 1 and the local minima appear at level p A point of the watershed appears when two distinct fronts join Avizo 2009 watershed The curve represents A is flooded at levels 1 A and C are flooded until three minima A B and and 2 but not B Then reaching point D D be Candtwomaxima D and jt reaches point C longs to the watershed E The set of markers but not E contains only A and C c From the next level Figure 29 The fastwatershed principle Avizo 2009 The fastwatershed command takes the inverted distance map and the merged maxima markers as input images Figure 30 shows the markers used for the watershed segmentation and Figure 31 shows the watershed lines calculated from the watershed operation The water shed lines is used for aggregate separation and determination of contact points between adjacent aggregates 37 Figure 30 Markers used for watershed segmentation Figure 31 Watershed lines used for separation of stones 3 6 Surface reconstruction After proper segmentation of the constituents of the AC mix sur faces were generated for visualization of the AC microstructure Fig ure 32 shows the final segm
5. 48 and 49 It can also be observed 95 that only load transfer regions in the binder are subjected to tensile stress greater that 10MPa The maximum tensile stress in the mas tic is 64 5MPa for the multi phase model while the maximum tensile stress in the continuum model is about 1 68MPa There is a significant increase in strength of the AC sample when modeled as a multi phase and the stress distribution is well captured From Figures 50 and 51 it can be seen that the strains are localized in the binder for the multi phase continuum model while for the continuum model the strains are continuous It can be seen that in the continuum model the strains dis tribution shows a regular pattern with decreasing tensile strains from the top to the bottom while the strain distribution in the multi phase model is dependent on the arrangement of the aggregates in the mix Table 6 shows the values of the maximum tensile stress and strain in the mastic for the continuum and multi phase model It can also be observed that the multi phase model takes up strains up to 1296 strain as compared to the continuum model where approximately 0 2 strain is predicted Surface Strain tensor Y component 1 w A 2 402x10 x10 25 30 35 40 45 50 S39 60 65 70 75 80 85 90 95 v 1 0984x10 Figure 50 Continuum model Strain distribution in AC microstructure 56 Time 1 Surface Strain tensor Y component 1 a T T T T T A 0 1261 90 0 01
6. 5 1 Thresholding ae ed ae res cereus m 3 5 2 Edge detection 2 0 220 ex ASA RR S 3 5 3 Distance Map approach uw osx moe 3 5 4 Watershed segmentation 3 6 Surface reconstruction ose e Boom eus s Digital Image Analysis and Results Finite Element Method FEM analysis and results 5 1 Two dimensional 2D finite element analysis 5 1 1 Two dimensional 2D uniaxial tension 5 1 2 Two dimensional 2D thermal analysis 5 2 Three dimensional 3D uniaxial compression 6 Conclusions List of Figures 13 14 15 16 17 18 19 20 21 22 Process workflow yc 8 248 a a 5 NSI X 5000 X ray CT facility in KTH Highway and Railway tab neue cram ag boat oh ay rA e mv woe A T Linear attenuation coefficient for asphalt concrete 10 Porous asphalt concrete sample used in the present study 13 Pixel representation in a digital image 15 Neighbourhood relationship 17 Subvolume of acquired image from CT scan 18 Histogram of acquired image from CT scan 18 Image contrast equalization technique 19 Contrast enhanced image of Asphalt concrete 20 Variation of gray level before image correction and noise Henn se gras cette qe a RS oed PSU AR okt ae ee dh tr DR NS e 20 Illumination profile of CT scanned image of Asphalt CONCTETE saute MA quar Jue en s doe hk Poe ke ace a 21 Beam hardening corrected image of Asphalt concrete 22 Result of
7. 85 0 009 80 5 0 008 70 0 007 65 60 0 006 55 0 005 50 0 004 Ra j 0 003 40 0 002 35 30 4 0 001 25 L E L 4 J 0 30 40 50 60 70 80 90 v 0 0328 Figure 51 multi phase model Strain distribution in AC microstructure Table 6 Maximum tensile stress and strain Model Max tensile stress Mpa Max tensile strain Continuum 1 68 0 024 Multi phase 64 5 0 126 difference 97 4 81 0 5 1 2 Two dimensional 2D thermal analysis Two dimensional 2D therm0 mechanical analysis is carried out to investigate the thermal stress and strain developed in the AC mi crostructure as a result of cooling and temperature variation in the AC microstructure In this case the AC microstructure is considered as a multi phase model with the mastic assigned viscoelastic mate rial properties and the aggregates assigned elastic material properties The mastic viscoelastic behaviour is modeled using the generalized Maxwells model with 3 branches The generalized Maxwells model represents the viscoelastic behaviour of the mastic as a series of spring 57 and dashpot pairs The generalized Maxwells model is shown in Figure 52 The viscoelastic material parameters used for the mastic is pre sented in Table 7 The aggregate elastic material properties in Table 5 is assigned to the aggregates for the thermal analysis Figure 52 Stress distribution in AC microstructure Table 7 Viscoelastic material parameters m Ts
8. and analysis The X5000 CT scanner is a seven axis universal x ray imaging system designed for the inspection of large objects It can accommodate a variety of part shapes sizes and weights It can produce X ray inten sities of up to 450kV The KTH X5000 CT system is shown in Figure 2 The asphalt concrete core sample with a diameter of 100mm and a height of S0mm as shown in Figure 4 is scanned The sample is scanned at an energy intensity of 225kV without beam filtration The scanning resolution is 1949 x 1799 with a slice thickness of 59microns and a total of 1932 slices The x ray scanning process includes sample preparation warming up the scanner pre scan settings scanning detector calibration and CT calibration Beam hardening artifact is manifested in CT images with brighter edges than the center of the image Beam hardening artifact reduces the quality of the scanned image and hence affects the phase segmentation results Tis artifact is corrected using digital image processing techniques in Avizo R Fire Other digital image pro cessing and analysis is also performed using Avizo Fire application The procedures for the 225kV and 450kV scanning configurations 12 are presented in the appendix A Figure 4 Porous asphalt concrete sample used in the present study 13 3 Digital image processing and analysis Image processing mainly involves editing and enhancement of dig ital image with the aim of improving the qualit
9. application window click on the Startup button to initiate the warm up process NB Startup includes warming up and this prepares the machine for the actual scan This procedure should always be performed when the x ray machine is turned on c When the warming up process is completed the warm up status on the FXE control application changes to Ready FXE Control Data modules Service login Additional manipulator functions Close X Ray main 34 Ray On S General Extended Centering Focussing Conditioning ModeDM TargetDM Statistics TubeDM HSGDM SysDM AmpDM VacDM 4 gt X Ray timer 34 Voltage kV Time s set 10 left 110 Max 100 Ia _ ee Use timer R SSS SS gern SSS SSS SSS Max warmup voltage 225 r Actual X Ray output 34 Current uA poser L 11 i Max T oo NENNEN 3000 Tube current pA 3 LECHE EE E EE EIE GIG Condensor ap cur uA i 1 1 i Dbjectve ap cu DA 0 0 m 50 4 4 i r r 4 T r T O 750 800 950 900 Target mode selection Target curent pA Generator ready T liena contiol stena z Target 1 E arget power w Vau 138 CM iv T Isowalt fi EE olid targe p System states Hardware link status Last PLC error Tubehead Solid Warmup Ready Reset controller Startup Not executed No safe
10. from the scanning chamber f Scanning is completed 6 Detector Calibration The two principal signal calibrations are offset and gain which determine the detector readings with X rays off and with X rays on at scanning conditions respectively Cite Ketcham Click on Detector Calibration configure offset map detective pixels and change No of frames acquired to 200 The number of frames refers to the number of slices required from the scanned sample To get rid of defective pixels on the detector we need to take the following steps to make necessary corrections a Start the X ray by pushing the black button on the 450kV console b Place the cursor on a spot in the middle of the screen and adjust the energy intensity using the voltage slider in the FXE control application till you get a raw reading of about 60 000 grey levels v LED Light W Radios 00760 tif 759 1080 1536x1920 Display zoom 568 d Click on Gain 0 and change the of frames to 200 and click on acquire Wait for the acquiring process to complete e Then place the cursor again on a spot in the middle of the screen and adjust the energy intensity like previous till you get a raw reading of about 45 000 grey levels f Click on Gain 1 and change the of frames to 200 and click on acquire Wait for the acquiring process to complete g Then place the cursor again on a spot in the middle of the screen and adjust the volt
11. is defined by the gradation of aggregates the orientation and number of contacts of ag gregate particles the properties of aggregate binder interface and the voids structure Wang 2010 Understanding the complex mechanical interaction that exists between the constituents of the asphalt concrete requires a reliable way to characterize the asphalt concrete microstruc ture The present study is aimed at developing the workflow from image acquisition to simulation for accurate characterization of the AC mi crostruture Computed tomography and modern computational sim ulation techniques are tools that can be used to significantly improve asphalt mixture performance directly relate the mechanical properties of the virtual samples to performance on the field and hence overcome the discrepancy that exist between asphalt mixture laboratory and field performance and reduce the cost associated with laboratory testing In this context mastic is referred to as the mixture of binder and fines At low temperatures the mastic behaves like an elastic mate rial while at high temperatures it behaves as a viscoelastic viscoplastic medium The complex temperature and loading rate dependent me chanical behavior of the mastic and the variability in the property of the binder as a result of varying chemical composition and source poses a challenge for modeling the asphalt mixture using realistic chemical or mechanical properties Instead linear viscoelasti
12. otherwise The feret diameter is used to determine the length and width of the stones Figure 35 shows the definition of the feret diameter of a stone The length is the maximum of the feret diameter and the width is the minimum of the feret diameter in x y and z coordinates Al Figure 35 Description of maximum Feret diameter In order simplify the AC microstructure aggregate particles with length less than 2 34mm are considered as part of the mastic The to tal number of aggregates in the sample after the clean up is 403 stones The distribution of the length of the stones is shown in Figure 36 and the distribution of the width in Figure 37 The 3D volume distribution of the stones is shown in Figure 38 Stone length distribution 70 60 4 50 40 30 hill mmm T DP ue Ru sp SZ E 2 L APS ly P XP AO qv 0 Number of stontes o o Length mm Figure 36 Distribution of length of stones in the sample 42 Stone width distribution 60 50 5 40 A B 30 E E 20 E if E 0 I A M wo e m eo t F Lu t a t co a w co N U e N ot do o m O G Wm O r d N 0 tT d e v eo a N N m L w w wo ceo o e eo e A N m m wT r 4 4 4 4 A a width mm Figure 37 Distribution of width of stones in the sample Volume of
13. stones 250 225 200 uw D 150 L 125 E g 100 C z 75 50 25 0 co m cec eo eo eo eoe ooo coc cocoouocococococoocooc u co o0 00509 c0 00005 005770057 0000 0 o o e N m cp X G 0 OO c N m cp WW G 0 OO ANM t Ad HAH de d HP HAN NNN Volume mm 3 Figure 38 Volume distribution of stones Table 1 shows the result of the volumetric analysis of the AC sam ple The result shows that the total volume of aggregates in the ana lyzed sample is 91109m the total volume of mastic is 36373mm and the total volume of air voids is 16468mm The percentage vol ume of the aggregates mastic and air voids in the analyzed sample are 63 396 25 396 and 11 496 respectively Table 2 shows the statistical 43 relationship for the area volume length and width of the aggregates Statistical relationship can be used to correlate the AC mix behaviour with geometric properties of the stones The relationship also gives relevant information that can be used to study asphalt mix design re liability and the microstructure variability Table 1 Phase Volumetric relationship Volume mm volume Aggregates 91109 63 3 Mastic 36373 25 0 Air voids 16468 11 4 Total volume of sample 143951 100 Table 2 Statistical analysis for aggregates Area3d mm Volume3d mm Length3d mm Width3d mm Min 1 74 0 08 2 34 0 30 Max 1221 86 2355 04 24 60 14 99 Mean 215 41 226 08 9 72 5 05 StdDev 235 10 354 75 5 37 3 26 Sum 86809 01 91
14. thresholding operation before beam hardening COTTECHION ok anh SOK anulo et ola a So Te ble eta Ae 23 Result of thresholding operation after beam hardening COrrection Dota tae RUS Aedes ilie iA ib ie Ba pe aos AR Bs ae tae 23 Contrast enhanced and gaussian filtered image of As phalt conerete si de wh a xo Ye e Roo 25 Contrast enhanced and median filtered image of Asphalt Ceu HD INL 26 Anisotropic diffused image of Asphalt concrete 27 Variation of gray level after image correction and noise FITEN E 6 atie ai cec aah a ae Seip ah a ea e Ochs caesa 28 Threshold operation 1 5 ooa ya Be E es 30 Slice showing probe line region of interest 32 Linear plot of gray intensity variation along probe line 32 ill 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 Al 42 43 AA 45 46 AT 48 First order derivative of the gray level intensities along the x coordinate deu atur eme ee es First order derivative of the gray level intensities along the y coordinate vor ALORS Y ARA deb eas Variation of gradient magnitude along probe line Gradient of the magnitude result showing edge of ob JOC he in TBG TD HOT o ure aor Ot os cir ue IER uM dom ue a Distance map of stones ll Line probe profile of a distance map b inverted dis tance Map A eon te ou qe pn e endi ups ei Hook ER i The fastwatershed principle Avizo 2009 Markers used for watershed segmen
15. tube detector and manipulator 225kv X ray energy generator This is the end of the Scan Procedure for X Ray Scan 450kV 1 Turn the key on the main console to turn on 2 Turn on the 450kV X ray scanning console KTH NSI X 5000 CT 450kV console 3 Warming up the x ray generator Pre warm up setting d Turn the key on the main console to 450kV tO change the scanning voltage on to 450kV e The 450kV console prompts for the downtime in days This is the number of days since the 450kV X ray was last used Enter the value for the downtime f Press Ctrl 1 then enter 101 to change to short warm up mode Mod 101 Short warm up takes about 15mins while the option 102 is for the long warm up mode Mod 102 which can take up to 2days To begin the warm up d Turn on the key on the 450kV console to power the x ray tube e Press the black button on the 450kV console to initiate the X ray emission Note the red button is to turn off emission Note Always place the beam blocker in front of the x ray source to block the rays from getting to the detector during warm up if this is not done then the x rays can damage the detector f The pre warming process begins and continues until the voltage increases to 450kV g When the voltage reaches 450kV then the warm up process is completed 4 Scanning configuration e Run the CT Rotate 450 program A shortcut can be found on the desktop f Alignment o
16. 109 19 3917 45 2034 24 Table 3 shows the estimation of the absolute error between the ac tual volume of the analyzed sample and the predicted measured vol ume using digital image processing techniques The AC image based reconstruction technique used in this study shows a high level of reli ability with an absolute error of 0 034 Table 3 Comparison of actual and measured AC sample volume Measure value Actual Volume of analyzed sample mm 144000 Predicted volume using image processing mm 143951 Absolute error 96 0 034 44 Distribution of the air voids with depth is obtained by calculating the 2D area of air voids per slice The 2D area of air voids is then cal culated for each slice along the depth of the sample Figure 39 shows the variation of the air voids with sample depth A practical applica tion of the air voids distribution is the assessment of compaction in AC sample The variation of the air voids can be used as a quality control measure to check the uniformity and adequacy of compaction effort on AC pavements It can also be used as a tool to assess the condition of existing AC pavements It can also be used to examine crack initiation and propagation in AC mixtures The distribution of the air voids can be captured in the x y and z coordinates before the application of mechanical load on the AC sample After loading the variation of the air voids distribution can be used together with other related info
17. 3 Image smoothening and noise reduction 4 Accurate phase segmentation and separation of aggregates using distance based watershed segmentation 5 Numerical simulation on 2D and 3D image based models of AC Further work can be used to compare experimental observations with results from numerical simulation to validate models for further understanding of the mechanical behaviour of the AC microstructure Once a model is developed and validated for a typical AC microstruc ture the microstructure composition can be varied to investigate the influence of composition variation on the mechanical behaviour of the AC sample Porosity and absolute permeability through the AC mi crostructure can also be used to determine the drainage characteristics and used to provide the appropriate drainage system to reduce mois ture damage in AC pavements X ray computed tomography CT and digital image processing DIP can also be used to assess the quality of finished asphalt pave ments by taking core samples for quantitative and qualitative analysis The distribution of the air voids in the internal structure can be related 65 to the quality of the compaction effort and asphalt concrete mixtures can be developed for specific purposes with high level of reliability by establishing the relationship between the model parameters and ma terial properties in order to optimize the mix properties and achieve desirable mix performance Masad et al 2005 Th
18. Em MPa ES MPa 1 1 12 84 1 2 0 1 11 81 3 0 01 1 829 The geometry used for the thermal analysis is the same as the one for the uniaxial tension with the aggregates mastic and air voids well represented The boundaries of the model is insulated and the initial temperature is 273 15K i e 0 The air voids boundaries are subjected to a temperature variation for 1 hour with a cooling rate of 10 3600 K s Figure 53 shows the temperature of the AC microstructure af ter cooling for 1 hour A temperature variation of approximately 10 K can be seen The temperature gradient would increase with increasing cooling speed and decreasing air voids content As the specimen is not constrained this type of thermal loading would result in only negligi ble stresses for the material 58 Time 3600 Surface Temperature K Ys A 264 54 95 90 85 264 4 80 L 75 264 2 70 65 264 60 inl 263 8 50 45 30 263 6 35 J 30 263 4 25 20 263 2 15 20 30 40 50 60 70 80 90 100 Y 263 13 Figure 53 Surface temperature of AC microstructure after 1hour cooling Figure 54 shows the regions of localized tension in the mastic due to difference in thermal contraction properties between phases Zones of intense tensile strains gt 0 5 strain are located between stones and around air voids From Figure 55 it can be seen that the stones are subjected to higher stresses due to their higher stiffness However maximum tensile stres
19. Image contrast enhancement Image contrast enhancement is a technique used to improve the con trast and of an image This enhancement process is used to transform the pixel value from the original image to the pixel value in the en hanced image The image contrast enhancement can be used to make visible more image detail or information and help reduce noise in some 18 cases However the image contrast enhancement can blur out some information and this is a usually the case with contrast enhancement The image contrast enhancement technique used depends on the type of results desired S T p A T p m 0 255 p Figure 9 Image contrast equalization technique Poor contrast image can be as a result of combination of effects which includes beam energy tube intensity scene illumination etc From Figure 7 it can be observed that the acquired image from the CT scan have a very poor contrast and hence it is difficult to visualize the image The image contrast equalization function shown in Figure 9 is used to stretch the contrast of the input image p by applying a gray scale transformation T so as to produce a contrast enhanced image 5 This image equalization technique is applied to the raw image to achieve an image with a good contrast for further processing Figure 10 shows the image contrast enhanced image Figure 7 shows the profile for the probe line in Figure 10 It can be seen from the probe line profile that the regions
20. Internal Structure Characterization of Asphalt Concrete using X Ray Computed Tomography E IBRAHIM ONIFADE KTH VETENSKAP 39 OCH KONST 9 Soy woe KTH Architecture and the Built Environment Master of Science Thesis Stockholm Sweden 2013 dip cit E TH Z VETENSKAP JO OCH KONST A cele KTH Architecture and the Built Environment Internal Structure Characterization of Asphalt Concrete using X ray Computed Tomography Ibrahim Onifade February 2013 TSC MT 13 002 Royal Institute of Technology KTH Department of Civil and Architectural Engineering Division of Highway and Railway Engineering Stockholm 2013 Internal Structure Characterization of Asphalt Concrete using X ray Computed Tomography Ibrahim Onifade Graduate Student Infrastructure Engineering Division of Highway and Railway Engineering Department of Civil and Architectural Engineering Royal Institute of Technology KTH SE 100 44 Stockholm onifade kth se Abstract This study is carried out to develop the workflow from image acqui sition to numerical simulation for asphalt concrete AC microstructure High resolution computed tomography CT images are acquired and the image quality is improved using digital image processing DIP Non uniform illumination which results in inaccurate phase segmentation is corrected by applying an illumination profile to correct the background and flat fields in the image Distance map
21. Table 5 53 a Figure 47 a Phase segmented asphalt concrete sample b 2d uniaxial tension model configuration Table 5 Thermal linear elastic model parameters Material property Aggregates Mastic Young s Modulus E 70e9 Pal 120e6 Pa Poissons ratio u 0 3 0 49 Density p 4000 kg m 1700 kg m Coefficient of thermal expansion a 7 9e 6 1 K 6e 4 1 K Thermal conductivity K 3 W m x k 0 5 W m x k Heat capacity at constant pressure Cp 790 J kg k 1000 J kg k Bulk Modulus K 58 3e9 Pa 10e9 Pa Shear Modulus G 26 9e9 Pal 10e6 Pa 54 Surface First principal invariant of stress N m is r A 1 6804x10 x10 85r 80 1 5 75 70 65r 60 55 50 45 40 L 35 30 257 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 v 3 412x10 Figure 48 Continuum model Stress distribution in AC microstructure Time 1 Surface First principal invariant of stress N m v T A 6 4477 x10 x10 90 il 85 80 0 9 75 0 8 we 0 7 65 0 6 60 55r 4 0 5 50 0 4 45 4 0 3 40 0 2 35 30 0 1 25r it 1 i L n 1 L J 0 30 40 50 60 70 80 90 v 9 426x10 Figure 49 Multi phase model Stress distribution in AC microstructure The continuum model shows a uniform load transfer pattern in the mastic medium while load transfer chains can be observed in the multi phase model as shown in Figures
22. able left riaht rm 169 655 Source up down mm 139 524 306 704 Table mag mm 1092 014 Detector mag mm 7 C ol able tilt degree DR project CT project CT project Part name Detector Calibration Advanced Parameters 360 range degree 0 image numbering start CT Scan 1440 projections delay ms ft average C Continuous scan l est duration 47m59s Images depth Max v 23 48 GB needed 908 68 GB available B CT Start E CT rab CT Rotate 225 application interface Monitor Xray down 10 variation max c Alignment of the x ray source and the detector The scanning should be carried out in a way that the full image of the object sample can be captured on the detector The x ray source detector and scanning platform can be aligned using the control gear The control gear is a variable speed Joystick used to control all drives Each axis control is independent allowing simultaneous movement of all axis Control gear with variable speed joystick d Once the object has been placed on the scanning platform and the necessary alignments have been made then the actual scanning can be carried out 5 Scanning a sample a Push the 225kV Enable button on the main console b Open the FXE Control application window and then click X ray On to start
23. age like previous till you get a raw reading of about 30 000 grey levels h Click on Gain 2 and change the of frames to 200 and click on acquire Wait for the acquiring process to complete i Then place the cursor again on a spot in the middle of the screen and adjust the voltage like previous till you get a raw reading of about 15 000 grey levels j Click on Gain 3 and change the of frames to 200 and click on acquire Wait for the acquiring process to complete k Then turn off the X ray 7 CT Calibration Notes taken from efX CT user manual An important step in the CT process is capturing images of the calibration tool These images are used to define the geometric setup of the inspection system and the relationship between pixels and units of measuring It is imperative that no movements be made with the exception of the rotational axis If the calibrations tool needs elevation it is recommended that it is place on a stable object Capturing of these images can be done before or after capturing of the inspection radiographs Additionally the energies and filters used during acquisition may be altered for the calibration images The idea is to maximize the contrast between the calibration spheres and the background Positioning the Calibration Tool Positioning the calibration tool will require a degree of trial and error The ideal situation will be to have the calibration tool travel from edge to edge of the image
24. al image processing DIP algorithm called Volumetrics based Global Minima VGM threshold ing algorithm for processing asphalt concrete AC X ray computed tomography CT images The thresholding algorithm utilizes known volumetric properties of AC mixtures as the main criterion for es tablishing the air mastic and mastic aggregate gray scale boundary thresholds Bhasin et al 2011 used X ray CT images to study the 3 dimensional distribution of the mastic in asphalt composites You et al 2012 developed a three dimensional 3D microstructure 3 based computational model to predict the thermo mechanical response of the asphalt concrete using a coupled thermo viscoelastic thermo viscoplastic and thermo viscodamage constitutive model You et al 2008 studied the dynamic modulus from the stress strain response un der compressive loads for two dimensional 2D and three dimensional 3D microstructure based discrete element models of asphalt mixtures Masad et al 2005 developed an approach for constitutive model ing of the viscoplastic behavior of asphalt mixes and measured the microstructure damage with X ray computed tomography and image analysis techniques However very limited amount of work has been done so far for 3D image based modeling of AC while some studies only considered the distribution of the aggregates and the mastic phase You et al 2012 The air voids mastic and aggregates phase are considered in this study Th
25. and analysis of composite struc tures Materials Science and Engineering 68 2 239 248 You et al 2012 You T A Rub R K A Darabi M K Masad E A and Little D N 2012 Three dimensional microstruc tural modeling of asphalt concrete using a unified viscoelas tic viscoplastic viscodamage model Construction and Building Materials 28 1 531 548 68 You et al 2008 You Z Adhikari S and Dai Q 2008 Three dimensional discrete element models for asphalt mixtures Journal of Engineering Mechanics 134 12 1053 1063 Yue et al 2003 Yue Z Q Chen S and Tham L G 2003 Fi nite element modeling of geomaterials using digital image process ing Computers and Geotechnics 30 5 375 397 Zelelew and Papagiannakis 2011 Zelelew H M and Papagian nakis A T 2011 A volumetrics thresholding algorithm for pro cessing asphalt concrete x ray ct images International Journal of Pavement Engineering 12 6 543 551 69 Appendix A X Ray scanning procedure 70 Procedure for X Ray Scan 225kV 1 Turn the key on the main console to turn on 2 Turn on X ray tube generator located behind the main scanning chamber 3 Warming up the x ray generator Pre warm up setting a Run PXE control application b On the PXE control application wait for the vacuum level to reach 110 then it gives a green signal FXE Control Data modules Service login Additional manipulator functi
26. ations The point operations are used to apply image processing techniques on discrete points in the image while the mask operation is used to apply image processing techniques on re gions Image filtering operations are typical examples of the mask operation while the histogram equalization is an example of a point operation Figure 6 shows an example of a 3x3 mask Connectivity and neighbourhood relationship between pixels is es sential in digital image processing Typically there are three types of pixel neighbourhood relationships the horizontal vertical and diago nal neighbours A pixel have two horizontal two vertical neighbours and four diagonal neighbours except at the boundaries where there are less neighbours Figure 6 shows the 8 neighbours of a pixel P where N Np and Ng are the vertical horizontal and diagonal neigh bours respectively Pixels are connected if they are adjacent in some sense and if they are neighbours and have similar intensity values There are different types of connectivity which includes 4 connectivity 8 connectivity and m connectivity 4 connectivity is the connectivity between the horizontal and vertical neighours 8 connectivity is the connection between horizontal vertical and diagonal neighours while the m connectivity is similar to the 8 connectivity except that it elimi nates the problem of multi path connection Pixel connectivity makes it possible to determine object properties like object boun
27. ave the most satisfactory result This method accurately separates the aggregates in the 3D image deter mines the contact point between aggregates and also create a region for mastic phase between the interface of two adjacent aggregates In this study the air void phase is segmented using basic threshold ing operation since the gray intensity levels for the air voids are quite distinct from those of the mastic and aggregates The aggregate phase is segmented and separated using the distance map and watershed seg mentation and the mastic phase is obtained by subtracting the air void phase and the aggregate phase from the full mask of the scanned sam ple 3 5 1 Thresholding Threshold operations are used for separating objects with a gray level of interest from the rest of the image The threshold operation produces a binary image with the gray levels of interest set to 1 and all other pixels are set to 0 The thresholding operation is used to obtain the air void phase and to create the aggregate phase mask 29 before the distance map based watershed segmentation is applied for aggregate separation Figure 20 illustrates the thresholding operation function S T p T p 0 p 255 Figure 20 Threshold operation 3 5 2 Edge detection Accurate phase segmentation is dependent on the ability to accu rately determine the location and direction of edges of objects in an image Edge detection is the technique used
28. based watershed segmentation is used to accurately segment the phases and separate the aggregates Quantitative analysis of the microstructure is used to determine the phase volumetric relationship and aggregates characteristics The results of the phase reconstruction and internal structure quantification using this procedure shows a very high level of reliability Numerical simulations are carried out in Two dimensions 2D and Three dimen sions 3D on the processed AC microstructure Finite element analysis FEM is used to capture the strength and deformation mechanisms of the AC microstruc ture The micromechanical behaviour of the AC is investigated when it is con sidered as a continuum and when considered as a multi phase model The results show that the size and arrangement of aggregates determines the stress distribu tion pattern in the mix Key Words X ray computed tomography digital image processing finite element method image based modeling Acknowledgement I would like to express my sincere gratitude to Assoc Prof Nicole Kringos for her insights and giving me the opportunity to carryout this study in the Division of Highway and Railway Engineering Royal Institute of Technology Stockholm Sweden I would also like to appre ciate Asst Prof Denis Jelagin for his kind supervision and unending encouragement during the course of this study Dr Alvaro Guarin has provided enormous support and enlightenment and have also as
29. c and viscoplas tic models are used for modeling the mastic and the properties of the mastic are represented using model constants Examples of such mod els are the Maxwell and Kelvin s viscoelastic models or more general Prony series representation Wang 2010 Aggregates can be modeled as linear elastic material or more com plicated models such as continuum damage model viscoplasticity model elastoplasticity model with back stresses Wang 2010 In this study the aggregates are considered as a linear elastic material with a stress strain relationship as linearly elastic Model parameters required for modeling the aggregates behaviour are the mechanical properties of the aggregates which include the Young s Modulus the Poisson s ratio and the strength parameters Under loading condition there may exist a rearrangement of the internal structure of the asphalt concrete mixture depending on the magnitude and duration of loading The most important aspects to consider in the modeling is the contact between two aggregates and the interface between the aggregate and the surrounding mastic The morphological properties shape angularity and texture determines the load transfer between aggregate particles at contact and the bond ing at the interface between the mastic and aggregate particles T hese properties are captured using the X ray CT and quantified using the image processing capabilities of Avizo R 2 The microstructure of th
30. daries area perimeter object identification etc 16 Figure 6 Neighbourhood relationship The image processing operations and relationships discussed above are used in this study to process and analyze a 60 x 60 x 40mm subvol ume image of the of the asphalt concrete scanned sample Avizo B fire is used for the processing and analysis of the acquired asphalt concrete image The digital image processing techniques used in this study are discussed below 3 1 Grey level histogram The histogram is a distribution of the number of pixels with their corresponding gray level values The gray level values is plotted on the abscissa while the frequency of pixels is plotted on the ordinate For an 8 bit image the gray scale range is from 0 225 because it contains 256 gray intensity levels The higher the gray level value the brighter the pixel i e gray intensity value of zero 0 denotes dark pixels while 255 represents bright pixels The gray level intensity relationship can be used in determining the right threshold value for segmenting different phases in the digital image Figures 7 8 shows a subvolume region of the acquired asphalt concrete image and the corresponding gray level distribution respectively 17 Figure 7 Subvolume of acquired image from CT scan 100000 10000 1000 Number of pixels 100 0 40 80 120 160 200 240 280 Gray level intensity Figure 8 Histogram of acquired image from CT scan 3 2
31. detector to the top If the tool does not span from the top to the bottom place the tool so that one portion of it reaches the top or bottom This allows the creation of larger ellipses during rotation and better data for the software to build the volume off of J m 6 35 mm E a T 1 59 mm 15 mm 40 mm 5mm 1 mm Small Medium Large CT calibration procedure a Place the calibration rod on the scanning platform There are three sizes of calibration tools to choose from large medium and small Tool size used is determined by the magnification used during acquisition and detector size b OnCT Project window click on CT Calib and select the 15mm large tool if you are using the large tool for the calibration else choose what corresponds to your choice and then click OK Then the calibration process begins c NB Before the calibration process begins necessary re alignment of the scanning platform might be required to make sure that the full image of the calibrating tool is captured on the detector d Wait for the calibration process to finish Y X 3 3X ji cce w P K ax Take note of the Calibration tool images above These stills are 180 degrees apart from one another The last item to note is the number of calibration spheres visible The more of them present the more accurate the software will interpret the geometric positioning of the x ray
32. dient magnitude along probe line Figure 26 Gradient of the magnitude result showing edge of objects in the image 3 5 3 Distance Map approach The distance operation is used for the particle segmentation before the watershed operation It calculates the distance function of a binary image onto a gray level image as the sum of successive erosions In the output image O a pixel x belonging to a particle X takes an intensity I equal to the distance in pixel units to the boundary Avizo 2009 The distxxx is command is used for the distance operation because it 34 takes care of the diagonals and hence more accurate Figure 27 shows the distance map of the aggregate mask The center of the aggregates show a high gray intensity value and the gray intensity value decreases as it moves towards the edge Figure 28 a 28 b shows the line probe profile for the distance map and the in verted distance map The inverted distance map is used for the water shed segmentation Figure 27 Distance map of stones 35 T 20 Probeline Length mm a 160 Value T T T Probeline Length mm b Figure 28 Line probe profile of a distance map b inverted distance map 3 5 4 Watershed segmentation The watershed segmentation is used for aggregate separation using the distance map and marker The marker is usually the regional maxima of the distance map image The watershed lines used for
33. e elements pixels Most part of the following X ray CT overview is adapted from Ketcham and Carlson 2001 X ray CT systems can be categorized into Medical CT system and Industrial CT system Medical CT systems are used to scan living ob jects while Industrial CT systems are mainly used to scan non living objects Medical CT systems are used in medicine for acquiring image of soft tissues and bones for diagnosis and prognosis Further advance ment and development of medical CT systems led to the evolvement of the Industrial CT systems High resolution industrial X ray CT differs from conventional medical CT scanning in its ability to utilize higher energy X rays with more penetrative power and to resolve details as small as a few tens of microns in size even when imaging objects made of high density materials Ketcham and Carlson 2001 The Industrial CT systems are used in different field and disciplines ranging from geo sciences engineering manufacturing aviation just to mention a few The quality of the scanned image is a measure of the noise the slice thickness the low contrast resolution and the high contrast resolution These qualities depend on the X ray source X ray detectors used and the scanning configuration The variables that determine how effective an X ray source will be for a particular task are size of the focal spot the spectrum of X ray energies generated and the X ray intensity The focal spot size partially defin
34. e AC has been simplified or over idealized by some researchers Bazant et al 1990 and Schlangen and van Mier 1992 have represented the aggregates as rigid spherical particles while others Wittmann et al 1985 and Wang et al 1999 have used algorithm to generate a random microstructure image of the as phalt concrete microstructure Most of these past research have been based on the two dimensional 2D analysis of the asphalt concrete mi crostucture due to the complexities in generating or accurately repre senting the three dimensional 3D microstructure of asphalt concrete The over idealized or randomly generated 2D microstructure does not effectively predict the mechanical behaviour of the asphalt concrete microstructure There has been a number of recent attempts to use X Ray CT to investigate the internal structure of AC and to investigate its impact on the AC mechanical properties Coleri et al 2012b used the X ray CT to study the changes in AC microstructure using full scale test sections and Heavy Vehicle Simulator HVS loading and X ray CT images taken before and after HVS testing Coleri et al 2012a also used the X ray computed tomography CT and digital image process ing to generate the internal microstructure of the asphalt mixtures and study the effectiveness of 2D and 3D models for the simulation of the shear frequency sweep at constant height FSCH test Zelelew and Papagiannakis 2011 used automated digit
35. e diffusion independent of the image gradient and smoothen out the edges a large value prevents smoothing in all edge like regions In order to make the diffusion process more stable the image is pre filtered by a Gaussian filter with parameter sigma All features of size sigma are removed This allows noise to be removed from the image But a too large value may also remove relevant features Avizo 2009 In order to investigate the appropriate parameters required to ef fectively denoise the scanned image without loosing image details a parametric study is carried out to determine the optimum parameter values A time stop of 25s with a time step of 5s is used while the values of the contrast and the sigma are 20 and 5 respectively Figure 18 shows the result of the edge preserving filter and Figure 19 shows the line probe profile for the line in Figure 18 Figure 18 Anisotropic diffused image of Asphalt concrete 27 Gray intensity value Probeline Length mm Figure 19 Variation of gray level after image correction and noise filtering 3 4 4 Erosion amp Dilation Erosion and dilation are neighbourhood operations similar to other filtering operations Erosion examines the pixel values of neighbour ing pixels and replace the value of the pixel with the minimum of the neighbouring pixel values On the other hand dilation is used to per form the same operation except that the pixel value is replaced with the maximum val
36. e interface between the mastic phase and the aggregate phase is challenging to model when considering the mechanics of the mastic phase which is highly anisotropic Further study is required to ade quately understand the interaction at the boundary between the mastic and the aggregates However the spatial location of the contact points between adjacent aggregates is determined in this study and referred to as the contact geometry 1 1 Objectives amp workflow The main objectives of this study are to develop procedures for 1 Segmentation of the three different phases in AC and determina tion of their volumetric relationship 2 Determination of air voids phase distribution with depth 3 Determination of aggregates particle size gradation and distribu tion 4 Determination of the distribution of contact zones between aggre gates 5 Micromechanical simulation using finite elements method FEM The steps involved in this study are summarized in Figure 1 Figure 1 Process workflow 2 Xray Computed Tomography X ray computed tomography is a non destructive technique used in visualizing the interior of an opaque solid object with the aim of obtaining digital information of the internal structure of the objects at a microscopic detail level A CT image is referred to as a slice Each slice has its associated thickness and hence makes a volume CT images are represented with volume elements voxels and not pictur
37. e use of Computed Tomography for material characterization presents a lot of possibilities in the future of asphalt concrete mix design 66 References Avizo 2009 Avizo 2009 Avizo User s Guide Vordeaux Frace Visualization Sciences Group Bazant et al 1990 Bazant Z P Tabbara M R Kazemi M T and Pijaudier Cabot G I 1990 Random particle model for fracture of aggregate or fiber composites Journal of Engineering Mechanics 116 1686 705 Berger et al 2010 Berger M J Hubbell J H Seltzer S M Chang J Coursey J S Sukumar R Zucker D S and Olsen K 2010 XCOM Photon Cross Section Database version 1 5 Bhasin et al 2011 Bhasin A Izadi A and Bedgaker S 2011 Three dimensional distribution of the mastic in asphalt composites Construction and Building Materials 25 10 4079 4087 Coleri et aL 2012a Coleri E Harvey J T Yang K and Boone J M 2012a Development of a micromechanical finite element model from computed tomography images for shear modulus sim ulation of asphalt mixtures Construction and Building Materials 30 0 783 793 Coleri et aL 2012b Coleri E Harvey J T Yang K and Boone J M 2012b A micromechanical approach to investigate asphalt concrete rutting mechanisms Construction and Building Materi als 30 0 36 49 Johns et al 1993 Johns R A Steude J S Castanier L M and Roberts P V 1993
38. ented image with the aggregates in red the mastic in blue and the air voids in aqua 38 Figure 32 AC sample showing the aggregates mastic and air voids segmented Surface reconstruction for the aggregates is shown in Figure 33 Fig ure 34 shows the air voids phase reconstruction with air voids variation from top to bottom of the scanned sample Figure 33 Reconstructed 3D image of aggregate 39 Figure 34 Reconstructed 3D image of air voids 40 4 Digital Image Analysis and Results After the phase segmentation of the AC sample image analysis is carried out so as to extract pertinent quantitative information In this study the interactive measure features of Avizo fire is used for the im age analysis The results of interest from the individual analysis of the ageregates are the Volume in 3D Area in 3D Feret diameter length and width in 3D and orientation of the aggregates For accurate ag gregates size analysis the bounding box dimensions can be used for aggregate characterization The volume in 3D of an object X is defined by the relationship in Equation 10 for a continuous case but estimated using Equation 11 for a discrete case The volume is the number of pixels in region X multiplied by the volume of a voxel VCX num dxdydz 10 Vo pes 3 xisyj 2n 11 i j k where I x i Yj z the intensity of the pixel of coordinates P M ZL I xi Yy 2n 1 if the pixel lies within the object X and 0
39. ermine the exact location of the boundary between two different objects but it is very sensitive to noise in the image The second order derivative is not used for edge detection but can be used to extract other pertinent image information 3l Gray level intensity f 15 Probeline Length mm Figure 22 Linear plot of gray intensity variation along probe line Figure 21 shows the image of a slice from the stack of slices ac quired The line probe tool feature in Avizo is used to measure the gray level intensities of a region of interest on the slice The region spans across two aggregates a region with binder and an air void region The linear plot of the gray level intensities along the probe line is shown in Figure 22 Figure 23 24 shows the variation of the first order derivative with respect to x and y coordinates The variation of the magnitude of the gradient along the probe line is shown in Figure 25 32 Figure 26 shows the result of edge detection using the magnitude of the gradient df dx 22 5 75 15 Probeline Length mm First order derivative of the gray level intensities along the x coordinate Figure 23 df dy 22 5 20 75 15 Probeline Length mm First order derivative of the gray level intensities along the y coordinate Figure 24 33 Gradient Magnitude 15 22 5 30 Probeline Length mm Figure 25 Variation of gra
40. es the potential spatial resolution of a CT system by determining the number of possible source detector paths that can intersect a given point in the object being scanned The more such source detector paths there are the more blurring of features there will be The energy spectrum defines the penetrative ability of the X rays as well as their expected relative attenuation as they pass through materials of different density Higher energy X rays penetrate more effectively than lower energy ones but are less sensitive to changes in material density and composition The X ray intensity directly affects the signal to noise ratio SNR and thus image clarity Higher intensities improve the underlying counting statistics but often require a larger focal spot The scanning configuration refers to the source object detector distances Ketcham and Carlson 2001 The resolution of the scanned object is magnified by minimizing the source object distance and maximizing the object dectector distance Figure 2 NSI X 5000 X ray CT facility in KTH Highway and Railway lab 2 1 Mass Total attenuation coefficient X ray Attenuation CT images of an asphalt sample are the mapping of the attenua tion coefficient of its constituents Mass attenuation coefficient is a measurement of how strongly a chemical species or substance absorbs or scatters light at a given wavelength per unit mass In addition to visible light mass attenuation coefficients can be de
41. f the scanning platform A red value on any of the adjustment parameters in the CT Rotate 450 program indicates a misalignment of the scanning platform in the x ray chamber Necessary amendments should be made so that all adjustment parameters are in the allowable range For instance the table tilt parameter relates to how the table is tilted in the horizontal The table tilt value should be set to zero 0 to ensure that the table is perfectly horizontal and that the sample is not tilted during the scan CT project Part name E Detector Calibration Snap Record CT Scan DES 1440 projections 0 delay ms E fr average C Continuous scan l est duration 47m59s 23 48 GB needed 308 68 GB available B CT Stan B CT Calib Advanced Parameters 360 range degree D image numbering start Images depth Max v Monitor Xray down 10 variation max Close CT Rotate 450 application interface g Alignment of the x ray source and the detector The scanning should be carried out in a way that the full image of the object sample can be captured on the detector The x ray source detector and scanning platform can be aligned using the control gear The control gear is a variable speed Joystick used to control all drives Each axis control is independent allowing simultaneous movement of all axis h Once the object has been placed on the scanning platform and the necessary ali
42. fined for other electromagnetic radiation such as X rays sound or any other beam that attenuates Wikipedia The defining equation for the mass attenuation coefficient is es sentially a different way to write the Beer Lambert law The Beer Lambert law is normally written Wikipedia I Inet 1 where Io is the original intensity of the beam I is the intensity of the beam at distance into the substance e is Euler s number about 2 718 u is the attenuation coefficient When discussing the mass attenuation coefficient this equation is rewritten as where p is the density 5 is the mass attenuation coefficient plis the area density known also as mass thickness The ability of the X rays to differentiate materials depends on their respective linear attenuation coefficients Ketcham and Carlson 2001 Asphalt concrete constituent materials with their linear attenuation co efficient can be used to determine the energy level that is most appro priate for scanning The linear attenuation coefficients can be obtained by multiplying the mass attenuation coefficient of each constituent ma terial with their respective density The values for the mass attenuation for different elements compounds and mixtures can be obtained from XCOM Photon Cross Sections Database managed by National Insti tute of Standards and Technology NIST Berger et al 2010 Figure 3 shows the plot of the linear attenuation coefficient as a functi
43. gnments have been made then the actual scanning can be carried out 5 Scanning a sample d e f Run the CT Rotate 450 program A shortcut can be found on the desktop Push the 450kV Enable button on the main console to turn X ray ON Check if the full image is formed on detector and if not you have to make necessary alignment changes These changes can be made using the control gear The process of re alignment of the detector X ray source and the scanning platform can be monitored using the CCTV monitor outside the scanning chamber There are four cameras installed in the scanning chamber Changing the Scan settings The voltage and the current to perform the scan can be changed on the 450kV console while the setting of the amount of details to capture can be changed on the CT rotate 450 application a Settings on 450kV console The voltage and the current for the 450kV scan can be adjusted to obtain a good scanned image On the 450kV console press the kV or mA button and turn the knob to adjust the voltage and the current respectively until a satisfactory image is obtained The voltage corresponds to the energy intensity of the X ray The adjustment should be made until one gets a good scan image on the computer screen A too dark image implies a bad scan and a too light image indicates that some details might have been lost during the scan This process is very important in the scan process and requires
44. hips and individual aggregate properties like volume of aggregates length and width of aggregates orientation of aggregates and spatial location of aggregates in the AC mix Air voids distribution in the AC mix can be captured in three di mension 3D and used to study the AC microstructure evolution Contacts between adjacent aggregates can also be captured using the watershed lines used for the aggregate separation Numerical simulation is used to study the strength and deforma tion mechanisms in order to characterize the AC microstructure us ing two dimensional 2D and three dimensional 3D finite element analysis The AC microstructure is considered as a continuum and then as a multi phase model in order to investigate the mechanical behaviour of the microstructure Accurate material characterization is obtained when the AC microstructure is considered as a multi phase model with different material properties for the mastic and aggregates Load transfer chains can be observed in the multi phase model with 64 strains localization in the mastic The maximum tensile strains in the AC microstructure for the multi phase model is about 12 compared to the continuum model where approximately 0 2 strain is predicted In conclusion the image to simulation workflow for AC microstruc ture have been developed in this study Procedures have been devel oped for 1 Image acquisition using CT 2 Image non uniform illumination correction
45. ibution in 3D model List of Tables NID oO FR WN rn Phase Volumetric relationship 44 Statistical analysis for aggregates 44 Comparison of actual and measured AC sample volume 44 Orientation measures of particle 49 Thermal linear elastic model parameters 54 Maximum tensile stress and strain 57 Viscoelastic material parameters 58 vi 1 Introduction Asphalt concrete AC is a heterogeneous material which consists of mastic binder and fines aggregates and air voids The distribution of the air voids in the matrix the interaction between the aggregates and the mastic and the properties of the aggregates and the mastic plays a vital role in determining the mechanical behavior of the asphalt con crete Mainly the aggregate properties determine the strength charac teristics the mastic determines the durability characteristics and the air void is related to the rate of moisture damage and rutting in the asphalt concrete Asphalt concrete mixes act as continuum materials at low temperatures when the stiffness of the asphalt binder and the aggregate are close However at high temperatures i e the critical conditions for rutting deformation accumulation stiffness of the aggre gate can be two to three orders of magnitude higher than the asphalt binder stiffness Coleri et al 2012b The microstructure of AC is very complicated It
46. line CD denotes the major axis of the particle and Line EF denotes the minor axis of the particle The centroid of the particle O is calculated using the moments of inertia 48 Figure 43 Description of orientation measure The particle orientation measures can be used together with the spatial location of the particles and particle identification to track par ticles movement in microstructure of the AC mix after loading Ad vanced mechanical relationships can be used to relate the translation and rotation of the particles to stresses and strains in different parts of the AC mix Further understanding of such motion and interaction at the interface between the mastic and stone interface will provide an in depth understanding of the mechanical behaviour of the AC mi crostructure Table 4 Orientation measures of particle Lengthorientation orientation orientation2 0 9 0 0 0 45 54 90 43 0 9 For simplicity only the results of the length orientation and orien tation of the major axis is presented in this report Length orientation phi varies from 90 to 4 90 and the Orientation 9 varies from 49 90 to 90 while the Length orientation theta ranges from 0 to 180 and the the Orientation theta ranges from 180 to 180 Figure 44 shows the length orientation of the aggregates in the asphalt concrete mix It can be seen from Figure 44 that the orientation of the maximum Feret diameter
47. ly reflected on the image on the computer screen Settings on the CT rotate 225 Change the number of frames per second fps to a suitable value This is directly related to how much detail you like to capture during the scan A fps of 2 is usually ideal for small objects but a fps greater than 2 will be suggested for large objects like asphalt cores There is a tradeoff between scanning speed and scan quality when considering the number of frames per second When you change the value of the fps check the image on the computer screen if it is satisfactory before proceeding If the image is satisfactory then turn off the X Ray emission by pushing the enable x ray to turn off the x ray Scanning a After necessary change in the scanning configuration and the scan settings the actual scanning process follows b To start a new scan click on CT Project On the CT rotate 225 application window and the CT Project window opens c Onthe CT Project window select a location to save the new scan make a new folder and give a name to the new folder Then click OK d Click on CT Scan to start the scanning process Then wait till the scan is finished e Turn off the X ray and take out the object from the scanning chamber f Scanning is completed 6 Detector Calibration The two principal signal calibrations are offset and gain which determine the detector readings with X rays off and with X rays on at scanning condition
48. n filter The median filter replaces the pixel value in the image by the median value of the neighbouring pixels Median filter is more effective in reducing noise than the low pass filter and the mean filter especially when the image contains non gaussian noise The median filter does not blur the edges as much as the gaussian filter The 3x3 median filter kernel is used in denoising the acquired asphalt concrete image The 3x3 kernel is the most appropriate for asphalt concrete microstructure Figure 17 Contrast enhanced and median filtered image of Asphalt concrete 3 4 3 Edge preserving filter Anisotropic diffusion filters are used to smoothen region of objects and reduce or stop the smoothening in the vicinity of edges thereby preserving the edges of objects Unlike the lowpass gaussian filter that treats every pixel with the same convolution the non linear anisotropic diffusion filter treats every pixel based on the intensity of the neigh bourhood pixel while taking the edge of the objects into consideration In this study the edge preserving filter in Avizo is used to denoise the image so as to further improve the segmentation results The pa rameter required for this process are the time step which determines 26 how accurately this process is sampled the contrast parameter which determines how much the diffusion process depends on the image gra dient i e how much the smoothing is stopped near edges A value of 0 makes th
49. of most of the aggregates is about 4 in the X Y plane Length orientation Theta Number of stones 40 l d t T T T g m Oo m Ln D D KN uw nm ceo 109 emm 117 e 20 cum 126 134 142 ee ceo 16 S d D N m t wo 150 158 4 o aA angle degree Figure 44 Length orientation of stones theta Figure 45 show the distribution of the direction of the major axis of the aggregates in the sample The direction of the major axis of most of the aggregates in the sample is between 4 70 and 90 The distribution of the orientation of the major axis of the aggregates is shown in Figure 46 50 Orientation Phi i S E 2 angle degree Figure 45 Orientation of stones phi Orientation Theta Number of stones angle degree Figure 46 Orientation of stones theta 51 5 Finite Element Method FEM analysis and re sults In this study numerical simulation is carried out to investigate the behaviour of the AC microstructure under the action of mechanical and thermal loading Two dimensional 2D plane strain and three dimen sional 3D analysis are performed The 2D analysis is carried out on a slice acquired from the CT X ray scan and due to limited computing power three dimensional 3D FEM analysis is not performed on the whole microstructure but instead two stones connected with mastic is extracted from the AC microstruc
50. of the aggregates are not well defined and this leads to inaccurate estimation of the the ag gregate boundaries Using the contrast enhanced image i e Figure 10 for phase separation will give inaccurate segmentation results Other 19 image processing techniques used to further improve the quality of the image are discussed below Figure 10 Contrast enhanced image of Asphalt concrete 200 Gray intensity value 20 Probeline Length mm Figure 11 Variation of gray level before image correction and noise filtering 3 3 Beam hardening correction Beam hardening is a common phenomenon in X Ray CT scans Beam hardening is eminent where CT scans exhibits edges that is brighter than the center of the scanned sample as seen in Figure 10 20 The artifact derives its name from its underlying cause the increase in mean X ray energy or hardening of the X ray beam as it passes through the scanned object This happens as lower energy X rays are attenuated more readily than higher energy X rays a polychromatic beam passing through an object preferentially loses the lower energy parts of its spectrum The end result is a beam that though dimin ished in overall intensity has a higher average energy than the incident beam This also means that as the beam passes through an object the effective attenuation coefficient of any material diminishes Ketcham and Carlson 2001 Beam hardening complicates significant the
51. on of X ray energy for aggregate limestone bitumen and air It can be seen from Figure 3 that the linear attenuation coefficient for aggregate and bitumen tend to show a converging trend before energy intensity of 200kV and a parallel and constant trend at energy inten sities greater than 200kV It can be seen from Figure 3 that a good distinction between aggregate and the mastic phase can be achieved when the asphalt concrete sample is scanned below 200kV This is how ever true for relatively small sample size but higher energy intensities is required to penetrate through relatively large samples e g 80mm depth In the final scan image the material with the highest attenua tion coefficient i e the aggregates will have high values of gray level bright the material with the lowest attenuation coefficient i e air voids will have low values of gray level dark and the mastic gray level falls between these limits Linear Attenuation Coefficients for different materials Aggregate Bitumen Air Attenuation in 1 cm L poi O O 1 r 1 i r 1 154 1 Teese eet Bas Ed ES 10 10 10 10 10 Photon Energy in MeV Figure 3 Linear attenuation coefficient for asphalt concrete 2 2 Acquisition of CT data 2 2 1 Sample preparation The main sample preparation prior to a CT scanning is to ensure that the object fits inside the field of view and that it does not move during the scan Because the full scan field fo
52. ons Close X Ray main L B 8 General Extended Centering Focussing Conditioning ModeDM TargetDM Statistics TubeDM HSGDM SysDM AmpDM VacDM 4 RRs tne 3 SEEN Tmels set HO r 70 T NES n eect necro C T0 225 x pu rcx CEDE ETE n ENERETTEN Max warmup voltage 0 p Actual X Ray output Current pA Acc voltage kV Tube current pA Condensor ap cur uA i 1 n n 1 1 n 1 n 1 n 1 n n DO W 50 100 150 200 250 300 D 450 500 550 i 750 800 950 900 Target mode selection Tea aon Generator ready o T Intensity contol G Target power W Vacuum 646 RCM iv du T Isowatt p System states Objective ap cur pA Hardware link status Tubehead Solid Last PLC error Warmup Not executed Reset controller Startup Not executed No safety close door Functions T Warmup Startup Ready PXE control application interface before warming up c Place the beam blocker in front of the x ray source to block the rays from getting to the detector this is done to protect the detector from the unfiltered x rays generated during the warming up process To begin the warm up a Push the 225kV button on the main console to put on the X ray The machine beeps 4 times and then starts emitting x rays Main console of NSI X 5000 CT scanner b On the FXE control
53. perature variation One of the slices 60mm x 60mm acquired and processed using X ray tomography and digital image processing is used for the 2D FEM analysis The slice is phase segmented with the aggregate mastic and air voids well represented The air voids are included as cavities with no mechanical properties assigned 5 1 1 Two dimensional 2D uniaxial tension In the tensile load case the micromechanical behaviour of the AC microstructure can be captured by studying strength and deformation mechanism in the AC microstructure when it is modeled as a contin uum and compared with when it is considered as a multi phase model The mastic and the aggregates are modeled as linear elastic materials for simplicity The same material properties is assigned to the mas tic and aggregates for the continuum model while different elastic and thermal material properties are assigned to the mastic and aggregates in the multi phase model Figure 47 shows the phase segmented AC microstructure and the uniaxial 2D analysis configuration In both continuum and multi phase model the model is fully restrained at the bottom with a tensile dis placement of 0 1mm applied at the top of the model The material properties required to model the elastic behaviour of the aggregates are the Young s modulus Poissions ratio and the density of the aggre gates The thermal elastic material properties used in this analysis for both the mastic and stones are presented in
54. pot in the middle of the screen and adjust the voltage like previous till you get a raw reading of about 15 000 grey levels j Click on Gain 3 and change the of frames to 200 and click on acquire Wait for the acquiring process to complete k Then turn off the X ray 7 CT Calibration Notes taken from efX CT user manual An important step in the CT process is capturing images of the calibration tool These images are used to define the geometric setup of the inspection system and the relationship between pixels and units of measuring It is imperative that no movements be made with the exception of the rotational axis If the calibrations tool needs elevation it is recommended that it is place on a stable object Capturing of these images can be done before or after capturing of the inspection radiographs Additionally the energies and filters used during acquisition may be altered for the calibration images The idea is to maximize the contrast between the calibration spheres and the background Positioning the Calibration Tool Positioning the calibration tool will require a degree of trial and error The ideal situation will be to have the calibration tool travel from edge to edge of the image during a single 360 degree of rotation The balls of the calibration tool must be at least 1 ball diameter from the edge of the image It is also beneficial but not necessary to have the calibration tool reach from the bottom of the
55. r CT is a cylinder i e a stack of circular fields of view the most efficient geometry to scan is also a cylinder Thus when possible it is often advantageous to have an object take of a cylindrical geometry Ketcham and Carlson 2001 2 2 2 Calibration Calibrations are necessary to establish the characteristics of the X ray signal as read by the detectors under scanning conditions and to reduce geometrical uncertainties Ketcham and Carlson 2001 The two principal signal calibrations are offset and gain which determine the detector readings with X rays off and with X rays on at different energy levels respectively 10 2 2 3 Collection The principal variables in collection of CT data are the number of views and the signal acquisition time per view The number of views used ranges from 600 to 3600 or more Each view represents a ro tational interval equal to 3608 divided by the total number of views The raw data are displayed such that each line contains a single set of detector readings for a view and time progresses from top to bot tom This image is called a sinogram as any single point in the scanned object corresponds to a sinusoidal curve Ketcham and Carlson 2001 2 2 4 Reconstruction Reconstruction is the mathematical process of converting sinograms into two dimensional slice images The most widespread reconstruc tion technique is called filtered backprojection in which the data are first convolved with a fil
56. rain 1 Ys A 5 7083x105 0 0 005 0 015 0 025 0 035 0 04 Ed v 0 146 a Surface Third principal strain 1 princip A 5 7083x10 0 0 005 0 015 0 025 0 035 0 04 v 0 146 al b Figure 57 Compressive strain distribution in x direction Figures 58 shows the Von Mises stress distribution for the 3D model The Von Mises stress localized in stones with high stress intensities 62 around contact points between aggregates Surface von Mises stress N m i 85 A 1 2921 x10 x10 0 y v1278 5 a Surface von Mises stress N m N A 1 2921x10 x10 0 L v 1278 5 b Figure 58 Von Mises stress distribution in 3D model 63 6 Conclusions X ray tomography is a non destructive method that can be used to accurately capture the microstructure of asphalt concrete mixtures The image acquired from the X ray tomography can be enhances and the quality can be improved using digital image processing techniques Phase segmentation of the AC microstructure using thresholding operations does not produce accurate phase segmentation results The distance map based watershed segmentation is used to accurately seg mented the phases in the AC microstructure and also to separate ad jacent aggregates Digital image analysis techniques is used to analyze the phase seg mented AC microstructure Digital image analysis is used to determine phase volumetric relations
57. re converted into gray scale images before dig ital processing to reduce computational effort and increase processing speed Digital images are arrays of pixel information in the form of a ma trix with the center of the coordinate at the top left corner In gray scale images the arrays of information contain the gray level of the 14 pixels that make up the image depending on the quantization level Figure 5 show a typical representation of the pixels in a gray scale image where i and j are coordinates and pi is the gray scale value for each pixel Figure 5 Pixel representation in a digital image The pixel values pj is a function of the reflectance of the surface point and intensity of the light falling on the surface point The inten sity of the light falling on the surface can be related to the x ray source characteristics i e x ray energy and tube current while the reflectance of the surface point is a material parameter which can be related to density The combination of the reflectance and the light intensity gives the pixel value The resulting pixel value and pixel neighbour hood relationship makes it possible to process digital images filter and enhance images threshold and segment pixels etc R is the reflectance of surface point I is the intensity of light falling on surface point P is the pixel value at surface point 15 Image processing operation can be divided into two categories the point and mask oper
58. reas with depth 46 Figure 41 3D image showing stones contact areas The orientation of the particles in the asphalt concrete mix plays a vital role in determining the behaviour of the mix under pure and simple shear action In this study Avizo Fire is used to characterize the orientation of each particle in the AC mix The orientation of the particles is defined using the lengthorientation orientation and orien tation measures Length orientation phi is the orientation from the OZ point to the centroid of the particle and Length orientation theta 0 is the orientation of the length maximum of the Feret di ameter in the X Y plane Orientation phi is the direction of the major axis of the particle and Orientation theta 0 is the orientation of the major axis in the X Y plane Orientation2 phi 9 and theta is the same for the Orientation phi and theta 0 except that it makes reference to the minor axis instead of the major axis The minor and major axis are computed with the moment of inertia for the particle 47 Figure 42 Description of orientation measure To describe the different orientation measures a particle is analyzed and discussed below The particle analyzed in this section is made up using the stacked image file format in Avizo R Fire The maximum and minimum Feret diameter of the particle is 13 0mm and 3mm respec tively In Figures 42 and 43 Line AB denotes the axis of the maximum Feret diameter axis
59. result of thresholding operation on the beam hardening corrected image with all the aggre gates captured It can be observed that there is a significant improve 22 ment in the thresholding operation result after the beam hardening correction Figure 14 Result of thresholding operation before beam hardening correction Figure 15 Result of thresholding operation after beam hardening correction 3 4 Filters Filters are mainly used to reduce noise and thereby improve image quality Different types of filters are used in image processing depend ing on the expected result or outcome Sources of noise in digital images include McNitt Gray 2006 1 The beam energy kVp The tube current mA Exposure time Collimation Reconstructed Slice Thickness Reconstruction algorithm or filter Oo a b ow mw Others Focal spot to isocenter distance detector efficiency etc It is important to note that the nature of the mastic makes the segmentation process of the asphalt concrete sample a little cumber some Considering the fact that the mastic is a mixture of bitumen and fines the threshold based segmentation becomes difficult as part of the mastic is identified as aggregates This is as a result of the CT attenuation of the constituents materials in the asphalt concrete mix Image filtering can be used to overcome this problem and thus improve the image segmentation results Lowpass filters are the most common type of filte
60. rmation to accurately de termine the microstructure evolution as a result of loading Void distribution 800 700 600 500 400 300 Area per slice mm 2 200 100 Sample depth mm Figure 39 Density distribution of air voids with depth The contact between adjacent aggregates is captured using a com bination of digital imaging techniques The watershed lines and the 45 aggregate mask used for the segmentation of the aggregate phase is used to determine the accurate location of contacts between aggre gates It is interesting to find out that the air voids distribution is inversely proportional to the contacts area distribution It is worthy to note that the processed sample is a porous asphalt sample that have undergone aggregates segregation as a result of clogging We have more fine particles at the top of the sample and more coarse particles at the bottom There are more inter particle contacts and less air voids at the bottom of the sample In the same sense there are more air voids at the top of the sample and less contacts between the fine aggregates Figure 40 shows the contact area distribution of the sample and Figure 4 shows the contact between aggregates in 3D Contact Area distribution 40 35 30 N E PIN 10 Area per slice mm 2 Sample depth mm Figure 40 Stone contact a
61. rs used in digi tal image processing Lowpass filters lets low frequencies go through but attenuates high frequencies and noise It reduces contrast but also tends to defocus the image and blur the edges Gaussian filter is ap plied to the asphalt concrete CT image and compared with the result when the median filter is applied The median filter shows a better denoising result than the gaussian filter when a 3x3 kernel is used Median filter and anisotropic filter are applied to the beam hardening corrected image for further processing and analysis 24 3 4 1 Gaussian filter Similar to other types of filters low pass filters are used to filter out noise and enhance the quality of the image Lowpass filters are not very effective because of their tendency to blur relevant image details and edge of objects Figure 16 shows the result of the gaussian filter on the beam hardening corrected image I z y Io z y G a y t 4 Clout zh rep E FP 5 The gaussian filter is based on the formulation in Equations 4 and 5 The filter applies a convolution kernel G x y to every pixel in the image The parameters x and y represent the offsets from the center of the convolution kernel If the standard deviations are important the filter coefficients will be close to each others The parameter t is the variance of the gaussian Avizo 2009 Figure 16 Contrast enhanced and gaussian filtered image of Asphalt concrete 25 3 4 2 Media
62. s respectively Cite Ketcham Click on Detector Calibration configure offset map detective pixels and change No of frames acquired to 200 The number of frames refers to the number of slices required from the scanned sample To get rid of defective pixels on the detector we need to take the following steps to make necessary corrections a Start the X ray by pushing the Enable 225kV button on the main console b Place the cursor on a spot in the middle of the screen and adjust the energy intensity using the voltage slider in the FXE control application till you get a raw reading of about 60 000 grey levels c v LED Light W Radios 00760 tif 759 1080 1536x1920 Display zoom SS d Click on Gain 0 and change the of frames to 200 and click on acquire Wait for the acquiring process to complete e Then place the cursor again on a spot in the middle of the screen and adjust the energy intensity like previous till you get a raw reading of about 45 000 grey levels f Click on Gain 1 and change the of frames to 200 and click on acquire Wait for the acquiring process to complete g Then place the cursor again on a spot in the middle of the screen and adjust the voltage like previous till you get a raw reading of about 30 000 grey levels h Click on Gain 2 and change the of frames to 200 and click on acquire Wait for the acquiring process to complete i Then place the cursor again on a s
63. seg mentation results r _ Figure 12 Illumination profile of CT scanned image of Asphalt concrete 21 Figure 13 Beam hardening corrected image of Asphalt concrete There are a number of possible techniques to reduce the beam hard ening in the scanned image which include the use of X ray beam that is energetic enough to ensure that beam hardening is negligible use of filters increased exposure time among others In this study the beam hardening artifact is corrected using the background and flat field cor rection feature in Avizo R Fire The flat field is computed using the bkgimg command in Avizo which computes a background image from a gray level image and a binary mask or no mask for all pixels of the image using second order polynomial The intensities of the 3D input image are then scaled according to the normalized intensities of the flatfield images The input image gets brighter at pixels where the flatfield is dark and vice versa In this way non uniform illumina tion is compensated for Avizo 2009 The background and flat field correction is applied to the contrast enhanced image in Figure 10 Figures 14 15 shows the result of thresholding operation before and after beam hardening correction It can be noticed in Figure 14 that thresholding operation only captures the aggregates at the edges of the image as a result of variation in grey intensities and non uniform illumination of the image Figure 15 shows the
64. ses in the binder reach approximately 2 5MPa 59 Time 3600 Surface First principal strain 1 90 85 80 75 70 65 60 55 50 45 40 35 30 25r 30 40 50 60 70 80 90 vo Figure 54 Thermal strain distribution in AC microstructure Time 3600 Surface First principal invariant of stress N m 90 A 1 2423x 10 85 l w 80 75 70 65 60 55 50 45 40 35 0 5 30 25 30 40 50 60 70 80 90 Y 1 9584x10 Figure 55 Thermal stress distribution in AC microstructure 60 5 2 Three dimensional 3D uniaxial compression Uniaxial compression simulation is used to investigate the stone breakage and polishing during compaction The 3D geometry is ex tracted from the AC microstructure and it consists of two 2 aggre gates interconnected with mastic A uniaxial compressive displacement of 0 01mm is applied at one end of the model with the other end fully restrained as shown in Figure 56 For simplicity both mastic and ag gregates are modeled as linear elastic The material properties for the both mastic and aggregates are taken from Table 5 Figure 56 3D uniaxial compression model Figures 57 shows the strain distribution in the 3D model It can be seen that compressive strains are localized in the binder with a maximum strain of 14 6 It can be seen in Figure 57 that high compressive strains are developed at regions with small mastic thick ness 61 Time 1 Surface Third principal st
65. sisted with the X ray scans Prof Bjorn Birgission is also appreciated for his advise and insight into further research and development in this area of study Pia Lundqvist is also appreciated for her friendly gesture and assistance during the study My gratitude also goes to all my colleagues during the Masters Programme at the Royal Institute of Technology Thank you all for your support as usual Dedication This work is dedicated to the memory of my late dad Yushau Ladipo Onifade Contents 1 Introduction 1 1 Objectives amp workflow 20 wo Rev y HL Xray Computed Tomography 2 1 Mass Total attenuation coefficient X ray Attenuation 2 2 Acquisition of CT data 5 de ak Gh Son he bed Ren Gel 2 2 1 Sample preparation 4 2 2 2 NGAP ATMOS aiu eu Xue Ae t NC ane cht OB ahd is 22 39 Gollection a xum Gh Sed aor ede dope Ge Bd 2 2 4 Reconstruction duae vetet Aad eer sig AS 2 9 Experimental data and scanning procedure Digital image processing and analysis 3 1 Grey level histogram 2 4222232599 o o 3 2 mage contrast enhancement 4 8 4 am RISE 3 3 Beam hardening correction 2 3 4 ro exe Tes Su IGE reati tad dv ie eh See ie et od Rate 3 4 1 Gaussian filter 4 2 ncs amp ELE ADR RR RS 3 4 2 Median filter Sud e qe ut ted e e UT d 3 43 Edge preserving filter 3 4 4 Erosion amp Dilation x 75 Xxx BA BG Re ad 3 5 Segmentation ae Aled A o t ec Aidala wt oy Aveda 3
66. tation Watershed lines used for separation of stones AC sample showing the aggregates mastic and air voids SEC MONE 26 ceto d Sette d set bo ue Ma Bante Ra es Reconstructed 3D image of aggregate Reconstructed 3D image of air voids Description of maximum Feret diameter Distribution of length of stones in the sample Distribution of width of stones inthe sample Volume distribution of stones Density distribution of air voids with depth Stone contact areas with depth 3D image showing stones contact areas Description of orientation measure Description of orientation measure Length orientation of stones theta Orientation of stones Ln Orientation of stones theta o oo aaa a Phase segmented asphalt concrete sample b 2d uniaxial tension model configuration Continuum model Stress distribution in AC microstruc 49 50 51 52 53 54 59 56 57 58 Multi phase model Stress distribution in AC microstruc Stress distribution in AC microstructure Surface temperature of AC microstructure after 1hour COO INO PET PE Thermal strain distribution in AC microstructure Thermal stress distribution in AC microstructure 3D uniaxial compression model Compressive strain distribution in x direction Von Mises stress distr
67. ter and each view is successively superimposed over a square grid at an angle corresponding to its acquisition angle During reconstruction the raw intensity data in the sinogram are converted to CT numbers or CT values that have a range determined by the computer system Most medical and older industrial systems use a 12 bit scale in which 4096 values are possible while more re cent systems use a 16 bit scale which allows values to range from 0 to 65535 The X ray facility in KTH Highway and Railway lab uses a 16 bit scale On most industrial scanners these values correspond to the grayscale in the image files created or exported by the systems Although CT values should map linearly to the effective attenuation coefficient of the material in each voxel the absolute correspondence is arbitrary Industrial CT systems are sometimes calibrated so that air has a value of 0 water of 1000 and aluminum of 2700 so the CT 11 number corresponds roughly with density Johns et al 1993 The calibration of CT values is straightforward for fixed geometry single use systems but far less so for systems with flexible geometry and scanning modes and multiple uses each requiring different optimiza tion techniques 2 3 Experimental data and scanning procedure In this study the KTH X5000 CTX ray CT scanner is used to ob tain the detailed microscopic structure of the porous asphalt concrete core sample for further visualization characterization
68. the judgment of the person carrying out the scan The linear attenuation coefficient of the materials to be scanned can be used to get an idea of the appropriate energy level required for the scan The changes made to the energy intensities are immediately reflected on the image on the computer screen b Settings on the CT rotate 450 Change the number of frames per second fps to a suitable value This is directly related to how much detail you like to capture during the scan A fps of 2 is usually ideal for small objects but a fps greater than 2 will be suggested for large objects like asphalt cores There is a tradeoff between scanning speed and scan quality when considering the number of frames per second When you change the value of the fps check the image on the computer screen if it is satisfactory before proceeding If the image is satisfactory then push the red button on the 450kV console to turn X ray OFF Scanning a After necessary change in the scanning configuration and the scan settings the actual scanning process follows b To start a new scan click on CT Project On the CT rotate 450 application window and the CT Project window opens c On the CT Project window select a location to save the new scan make a new folder and give a name to the new folder Then click OK d Click on CT Scan to start the scanning process Then wait till the scan is finished e Turn off the X ray and take out the object
69. to determine the position of edges in an image Edges are the boundary between 2 objects in an image with distinct intensity values There are two types of edges associated with images the strong and the weak edges Strong edges can be noticed at the interface between two materials while soft edges may represent intensity variations in the same object The sobel and the canny techniques are two edge detection methods commonly used The canny method is the most effective in detecting strong and weak edges while the sobel method is effective for strong edges and not very sensitive to weak edges One issue for asphalt is that we have higher density variations inside the phase than between phases The process of edge detection is based on the principle that the first order derivative of the gray levels with respect to the coordinates 30 at the aggregate and matrix interfaces should have either a maximum or a minimum Yue et al 2003 The magnitude of the gradient is used to determine the strength of the edge at the position x y the magnitude of the gradient and the orientation of the gradient can be estimated using the relationships in equations 6 and 7 respectively Equations 8 and 9 gives the first order derivatives of an image f x y along the x and y coordinates respectively WN G3 G3 6 scat tan 7 where al G SG L9 fea hy 8 3 Gy f y 1 fey 1 9 The second order derivative can be used to accurately det
70. ture and analyzed The numerical simulation is carried out using COMSOL Multiphysics In order to accurately characterize the AC microstructure a linear elastic 2D coupled thermo mechanical analysis is carried out The AC microstructure is first considered as a continuum with same material properties prescribed to the mastic and the aggregate phases The AC microstructure is then modeled as a multi phase material with different material properties assigned to the mastic and the aggregate A ther mal analysis is also carried out on the 2D microstructure to investigate the thermal stresses and strains developed as a result of temperature gradient in the material In the case of the thermal analysis the AC microstructure is modeled as multi phase with the mastic is assigned viscoelastic material properties and the aggregates assigned elastic ma terial properties The 3D FEM analysis is carried out to investigate the stress distribution in the AC matrix under uniaxial compression The interface between the mastic and aggregates is modeled using the identity pairs feature in COMSOL Multiphysics and considered as continuous to enable stress transfer at the interface 52 5 1 Two dimensional 2D finite element analysis The 2D AC microstructure considered in the FEM analysis assumes a plane strain condition The behaviour of the AC microstructure is studied under two 2 loading conditions uniaxial tensile loading and thermal loading due to tem
71. ty close door Functions Warmup Startup Filament adjust Sweepon _ Autocenter kv Autocenter all Allcentering and focus tables valid PXE control application interface after warming up 4 Scanning configuration a Run the CT Rotate 225 program A shortcut can be found on the desktop b Alignment of the scanning platform A red value on any of the adjustment parameters in the CT Rotate 225 program indicates a misalignment of the scanning platform in the x ray chamber Necessary amendments should be made so that all adjustment parameters are in the allowable range For instance the table tilt parameter relates to how the table is tilted in the horizontal The table tilt value should be set to zero 0 to ensure that the table is perfectly horizontal and that the sample is not tilted during the scan Technique development Comet XAG kV 120 4 pA 2170 focal spot microns 0 120 4 kv 2170 pA focal spot Standard interlock OK Not in remote operation ray detector gain O 25pF w fps 200 v bin none v C live averaging 2 Hframe Scan setup src det mm 1092 014 src obi mm 306 704 fixturing filter Mitsubishi PLC ax motion controller rotate C lock other axis CT mode light curtain lear 19297 519 CT rotate degree 3 543 T
72. ue of the neighbouring pixels Both operations can be used to enhance the image quality to get pertinent information These image processing methods can be used in a label field to separate ad jacent aggregate particles and fill holes in the aggregates 3 5 Segmentation Segmentation is a process used to separate pixels with the same gray level value from those pixels with a different threshold This is a very important step in AC image processing as it is used to separate the constituents of the AC mixture A good image resolution and con trast between the gray levels is essential for a good segmentation result In three dimension 3D one of the major challenge in processing 28 and modeling of the AC microstructure is the separation of the in dividual aggregates Improperly separated aggregates can be seen as one big interconnected mass of stone and quantitative analysis of this interconnected stones does not give any meaningful result Performing numerical analysis with such improperly segmented 3D microstructure image does not simulate the true behavior of the 3D mix under loading conditions The 3D model becomes too stiff and rarely deform under normal load conditions Hence the true interaction between the ag gregate and the binder is not well captured and accounted for To overcome this problem different methods were employed for seg mentation of the AC microstructure The distance map based water shed segmentation technique g
73. y of the image or to extract relevant information Different techniques are applicable in image processing and the selection of the appropriate technique or method depends on the application General image processing tech niques for image enhancement includes image contrast enhancement filtering erosion and dilation background correction etc Other image processing techniques like thresholding edge detection watershed are used for segmentation and separation of objects Generally images are two or three dimensional matrix of pixels Colour images are represented as three dimensional matrix of pixels with the first two dimensions representing the pixel location and the third dimension representing the red blue and green component of the pixel while gray scale images are represented as two dimensional repre senting gray scale intensities in the image A pixel Picture Element is a single point in a raster image or the smallest addressable screen element in a display device Wikipedia Colour images contain more information than gray scale images and hence more time and computa tional power is required for processing colour images Gray images are usually encoded as 8bits and hence it contains 256 gray levels whereas colour images which contains different shades of red green and blue RGB are encoded as 24bits with 8bits for each colour Coloured images contains three times more information than gray scale images Usually colour images a

Download Pdf Manuals

Related Search

Related Contents

"5 - Mamanlou.com    Algemene verkoopvoorwaarden  CXR120/200  TomTom BRIDGE  

Copyright © All rights reserved. Failed to retrieve file