Computational simulation has been considered a

2007 International Nuclear Atlantic Conference - INAC 2007 Santos, SP, Brazil, September 29 to October 5, 2007 ASSOCIAÇÃO BRASILEIRA DE ENERGIA NUCLEAR – ABEN

DIGITAL RADIOGRAPHY SIMULATION FOR INDUSTRIAL APLICATIONS WITH MCNPX Edmilson M. de Souza1, Samanda C. A. Correa1, Ademir X. da Silva2, Ricardo T. Lopes1 and Davi F. Oliveira1 1

[Programa de Engenharia Nuclear]/COPPE, Universidade Federal do Rio de Janeiro Ilha do Fundão, Caixa Postal 68509, 21945-970, Rio de Janeiro, RJ, Brasil 2

[PEN/COPPE - DNC/Poli]CT, Universidade Federal do Rio de Janeiro Ilha do Fundão, Caixa Postal 68509, 21945-970, Rio de Janeiro, RJ, Brasil [email protected]

ABSTRACT The energy dependent response of a BaFBr Image Plate detector was modeled and introduced in MCNPX radiography tally input. To convert MCNPX radiography tally output in 16 bits digital images, a post processing program called PROGRAMA IMAGEM is presented. Simulate images of a steel tube containing corrosion alveoli and grinded defects were compared with experimental images. The radiography technique used in all tests was double wall single image, DWSI, using an Iridium source (192Ir) touching the adjacent wall. Visual and perfilometric analysis showed that the methodology used for sensible material simulation and data post-processing makes simulate digital images comparable to experimental images.

1. INTRODUCTION The computational simulation of a complete radiographic system has been of much interest for industrial applications. Nowadays, with the advent of a specific tally for image simulation, the Monte Carlo code MCNPX [1] becomes a potential tool for radiographic test modeling. However, although the MCNPX code offers a specific radiography tally, different steps must be taken to convert MCNPX radiography tally output in digital image, and make it comparable to experimental image. The objective of this work is to propose a methodology for digital radiography simulation for industrial applications using Monte Carlo code MCNPX. The energy dependent response of a BaFBr Image Plate detector and the linear response curve of a 16 bits digital systems will be considered. 2. MODELED SYSTEM 2.1 Image Plate Detector Models The basic theory of the MCNPX radiography tally is well presented in [1] where the resultant image is an aerial photon distribution leaving the simulate test object mapped in two dimensions, forming a virtual image or pseudo digital image. In order to reproduce the photon absorption rate in an image plate systems, the energy dependent

0.20

2

Absorbed Energy / Fluence (MeV.cm /g)

response of a BaFBr Image Plate detector was modeled using the DE and DF MCNPX cards and introduced in MCNPX radiography tally input. Fig. 1 shows the energy absorption curve obtained from BaFBr detector simulation. The energy peak absorption obtained in the simulation was 38.4 keV. The energy peak absorption from BaFBr found in the literature is between 37.4 keV and 40 keV [2,3]. The good agreement between simulate and the theoretical data was a necessary requirement to make the comparable simulate and experimental images.

0.18 0.16 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Energy (MeV)

Figure 1. Energy absorption curve from BaFBr detector obtained by Monte Carlo simulatons.

2.2 Data Post Processing Program Although the MCNPX code offers a friendly interface to the user, different steps must be taken to data processing and to form the final simulate image, due to the great amount of information in radiographic MCNPX output that does not contribute to the image reconstruction. To solve this, a post processing program named PROGRAMA IMAGEM was developed. The PROGRAMA IMAGEM is an evoution of a primitive post processing program developed in the Labview version 8.0, named PROGRAMA MATRIZ [4], that converts the radiographic output files from MCNPX to forms compatible with various external graphics packages. The PROGRAMA IMAGEM interface is shown in Fig. 2.

INAC 2007, Santos, SP, Brazil.

Figure 2. PROGRAMA IMAGEM interface.

Besides having all PROGRAMA MATRIZ characteristics, the PROGRAMA IMAGEM also permits to convert MCNPX radiography tally output in 16 bits digital images. Another important characteristics of the PROGRAMA IMAGEM are to resample the image grid output through the nearest neighbour interpolation method, to combine images and display in its interface the distribution of incident radiation intensity in each image pixel. 3. COMPARISON BETWEEN THE SIMULATE AND EXPERIMENTAL IMAGES The methodology for digital radiography simulation was tested with real radiography experiments through the visual and perfilometric comparisons of the contrast sensitivity of the real and simulated computerized radiography system.The experimental tests were performed using an Acr2000i Kodak system and an 53.65 GBq (1.45 Ci) activity Iridium source (192Ir) with diameter of 0.3 cm. The radiography technique used in all tests was double wall single image, DWSI, with the source touching the adjacent wall. Kodak standard phosphor imaging plates were used in the expositions. Image processing in experimental images was performed using Image Pro Plus software (Media Cybernetics). The irradiated test specimen consists of the steel pipe with 25.4 cm diameter, 2 cm thickness and 25 cm height containing artificial defects such the corrosion alveoli and stress corrosion cracking. Fig. 3 shows the standard and the engineering design of artificial defects on pipe reference with the dimensions. The exposure time was 15 minutes.


Figure 3. Simulation configuration of steel pipe with corrosion alveoli and stress corrosion cracking. The holes correspond to the alveoli and the slots correspond to the cracks.

As the aim of the present work was to simulate steel pipe image, the MCNPX radiography tally TIC was used. In this study the images were obtained simulating a detector grid with pixel sizes of 0.01 cm (100 µm pixels) for unscattered images and 0.1 cm for scattered images. The scattered image and the unscattered image were combined using the PROGRAMA IMAGEM. 4. RESULTS Fig. 4a-4d show the simulated and experimental images of the corrosion alveoli and stress corrosion cracking inserted in the pipe. The simulated images were reconstructed using imagesc Matlab routine.


(a)

(c)

a

a

b

b

c

c

(b)

(d)

e d c b

a e d a

c b

Fig. 4. Simulate and experimental images. (a) Simulate image from corrosion alveoli; (b) Experimental image from corrosion alveoli; (c) Simulate image from stress corrosion cracking; (d) Experimental image from stress corrosion cracking.

It is clearly seen that the simulations predicted the qualitative results of the experimental image very nicely. All holes and stress corrosion cracking were detected in addition to the first, less deep. This has been quite a positive result considering sample thickness. The gray scale profiles obtained in conformity with the lines defined in Figure 4 are shown in Figure 5. These profiles were used to see the differences in gray scale between the simulated and experimental images.


c

35

(a) b

30 25 20

a

15 10 5

(b)

Normalized Gray Scale


40

0

0

200

400

600

800

45 40 35 30 25 20 15 10 5 0

Pixels

c

(c) b a

0

200

400

600

800

Pixels


e

50

d

40

d

c

30

c b a

b

20

a

10 0

(e)

e

0

175

350 Pixels

525

(f)


Pixels

60

(d)



Pixels

700 Pixels

Fig. 5. Profiles of the simulated and experimental images. (a) Profile of the simulate image for 20mm holes; (b) Profile of the experimental image for 20mm holes; (c) Profile of the simulate image for 10mm holes; (d) Profile of the experimental image for 10mm holes. (e) Profile of the simulate image for slots: 2mm wide series and 1mm wide sequence; (f) Profile of the experimental image for slots: 2mm wide series and 1mm wide sequence. The gray scale was normalized by factor 1000.


Good agreement between the simulated and measured profiles can be observed in the holes and slots regions of the pipe, as shown the doted lines and the dashed circles in Fig. 5. The small differences found between gray scales for the smallest peaks and the noise level in measured profiles which increases with the wall thickness are due the processing stages of the stimulated signal from the image plate of the scanner manufacturer. The gray levels in the image depend on the characteristics of the Analogue to Digital Converter (ADC) used in the IP reader, that it was not taken into consideration in the simulation [2,3]. 5. CONCLUSIONS The methodology for digital radiography simulation using the MCNPX radiography tally presented in this work demonstrated to be particularly useful to make predictions on the detectability of different imaging parameters and geometries successfully to image indicative details such as wall thickness diminution caused by corrosion and cracks. This kind of information is very important to radiography setups where the sample can not be easilly investigated, providing additional guidance to experts in the field. ACKNOWLEDGMENTS This work was supported by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) and the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq). REFERENCES 1. Pelowitz, D. B. ed. MCNPXTM User’s Manual, Version 2.5.0. Los Alamos National Laboratory report LA-CP-05-0369, April. 2005. 2. Rowlands, J. A. “The Physics of Computed Radiography”, Phys Med Biol, v. 47, pp. R123-R166 (2002). 3. Tingberg, A., Sjostrom D. “Optimization of image plate radiography with respect to tube voltage”, Radiation Protection Dosimetry, v.114, pp. 286-293 (2005). 4. Souza, E. M., Correa, S. C. A., Lopes, R. T., Silva, A. T. “Development of a Data Post Processing Program of Image Simulation with MCNP5”, Proceedings of the XVIII Imeko World Congress. Rio de Janeiro, Sept (2006).
