Study of gas pipelines cracks using transmission and Compton scattering

Most of the natural gas production is transported through pipelines that require periodic inspections to evaluate the structural integrity of the pipelines due to possible defects caused by degradation that can rupture causing leakage of the fluid causing major disasters. Based on this, the project presents a methodology for predicting cracks in pipe used in gas pipelines. The approximation is based on the principles of gamma densitometry to calculate the density of the pipe wall in order to investigate possible cracks. The natural gas fluid is found in such systems and interferes in the density calculations and therefore will be considered in the simulations. The detection system uses a narrow beam geometry appropriately, comprising gamma ray source (Cs) and NaI (Tl) 3"x3" detectors for calculating transmitted and scattered photons. Different positioning angles of the detector are investigated. In this study, the MCNP-X code is used to perform the simulations, in order to develop a counting geometry. Simulations of different thicknesses of the crack were also used to determine the minimum thickness detected by the two NaI (Tl) detectors. Having equipment that can estimate cracks present in pipes used in gas pipelines, in addition to predicting their location can reduce costs and make a major contribution to this sector.


INTRODUCTION
Due to the growing demand, natural gas is a source of energy that has been showing growth in the Brazilian energy matrix for use in industry, homes, motor vehicles and the electric sector. The development of this matrix is essential for industries to become competitive in order to increase operational efficiency and reduce costs. To do so, investments are needed in the expansion of the transmission network and in the distribution of this source of energy.
Much of the natural gas production is transported by means of gas pipelines that require periodic inspections to evaluate the structural integrity of the pipelines due to possible defects caused by degradation that may rupture causing leakage of the fluid causing large-scale environmental disasters with population damage and the economy of the local region. In Brazil, there are a total of 9,700 km of gas pipelines, 7,107 km operated by TRANSPETRO (TRANSPETRO, 2014) and 2,593 km operated by TBG (TBG, 2015), through 443 municipalities, with a transportation capacity of 130 million m 3 /d natural gas, making the control of gas pipeline integrity important to predict abnormality that could cause future risk to people and the environment.
Using conventional technology to perform preventive or corrective measures, crack detection can occur through internal pipeline monitoring by a variety of techniques, such as: instrumented, magnetically based Pipeline Inspection Gauge (PIG) is the most widespread tool for corrosion inspection of ducts. It uses changes in magnetic field to detect changes in duct thickness (Silva, 2015).
However, the cost to the company for the launch and follow-up of the instrumented PIG passage in the pipelines is very high, in the order of US$ 2,000.00 per km of pipeline (Carlos et al., 2002), besides the possibility of losing monitoring of the PIG when there is variation in the velocity of the fluid, above 3.0 m/s and below 1.0 m/s of that allowed for PIG monitoring.
In the context of Nuclear Techniques, a potential solution for the preventive control and evolution of cracks has been shown to be adequate to monitor and quantify the crack by means of the density of the material being analyzed, such as: gamma ray densitometry. The method is based on the change in attenuation of the gamma radiation caused by the presence of internal or external discontinuities when the radiation passes through the material and is recorded by specific detectors. Density measurement is based on the absorption of gamma radiation as it passes through the material to However, in this type of measures difficulties are encountered, such as: extension and large diameter of the pipes, presence of fluids that due to differences in density interfere with the accuracy of density estimation. In contrast, in Compton scattering, the detector and the source can be arranged on the same side of the tube and an image of the cross-section of the duct can be directly obtained from the scattered radiation (Sharaf, 2001). Photons scattered alone from a well-defined volume of a sample contain information on the density of the material being analyzed (Arendtsz et al., 1995).
Theoretical models were developed using the mathematical code MCNP-X based on simulations using the Monte Carlo method to develop a suitable counting geometry for crack detection by calculating density.

GENERAL CONSIDERATIONS
Devices based on nuclear technology allow detailed "observation" of flow characteristics, especially density, distribution of the solid particles, concentration in the medium and the speed of displacement, among others. The method is based on the attenuation of a gamma-ray monoenergetically transmitted/scattered beam by a tubing containing fluid being the signal recorded by detector(s).
The two methods of measurement: transmitted and scattered beams will be tested.
In the transmitted beam, a sealed, shielded monoenergetic gamma radiation source with sufficient energy to penetrate and be transmitted throughout the tube is positioned on one side of the pipeline and a NaI(Tl) scintillation detector is generally installed, diametrically opposite the source for the Where I is the transmitted intensity of gamma rays (photons cm -2 .s -1 ), Io is the initial incident intensity of gamma rays from the source (photons. cm -2 .s -1 ), E is the incident radiation energy, X is the material thickness D duct, G gas, F fissure (cm), and  is the total attenuation coefficient D duct, G gas and F fissure, (cm -1 ).
In Compton scattering, where the energy of the scattered photon is smaller than that of the incident photon; that is, the incident photon transfers a fraction of its energy to an electron from the weakly Compton scattering is the largest mechanism of interaction. The probability of Compton scattering is directly proportional to the photon energy and inversely to the atomic number. The number of scattered photons reaching the detector can be obtained by Equation 3.
Where t is the count time (s), I0(E) is the incident photon flux with energy (E), de KN /d is the Klein-Nishina differential shock section in energy (E) for a free electron, which is the scattering probability of a photon, d is the solid angle subtended by the detector as seen from the interaction point,  is the density, Z is the atomic number, NA is the Avogadro's number, A is the mass number of the material on analysis, 1 e 2 are the linear attenuation coefficients for primary and scattered photons within the sample,  is the detector photopeak counting efficiency in scattered photon energy, l1 e l2 are the path lengths of the photons in the source sample to the scattering center and back to the detector and dV is the differential volume considered for radiation and its interaction with matter.
In the case of a homogeneous sample and geometry with point source, the terms of the integrals are constant. In addition, the Klein-Nishina differential shock section will be constant for a fixed geometry, a given incident photon energy and flux. Therefore, the counting rate depends only on the density of the material examined and information obtained by this technique is strongly dependent on this density, so that the variation of density within the sample can be monitored. The simulation consisted of a carbon steel pipeline 5" in diameter and 0.17" thick, inside the pipeline the natural gas fluid was used, as reference was the natural gas delivered to Companhia de Gas de São Paulo (COMGAS), with a density of 0.766 kg.m -3 (@ 20 ° C, 1 atm). A source of 137 Cs (662 keV) with narrow beam was appropriately used in the simulation with the angle of 4.29º divergence. Three lead collimated 3"x 3" Nal (Tl) detectors were properly positioned, the first at 180º (D1), the second at an angle of 30º (D2) and the third at 60º (D3) as shown in Figure 1.

Figure 1: Simulated Geometry
The detector D1 was used to measure the attenuation of the transmitted beams, and the others were used to measure the scattered beams.

Potentiality of the code in scattering calculation
In order to evaluate the potential of the MCNP-X code in the scattering calculation, a geometry formed by a solid aluminum tube was developed, simulating a scattering bar with a diameter of 1 cm. The same 137 Cs source (662 keV) was simulated appropriately with the angle of 4.29° divergence and a scatter detector was positioned at 30° and 60°. The results obtained were compared with theoretical data obtained by Equation 2. The study also considered the influence of the scattering of the 5"carbon steel pipeline and the contribution of the natural gas fluid inside this pipeline, in the D2 and D3 detectors.

Determination of cracks
For the study of the fissures in the pipelines, fissures were added in the rectangular (parallelogram) format of 3 mm of thickness, 50 mm of height, the width was changed from 2 mm to 16 mm, in steps of 2 mm. The first simulation was done without the crack. As shown in Figure 3.

RESULTS AND DISCUSSION
In the analysis of the study of the pontenciality of the code in the scattering calculation, it can be observed that according to Equation 2, the scattered energy referred to 662 keV at 30º and at 60º is 564 keV and 411 keV, respectively. In the simulation performed, the scattering energy detected by  In Figure 5 show the influence of the carbon steel duct on the scattering. As can be observed, the scattering generated by the carbon steel is distributed in two peaks, 480 keV and 610 keV for the detector D2 and 310 keV and 480 keV for the detector D3.
Where y is the numbers of photons detected by detector D1 and x is the crack width.
In the detectors D2 and D3, the "behavior" of the scattering photons in relation to the crack width variation was analyzed, and as can be observed in Figure 8, there was an increase in the number of photons scattered as a function of reduction of the crack width, as can best be seen at the energies of 480 keV for the D2 detector and 310 keV for the D3 detector.
Where y is the numbers of scattering photons detected by detector and x is the crack width.

CONCLUSION
The MCNP-X code was shown to be an important and useful tool to continue the studies of cracks analysis, given that it was possible to carry out the surveys of the scattering energies correctly.
The model developed in the code represented satisfactorily the scattering beam.
It was possible to identify that the nuclear techniques are sensitive for determination of crack size, which in this study was 2 mm to 16 mm, being possible the differentiation between them, by means of the adjusted curves. For future work, will be changed the position of the crack and will be used Artificial Neuronal Networks (RNAs) to identify its location.

ACKNOWLEDGMENT
I hereby express my gratitude to the Institute of Nuclear Engineering, without which this work would not have been possible.