Journal of Sensors

Journal of Sensors / 2016 / Article

Research Article | Open Access

Volume 2016 |Article ID 8126214 |

Qingshan Feng, Rui Li, Hong Zhang, "Modeling and Calculation of Dent Based on Pipeline Bending Strain", Journal of Sensors, vol. 2016, Article ID 8126214, 8 pages, 2016.

Modeling and Calculation of Dent Based on Pipeline Bending Strain

Academic Editor: Guiyun Tian
Received27 Jan 2016
Revised10 Mar 2016
Accepted03 Apr 2016
Published16 Aug 2016


The bending strain of long-distance oil and gas pipelines can be calculated by the in-line inspection tool which used inertial measurement unit (IMU). The bending strain is used to evaluate the strain and displacement of the pipeline. During the bending strain inspection, the dent existing in the pipeline can affect the bending strain data as well. This paper presents a novel method to model and calculate the pipeline dent based on the bending strain. The technique takes inertial mapping data from in-line inspection and calculates depth of dent in the pipeline using Bayesian statistical theory and neural network. To verify accuracy of the proposed method, an in-line inspection tool is used to inspect pipeline to gather data. The calculation of dent shows the method is accurate for the dent, and the mean relative error is 2.44%. The new method provides not only strain of the pipeline dent but also the depth of dent. It is more benefit for integrity management of pipeline for the safety of the pipeline.

1. Introduction

With the development of the oil and gas production, the long-distance buried pipeline is used to the transport the production of oil and gas [1, 2]. Due to the reasons of time accumulated or construction, defects such as corrosion, gouge, dent, and displacement seriously threat the safety operation of the pipeline. In order to reduce the risk of these defects, the pipeline company usually selects in-line inspection tool to inspect the defects for body of pipeline. The magnetic flux leakage (MFL) and ultrasonic tool are used to inspect metal loss of pipeline such as corrosion, crack, and gouge.

Dent is another common defect in long-distance buried oil and gas pipeline. A lot of dents have been found in the pipeline due to the construction of mechanical damage. Dents can deform the pipeline and affect the integrity of body for the pipeline [36]. Specifically, the dent located in the weld of pipeline or the dent that is very sharp can lead to the serious leak consequence. The information about the severity of the damage to a pipe due to dents is very important for the pipeline industry. The defect of dent is the radial deformations of the pipe wall. Generally, these dents can be measured by in-line inspection tools named geometry tool which equipped with instrumentation and mechanical fingers is able to provide data on the pipe wall geometry deformation. However, this tool only inspects the size (length, depth, and width) of the dent. In order to assess the hazard for the dent, the technical requires a complex modeling and computes the strain or other information. Another method for dent testing can be using the profile gauge to measure the depth, width, and length of the gouge or scratch damage. This method also can be used to determine the extent of damage and determine if any sharp edges are present. Recently, 3D laser scanner is used to map and measure the dent. This technology can more accurately measure the dimensions of the dent by the laser scan. However, the method for dent testing of profile gauge and 3D laser scan only use dents which have been dug. They are the verification for dents which are inspected by geometry tool.

Nowadays, another Pipeline In-Line Inspection tool which installed inertial measurement unit (IMU) is used to inspect the pipeline [710]. As shown in Figure 1, this tool, which consists of a Data Acquisition System (DAS), an IMU, some weld detectors, and odometers, is driven to move through the pipeline and collect data regarding dents, bends, and navigation. Petrochina Pipeline Company has developed an IMU system based on a navigation system, which is suitable for a geometry PIG. The IMU can be used to record the attitude data for the tool during the inspection. A method of calculation for pipeline centerline is proposed by fusion for multisensors. Czyz et al. [11, 12] developed a pure geometry-based algorithm for the bending strain of a pipeline. Another pipeline strain testing method is using the strain gauge to be installed on the surface of pipeline. Two types of strain gauge, resistance and optical fiber, are used to measure the strain of pipeline. The strain gauge of optical fiber is more precise for measuring the strain. These two methods can be used to monitor the strain changes of the pipeline. But the limit for these two methods is only used to measure the part sections of pipeline. They cannot inspect the whole strain for long-distance buried pipeline.

The bending strain can be reported for the whole pipeline when the tool runs for once. Generally, the technical just uses the bending strain to detect the position of pipeline enduring the extra strain larger than the operation requirement and assessing the displacement for local pipeline. However, it is found that the bending strain can be affected by the dent [1316]. In this paper, a method for modeling and computation of dent which is based on bending strain is proposed. The technique takes Bayesian statistical theory and neural network to improve a new method to compute the depth of the dent. It is more useful to assess the dent for the pipeline integrity.

2. The Calculation Method of the Pipeline Bending Strain

In this section, the calculation method of the pipeline bending strain is described. Several factors such as internal pressure, temperature differential, external loads, and boundary conditions are affecting the strain. The strain can be consisted of two primary components: longitudinal and hoop strain [1719]. The bending strain is one of the components for longitudinal strain. It can be described by the following formula:where is the total bending strain, is the total curvature, is vertical curvature, and is horizontal curvature.

The total curvature of the centerline of a pipe is described at each point along the pipeline by the curvature vector. In order to calculate the pipeline curvature, the centerline of a pipe is considered as a 3D parametric curve described in a Cartesian system by a vector , which is a function of a distance along the curve [20, 21]:

Assume that the vector is tangent of , separating the vertical and horizontal curvature components as shown in Figure 2. The calculation of the pipeline bending strain can be given aswhere the pitch and azimuth of the pipeline centerline can be measured by the PIG.

Assume that the vector is the curvature vector of a 3D curve at a given point, and consists of the vertical curvature and the horizontal curvature , which can be given as

The above equation can be written separately for each component of the curvature vector in the Cartesian system:

Based on (1)–(4), the components of the defined curvature vector can be calculated as follows:

The vertical curvature and the horizontal curvature can be given as

From (7), it can be seen that the pipeline bending strain can be calculated with the attitude data, which can be acquired with a PIG.

3. Modeling and Computation of Dent

In Section 2, a computation of the pipeline bending strain is discussed. The total of the features such as displacement and dent based on bending strain can be computed and identified. However, it is difficult to obtain the depth of dent by the bending strain measurement. Because of the lots of factors, it is hard to find a simple relationship between bending strain and size of dent. Neural network is a good solution for this practical project. It is based on complex connection of large number of neurons and deal with difficult language of the modeling information by self-learning, self-organized, and nonlinear dynamics.

3.1. BP Neural Network

The BP neural network consists of an input layer, a hidden layer, and a linear output layer. Each layer is composed of several neurons [2224]. The most basic three-layer BP neural network structure is shown in Figure 3. It is assumed that each layer consisted of “” processing elements; the training set included “” samples mode for (,). For the “” training sample, the sum of total input is for element. If the output is , it follows where is the weight for neurons between and . is the function as

The error of neural network is as follows:where is expect output of the output for input of . If the learning rule used a method for gradient descent, the weights can be changed by error function. The modified function is as follows:where is the times of learning, is the learning factor, and is the value of modified error. is the modify error for output and is the modified error for hidden layer.

3.2. Improving the Neural Network Based on Bayesian Theory

The traditional neural network has difficulty in controlling its model complexity, leading to network overfitting, long time training, and network model low stability [25, 26]. However, the Bayesian neural network based on Bayesian reasoning can solve these problems effectively through correcting network training performance function.

In the actual complex system, the input quantity includes given input quantity and unknown disturbance quantity such as (noise signals), and normally the unknown disturbance quantity has an effect on the system output [27, 28]. Let as an observable given input; then the relationship between system output quantity and input quantity can be described by the following formula:where represents the effect of input on the output and the random quantity submitting to some distribution. The training performance function of neural network adopts mean square error function. Assume error function is as follows:where is the sample numbers, is the neural network output numbers, is the expectation output, and is the network actual output. Generally the not unique solution may lead the neural network training to fall into partial minimum value. To solve this problem, a constraint item can be introduced to make function have interpolation ability, and then the solution for is stable. The condition that has interpolation ability is that is smooth. While the square sum of network weight and threshold is small, its output is smoother. Therefore constraint item should represent smooth constraint, and the objective function is expressed as follows:

is the neural network weight and threshold numbers, is network weight and threshold, is the square sum of weight and threshold, is the square sum of network actual output and objective expectation value residual, and and are hyperparameter which decide the value size of the neural network objective performance function and control the distribution form of weight and threshold. Through using new objective performance function, the network weight and threshold are as little as possible, on condition that the network training error is as small as possible. According to Bayesian, posterior distribution probability of and meets the following formula:

is a sample data set which consists of samples, is neural network hidden layer number, is hyperparameter prior probability, is normalization factor, and is likelihood function. After partial derivatives to , , the most salient hyperparameter can be carried out:

is the number that can reduce parameters numbers (0–W) in performance index function parameters in network. is the weight and threshold when is minimum. is the subscript for . is the weight and threshold when is minimum. After working out the most salient hyperparameter , determine whether is convergence again. Through iterative judgment, the best and most salient neural network model can be obtained.

Relative to traditional neural network, the Bayesian neural network is focused on the probability distribution of the whole parameter space, and the predicted results are based on the statistical average of parameters posterior distribution. A single model is corresponding to one point in the space; all models are corresponding to the whole parameter space. Therefore the Bayesian neural network guarantees network stronger generalization in theory.

3.3. Modeling and Computation for Dent by Bending Strain

A typical signal of dent feature from bending strain is shown in Figure 3. The horizontal axis represents the dent length along the pipeline and the vertical axis is the bending strain for dent, respectively. The bending strain can be represented for the severity for the dent. To the identity of the dent feature, the input can be related to the bending strain information for the dent. The output selected the actual depth of the dent from the site measurement.

According to the relationship for the bending strain and dent as shown in Figure 3, pattern in a plot of bending strain against distance shows the characteristic of environmentally induced deformation for dent. The pattern is caused by a primary action in the center, with a reaction (bending strain in the opposite direction) on both sides. In the case of a dent the central deformation is an underbend with overbends at either end. The main influencing parameters are including the beginning bending strain and ending bending strain which is related to the reaction area of dent along the pipeline, respectively. The largest bending strain is related to the deepest of the dent. is the distance for reaction of the beginning bending strain for dent. is the distance for primary action for dent. is the distance for another reaction of the ending bending strain for dent. The process of modeling and computation for the dent based on Bayesian theory neural network is shown in Figure 4.

4. Field Test and Data Analysis

4.1. Equipment and Performance

To test the proposed method experimentally, an in-line inspection tool (shown in Figure 5) with the proposed method is used to inspect oil pipeline, which is about 150 kilometers. About 150 dents have been dug to measure and verify depth gauge in past five years. These actual data can be selected to be used for sample for the proposed algorithm.

The IMU pipeline ILI tool mainly consists of the following sensors to be used to calculate the bending strain and position of the pipeline:(1)Inertial measurement unit is the main device to gather the ILI data. Three gyroscopes and three accelerometers orthogonally mounted on the IMU to measure the attitude and positon for ILI tool. These gathered data can be used to calculate the pipeline bending strain with other sensors mounted on the ILI tool.(2)Two odometers mounted on the ILI to measure the distance and instantaneous and average speed, which can be used to modify the inertial system errors.(3)Two weld detector sensors are used to measure and record time of passing each girth weld for tool. They can use the alignment for every spool for repeat inspection.

As is known, most domestic oil pipelines are heating transportation and pass through mountains, hills, rivers, and other complex environmental areas [29]; the technical and safety requirements of electrical equipment are extremely strict. To safely inspect the actual pipeline, the IMU should meet not only the technical performance, but also the actual situation of the pipe and the external environment should be considered. Technical performances of inertial devices, which are used in the field test, are shown in Table 1.


Random walk/

AccelerometerBias stability<50 g
Scaling factor<50 ppm

OdometerScaling factor<0.3%
White noise<0.1 m/s

LandmarkWhite noise1 m

4.2. Data Analysis

According to Section 2, a typical bending strain for dent is shown in Figure 6, whose depth is 52.79 mm for 6.5% of outer diameter measured with gauge in field. The bending strain for this dent is −0.4125% which is located in bottom of the pipeline.

For the method of Section 3, 120 data sets for bending strain of dent are selected for the input of the new neural network. Part of dents of training data are shown in Table 2. The input of neural network is consisted of , , , , , and introduced in Section 3.3. The output of neural network is the depth of the dent.

Depth ratio (%)Depth (mm) (%) (%) (%) (m) (m) (m)


The measured data input into the BP neural network to calculate the result of the dent. Parts of sample data input into the network for training and the other sample data to verify the accuracy. The calculation is shown in Table 3 and Figure 7. The mean relative error for BP calculation and actual depth is 12.26%. Although the BP neural network can be used to calculate the depth from the bending strain, the accuracy degree is low to assess the dent.

Actual depth (mm)Calculation of BP (mm)Error for calculation of BP and actual depth (mm)Relative error (%)


To verify the accuracy of the proposed method, all of data input into the modified Bayesian neural network. The neural network can be trained by the sample data. Mean square error (MSE) is selected to assess the quality of the neural network. To adjust the parameter of neural network, the result of training is shown in Figure 8. The best validation performance of MSE for the method is 0.0040885 at the epoch 95.

The calculation of modified neural network based on Bayesian is shown in Figure 9 and Table 4. It is clear to see the method can be used to calculate the depth of dent from bending strain. From Table 4, the mean relative error for calculation and actual depth is 2.44%. The accuracy of modified neural network is raised more than BP neural network. The calculation of modified Bayesian neural network can be used to calculate the depth of dent from dent. With this method, the pipeline company runs the IMU tool for once not only to obtain the pipeline bending strain but also to calculate all of the depths of dent. The proposed method offers a useful method for pipeline integrity evaluation.

Actual depth (mm)Calculation of Bayesian (mm)Error for calculation of Bayesian and actual depth (mm)Relative error (%)


5. Conclusions

The in-line inspection tool, which loads the IMU, used to inspect the centerline of the pipeline. The attitude information can be used to compute the bending strain of the pipeline. However, there is no paper or report to research the relationship between the bending strain and dent. In this paper, based on the analysis of the calculation method of pipeline bending strain, we propose a method based on modified Bayesian neural network to calculate dent depths from bending strain. To test the proposed method experimentally, a PIG with the proposed method is used to inspect a 150 km pipeline. It can be obtained that(1)A new method is proposed to verify the relationship between pipeline bending strain and dent based on the calculation of bending strain.(2)According to the characteristic of signal between dent and pipeline bending strain, a new model is presented in detail to be used for proposed calculation of algorithm.(3)The calculation of modified Bayesian neural network is proposed to compute the depth dent with pipeline bending strain to compare with actual data. And traditional BP neural network is used to calculate the depth of dent. According to the result of calculation, the mean accuracy of calculation for modified Bayesian neural network is better than BP neural network for 9.82%.(4)According to calculation, the proposed method is more accurate and suitable for the calculation of depth based on pipeline bending strain. The mean relative error for calculation of dent depth based the modified Bayesian neural network is 2.44%.

This paper provides a novel method for calculating dent depths from the pipeline bending strain. The dent can be evaluated by the bending strain not used to dig or use another tool to reinspect. The bending strain and calculation of dent depth is also useful to the evaluation of pipeline integrity.

Competing Interests

The authors declare that they have no competing interests.


This work was supported by the project of Petrochina Pipeline Company “the research of safety service of buried pipeline in permafrost region.”


  1. Q. Aihua, “Chinese oil and gas pipeline transport development status and analysis of associated problems,” International Petroleum Economics, vol. 17, no. 12, pp. 17–12, 2009. View at: Google Scholar
  2. L.-J. Yang, R.-Q. Li, S.-W. Gao, and Y. Yang, “Navigating and positioning technique for inner detection of pipeline,” Journal of Shenyang University of Technology, vol. 34, no. 4, pp. 427–432, 2012. View at: Google Scholar
  3. S. Sha and Q. Feng, “Excavated verification technology of pit defects of buried pipeline,” Oil and Gas Storage and Transportation, vol. 8, pp. 834–838, 2014. View at: Google Scholar
  4. H. Ghaednia, S. Das, R. Wang, and R. Kania, “Safe burst strength of a pipeline with dent-crack defect: effect of crack depth and operating pressure,” Engineering Failure Analysis, vol. 55, pp. 288–299, 2015. View at: Publisher Site | Google Scholar
  5. B. Pinheiro, I. Pasqualino, and S. Cunha, “Fatigue life assessment of damaged pipelines under cyclic internal pressure: pipelines with longitudinal and transverse plain dents,” International Journal of Fatigue, vol. 68, pp. 38–47, 2014. View at: Publisher Site | Google Scholar
  6. M. Allouti, C. Schmitt, and G. Pluvinage, “Assessment of a gouge and dent defect in a pipeline by a combined criterion,” Engineering Failure Analysis, vol. 36, pp. 1–13, 2014. View at: Publisher Site | Google Scholar
  7. S. Paeper, B. Brown, and T. Beuker, “Inline inspection of dents and corrosion using ‘high quality’ multi-purpose smart-pig inspection data,” in Proceedings of the 6th International Pipeline Conference (IPC '06), pp. 243–248, September 2006. View at: Publisher Site | Google Scholar
  8. X. Wang and H. Song, “The inertial technology based 3-dimensional information measurement system for underground pipeline,” Measurement, vol. 45, no. 3, pp. 604–614, 2012. View at: Publisher Site | Google Scholar
  9. L. Mengjie, “The study of accurate in-line inspection technology to offshore pipeline route,” China Offshore Platform, vol. 19, no. 6, pp. 46–49, 2004. View at: Google Scholar
  10. J. Yu, J. G. Lee, C. G. Park, and H. S. Han, “An off-line navigation of a geometry PIG using a modified nonlinear fixed-interval smoothing filter,” Control Engineering Practice, vol. 13, no. 11, pp. 1403–1411, 2005. View at: Publisher Site | Google Scholar
  11. J. A. Czyz, C. Fraccaroli, and A. P. Sergeant, “Measuring pipeline movement in geotechnically unstable areas using an inertial geometry pipeline inspection pig,” in Proceedings of the ASME 1st International Pipeline Conference, Calgary, Canada, June 1996. View at: Google Scholar
  12. J. A. Czyz and J. Falk, “Use of geopig for prevention of pipeline failures in environmentally sensitive areas,” in Proceedings of the Pipeline Pigging, Integrity Assessment and Repair Conference, Houston, Tex, USA, February 2000. View at: Google Scholar
  13. I. B. Iflefel, D. G. Moffat, and J. Mistry, “The interaction of pressure and bending on a dented pipe,” International Journal of Pressure Vessels and Piping, vol. 82, no. 10, pp. 761–769, 2005. View at: Publisher Site | Google Scholar
  14. J. Błachut and I. B. Iflefel, “Collapse of pipes with plain or gouged dents by bending moment,” International Journal of Pressure Vessels and Piping, vol. 84, no. 9, pp. 560–571, 2007. View at: Publisher Site | Google Scholar
  15. J.-H. Baek, Y.-P. Kim, W.-S. Kim, J.-M. Koo, and C.-S. Seok, “Load bearing capacity of API X65 pipe with dent defect under internal pressure and in-plane bending,” Materials Science and Engineering: A, vol. 540, pp. 70–82, 2012. View at: Publisher Site | Google Scholar
  16. A. Limam, L.-H. Lee, and S. Kyriakides, “On the collapse of dented tubes under combined bending and internal pressure,” International Journal of Mechanical Sciences, vol. 55, no. 1, pp. 1–12, 2012. View at: Publisher Site | Google Scholar
  17. D. K. Kim, S. H. Cho, S. S. Park, H. R. Yoo, and Y. W. Rho, “Design and implementation of 30 geometry PIG,” KSME International Journal, vol. 17, no. 5, pp. 629–636, 2003. View at: Publisher Site | Google Scholar
  18. J. A. Czyz et al., “Multi-pipeline geographical information system based on high accuracy inertial surveys,” in Proceedings of the ASME 3rd International Pipeline Conference, Calgary, Canada, 2000. View at: Google Scholar
  19. J. D. Hart, N. Zulfiqar, D. H. Moore, and G. R. Swank, “‘Digital Pigging’ as a basis for improved pipeline structural integrity evaluations,” in Proceedings of the International Pipeline Conference, vol. 2 of Integrity Management; Poster Session; Student Paper Competition, Calgary, Canada, September 2006. View at: Publisher Site | Google Scholar
  20. J. A. Czyz and J. R. Adams, “Computations of pipeline—bending strains based on geopig measurements,” in Proceedings of the ipeline Pigging and Integrity Monitoring Conference, pp. 14–17, Houston, Tex, USA, February 1994. View at: Google Scholar
  21. J. D. Hart and G. H. Powel, Geometry Monitoring of the Trans-Alaska Pipeline, Trans-Alaska Pipeline System (TAPS), 2005.
  22. D. Yu, L. Zhang, W. Liang, Y. Ye, and Z. Wang, “Noise reduction of signal and condition recognition of long-distance pipeline,” Acta Petrolei Sinica, vol. 30, no. 6, pp. 937–941, 2009. View at: Google Scholar
  23. J. Zhou, B. Gui, M. Li, and W. Lin, “An application of the artificial neural net dominated by lithology to permeability prediction,” Acta Petrolei Sinica, vol. 31, no. 6, pp. 985–988, 2010. View at: Google Scholar
  24. D. Li, D. Lu, X. Kong, and Y. Du, “Processing of well log data based on backpropagation neural network implicit approximation,” Acta Petrolei Sinica, vol. 28, no. 3, pp. 105–108, 2007. View at: Google Scholar
  25. K.-X. Peng and S.-Z. Yu, “Prediction and control of strip thickness based on Bayesian neural networks,” Journal of University of Science and Technology Beijing, vol. 32, no. 2, pp. 256–262, 2010. View at: Google Scholar
  26. H. Yang and H.-Z. Fu, “Stock index forecast based on bayesian regularization BP neural network,” Science Techno Logy and Engineering, vol. 6, pp. 3306–3310, 2009. View at: Google Scholar
  27. H. Wang, L. Shi, H. Zhao, and Y. Yue, “Temperature prediction model for sunlight greenhouse based on bayesian regularization BP neural network,” Hubei Agricultural Sciences, vol. 9, pp. 4300–4303, 2015. View at: Google Scholar
  28. Y. Shenghua and D. Juan, “GDP prediction based on principal component analysis and bayesian regularization BP neural network,” Journal of Hunan University (Social Sciences), vol. 11, pp. 42–45, 2011. View at: Google Scholar
  29. R. Li, Q. Feng, M. Cai et al., “Measurement of long-distance buried pipeline centerline based on multi-sensor data fusion,” Acta Petrolei Sinica, vol. 35, no. 5, pp. 987–992, 2014. View at: Publisher Site | Google Scholar

Copyright © 2016 Qingshan Feng et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

More related articles

 PDF Download Citation Citation
 Download other formatsMore
 Order printed copiesOrder

Related articles