Transient analysis helps IM for crater-type corrosion defects

Nov. 4, 2013
Introducing transient analysis to pipeline integrity management (IM) is strongly recommended when crater-type corrosion defects are present and when threshold criteria are more rigorous than usual.

Abdel-Fettah Touabti
University Ferhat Abbas
Setif, Algeria

Karim Younsi Abdelhak Bentriou
University of Boumerdes
Algeria

Abdelnacer Smati
Pegaz Engineering
Algiers

Introducing transient analysis to pipeline integrity management (IM) is strongly recommended when crater-type corrosion defects are present and when threshold criteria are more rigorous than usual. Applying transient analysis to gas pipeline integrity management does not significantly increase the number of repairs, only the manner in which they are planned for, and therefore does not result in prohibitive additional costs.

Transient analysis, however, is not justified in integrity management of gas pipelines when all corrosion points are of pinhole type.

Structural-reliability model

Pipeline integrity management programs focus mainly on using inline inspection (ILI) tools, with significant progress in this field allowing detection of nearly all corrosion defects. ILI inspections typically display results as graphics showing the geodetic positions of the main corrosion defects and their dimensions. A number of standards estimate the failure stress of pipelines according to the geometrical dimensions of critical defects and guide repairs needed to return a particular line to its original maximum allowable operating pressure (MAOP).

Corrosion estimation

Comparing the size of defects in successive ILI inspection runs allows estimation of the failure stress of a corroded pipeline by monitoring corrosion rate.1 A minimum of two sequential inspections are required to gauge corrosion rate, with repair priorities assuming a defect-by-defect linear evolution of corrosion depth (Equation 1 in accompanying equations box).

Rigorous monitoring of corrosion rate requires frequent ILI inspections. But this is an expensive proposition for large pipeline systems, and ILI tool programming must take limits on available resources into account. Operators of a pipeline that has only been inspected once must estimate corrosion growth rates heuristically or by using Bayesian approaches.2

Failure-stress assessment

A deterministic failure model generally uses recommendations of the various allowed standards to estimate failure stress according to geometrical dimensions of the defect using Equations 3 and 43-4 and to plan the repair of corrosion defects using Equation 5. Differences in the recommendations of the various standards lie in their expression of the Folias factor, which modified standard ASME B31G calculates differently according to whether the corrosion defect type is pinhole or crater (Fig. 1).

This approach, however, involves uncertainties. Despite significant progresses in ILI-tool design, uncertainty on measures of defect depths remains at about 10% of WT for a confidence level of 80%, suggesting a potential risk of over- or underestimating failure stress. Specified mechanical properties of the steel grade correspond to minimal requirements of acceptance or refusal of the pipe during manufacture, resulting in yield strength generally much higher than specified values and implying that pipe resistance is generally underestimated.

Corrosion is a complex process and its evolution, from one corrosion defect to another, is not uniform, depending on many factors including coating state, soil aggressiveness, and cathodic protection efficiency. Operating pressure varies depending on scheduled flow rates and particularly overpressures generated in transient and surge situations (Fig. 2).

It is therefore necessary to implement an approach allowing for:

    • Uncertainties in measurement generated by the inspection tool, by the mechanical properties of steel, and by the corrosion growth rate.

    • Quantitative estimation of the failure probability associated with a given corrosion state established on results of the last inspection.

    • Evaluation of the remaining probability of failure after repair of critical defects.

    • Influence of overpressures caused by transient and surge situations.

Transient flow

During project phase, transient and surge analyses check the resistance of the pipeline. The new pipe is considered devoid of corrosion defects. Transient and surge situations occur when fluid flow velocity changes abruptly because of a valve closure or other status change in a control component, causing a pressure wave that moves from one point to another at the sound speed inside the fluid. The wave therefore potentially subjects the pipe to pressure values exceeding the limits of pipeline resistance (Fig. 3).

Many leaks and explosions in Niger delta pipelines are attributable to a conjunction of corrosion state surge-generated overpressures.5 This article, however, is restricted to presenting a procedure defining the maximum steady-state operating pressure needed to ensure excessive transient pressures don't exceed the resistance limits of corroded pipe, using DNV-F101.

As part of preliminary planning for a project to increase flow rate of a pipeline in Brazil, engineers proposed analyzing the integrity of the system using the transient simulation software STONER and a deterministic model to estimate failure stress.6 Analyzing the transient regimes in pipelines mathematically requires solving a system of partial differential equations subjected to initial and boundary conditions characterizing the studied case.

The mathematical difficulties inherent in the resolution of such systems prompted development of powerful dynamic simulators, which can deal with all the possible configurations. This article uses SIMONE 5.66 dynamic simulation software for gas pipelines networks.

Probabilistic analysis

Variability in the corrosion growth rate and uncertainties regarding resistance limits, tool accuracy, and pipeline geometry, have prompted some to use probabilistic approaches instead of deterministic approaches.7 Since evolution of the degradation processes over time is uncertain, these can best be represented by stochastic processes (OGJ, Oct. 1, 2012, p. 122; Nov. 5, 2012, p. 132).

Probabilistic structural analysis forms a mathematical model through which it is possible to calculate the probability that a structure is found in a specified state, knowing both that one or more of its properties are random and that loads on the structure are also random. Parity between the load and resistance defines the limit state. When load becomes greater than resistance, failure occurs.

In the case of corroded pipeline the dimensions of metal loss are load conditions and the allowable dimensions of the defect resistance conditions.8-9 Failure occurs when corrosion depth, x, reaches critical depth, xcr. Failure probability corresponds to the surface represented by the shaded area in Fig. 4.

The dynamic nature of degradation processes therefore makes the load a function of time, τ, tending to increase when resistance is constant (Equation 6).

Resistance-curve modeling

If the main dimension characterizing the default risk is represented by the depth of corrosion, then rearranging Equation 3 yields the critical depth of corrosion expressed in Equation 7. Considering a normal distribution with known average and variance, yield strength's uncertainty in Equation 4 is described by determining momentums µXcr and σXcr of Xcr distribution on the basis of Monte Carlo simulation, yielding the resistance curve expressed by Equation 8.

Load curves

Measuring evolution of metal-loss depth guides estimation of corrosion growth rates. In a corroded pipeline, each pipe element can contain hundreds or even thousands of corrosion points of various sizes. Assuming normal distribution and the requirement of a minimum of two inspections, classical statistical treatment can generate corrosion growth rate distributions. Equation 2 calculates the average, µv, while Equation 9 estimates standard deviation.

Assuming a linear process of corrosion, Equations 10-12 describe the evolution of the load curve over time.

The equalization of load distributions and resistance defines the intersection coordinates of Z(τ) (Fig. 5). Calculation of failure probability uses Equation 6, with the surface of failure between the resistance curve and the intersection point obtained by integrating the curve of resistance (Equation 13). Integrating the load curve obtains the part of the surface of failure ranging between the intersection point and the load curve (Equation 14).

Adding the two surfaces determines the probability of failure (Equation 15).

Failure probability

A repair program follows ILI inspection. Simulating failure probability before and after repair and comparing results to a threshold criterion adopted by the operator can provide for optimal planning of preventive repairs. The threshold value used in worldwide practice refers to failure probability/kilometer.4 In a corroded pipeline, each kilometer can contain several corrosion defects of various dimensions (Fig. 6).

Following reliability theory and assuming that corrosion defects are independent elements installed in series,10-11 the probability of failure at the moment, τ, of the pipeline section corresponding to the Jth km is expressed by Equation 16.

The repair program can use the estimated failure probability as its basis, determined in Equation 16 by assigning a value of zero probability for each corrosion defect to be repaired. This approach allows hierarchical optimal planning of repairs to maintain a failure probability below risk tolerance, followed by scheduling inspection with intelligent tools.

It follows from the assumption that corrosion defects are independent elements that the failure probability of a section of pipeline must be determined starting from the probabilities of failure of all the corrosion defects present on this section.

A section of pipeline with a great number of defects of average size could prove, in certain cases, more dangerous than a section exhibiting a reduced number of points of corrosion of more significant size. This perspective is cut off from view by interposed matter in practice when analysis is performed defect by defect.

A section of Algerian gas pipeline between two compressor stations (Fig. 7), with parameters shown in Table 1, was inspected twice (Table 2). Simulating the closing of the successive valves in 60-sec intervals (Fig. 8) defined the envelope of maximum transient pressure. The model described in this work allowed computation of the failure probability per corrosion defect along the pipeline (Fig. 9) and the failure probability/kilometer before repair (Fig. 10).

Applying Equation 16 and the 10-3 threshold criterion allowed identification of the corrosion defects to be repaired and computation of the failure probability/kilometer after repairs (Fig. 11). Only the failure probability of crater-type corrosion defects differs significantly between using steady and transient approaches (Fig. 12), with the failure probability of pinhole defects remaining practically the same (Fig. 13).

References

1. Worthingham, R.G., Morrison, T.B., and Desjardins, G.J., "Analysis of corrosion rates on a gas transmission pipeline," NACE Corrosion2002, Denver, Apr. 7-11, 2002.

2. Younsi, K., and Smati, A., "Modèle bayesien d'estimation du risque de défaillance sur un pipeline corrode," (French only), Multidisciplinary International Conference, Bordeaux, Mar. 16-17, 2005.

3. ASME B 31G, "Manual for determining the remaining strength of corroded pipeline. A supplement to ANSI/ASME B 31G code for pressure piping," 1991.

4. DNV, "Corroded pipelines recommended practice," Det Norske Veritas, RP-F101, 1999.

5. Achebo, J.I., "Mathematical Analysis of Pressure Surge Problems in Corroded Pipelines," International Journal of Engineering Studies, Vol. 1, No. 1 (January 2009), pp. 1-5.

6. Nicoletti, E.S.M., Geraldo de Souza, A., and Tolmasquim, S.T., "Maximize Flow Capacity on Corroded Liquid Pipelines: A Massive Defect Assessment Methodology Based Upon Hydraulic Simulations and ILI Results," International Pipeline Conference, Calgary, Sept. 28-Oct. 3, 2008.

7. Alamilla, J.L., and Sosa, E., "Stochastic modelling of corrosion damage propagation in active sites from field inspection data," Corrosion Science, Vol. 50, No. 7 (July 2008), pp. 1811-1819.

8. Dawson, J., Race, J., Peet, S., and Krishnamurthy, R., "Pipeline corrosion management," NACE Corrosion2001, Houston, Mar. 11-16, 2001.

9. Younsi, K., Chebouba, A., Zemmour, N., and Smati, A. "Pipeline integrity assessment using probabilistic transformation method and corrosion growth modeling through gamma distribution," Oil and Gas Facilities, Vol. 2, No. 2, April 2013, pp. 51-60.

10. Lecchi, M., "Evaluation of predictive assessment reliability on corroded transmission pipelines," Journal of Natural Gas Science and Engineering, Vol. 3, No. 5 (October 2011), pp. 633-641.

11. Lee, O.S., and Pyun, J.S., "Failure probability of corrosion pipeline with varying boundary condition," KSME International Journal, Vol. 16, No. 7 (July 2002), pp. 889-895.

The authors

Abdel-Fettah Touabti ([email protected]) is a lecturer-researcher at the University Ferhat Abbas, Sétif, Algeria since 1994. He holds a BS in engineering (1980) from University of Boumerdes and an MS in gas engineering and reliability (1992) from Mining School of Paris, France.

Karim Younsi ([email protected]) is a lecturer-researcher in the oil and gas faculty at University of Boumerdes, Algeria. He holds a BS in engineering (1997) and an MS in reliability (2002) from the Algerian Faculty of Hydrocarbons and Chemistry, University of Boumerdes.

Abdelhak Bentriou ([email protected]) is a senior lecturer in the chemistry and hydrocarbons faculty, University of Boumerdes. He holds a PhD in processing from the former Gas, Petroleum, and Chemistry Institute of Moscow (Goubkine), now known as Petroleum and Gas University of Moscow.

Abdelnacer Smati ([email protected]) is a technical manager for Pegaz Engineering. Before joining Pegaz, he taught for 22 years in the department of transport and equipment for the hydrocarbon faculty of Boumerdes University. He has a PhD. in mechanical engineering, with a specialization in petroleum engineering (1986)from Moscow University of Petroleum and Gas