A Novel FR 13 Risk Assessment of Corrosion of Pipeline Steel in De-Aerated Water

Abstract

Steady-state operations are used globally in chemical engineering. Advantages include ease of control and a more uniform product quality (Ghasem and Henda, 2008; McCabe et al, 2001). Importantly however, there will be naturally occurring, random (stochastic) fluctuations in parameter values about the ‘set’ mean when process control is inadequate. These are not addressed explicitly in traditional chemical engineering because they are not sufficient on their own to be considered transient (unsteady) and because, generally, fluctuations in one parameter are off-set by changes in others with plant output behaviour seeming to remain steady (Amundson et al., 1980; Sinnott, 2005; Zou and Davey, 2016). Davey and co-workers, however, have demonstrated these fluctuations can unexpectedly accumulate in one direction and leverage significant (sudden and surprise) change in output behaviour with failure in product or plant (e.g. Abdul-Halim and Davey, 2016; Zou and Davey, 2016; Chandrakash and Davey, 2017). To underscore the unexpected element of the failure event they titled their risk framework Fr 13 2 (Friday 13th Syndrome). Case studies of their probabilistic risk framework to 1-step operations include loss of thermal efficiency in a coal-fired boiler (Davey, 2015 a) and failure to remove whey deposits in Clean-In-Place (CIP) milk processing (Davey et al., 2015). More recently, to advance their risk framework for progressively, multi-step and complex (in the sense of ‘integrated’, not ‘complicated’) processes they demonstrated its usefulness to 2-step membrane fouling with combined ultrafiltration-osmotic distillation (UF-OD) (Zou and Davey, 2016), and; a 3-step microbiological raw milk pasteurization (Chandrakash and Davey, 2017). Findings overall revealed no methodological complications in application - and it was concluded the risk framework was generalizable (Zou and Davey, 2016; Chandrakash and Davey, 2017). A significant advantage of the framework is it can be used in ‘second-tier studies’ to reduce risk through simulations of intervention strategies and re-design of physical plant or operating practice. It can be applied at both synthesis and analysis stages. 2 see Appendix A for a definition of some important terms used in this research. Although the risk framework has been successfully applied to corrosive pitting of AISI 316L metal widely used in off-shore oil and gas structures (Davey et al., 2016) 3 it was not known if it could provide new insight into corrosion of metal, more specifically microbiologically influenced corrosion (MIC), a major problem globally that accounts for ~ 20 % of overall corrosion (Flemming, 1996). It is estimated to cost AUD$7 billion to Australia annually (Javaherdashti and Raman-Singh, 2001). A review of the literature showed that a thorough understanding of MIC has been slow to emerge, both because of the role of micro-organisms in corrosion and because of a lack of methodology to determine any impact of natural fluctuations in the internal pipe environment. Importantly, the insidious nature of MIC was known to pose a practical risk of failure of pipes used to transport wet-fluids. However, because modelling of direct MIC would be uniquely complex it was planned that a general model for corrosion should be synthesized and understood that could be extended. A limited research program was therefore undertaken with the aim to advance the Fr 13 framework and to gain unique insight into how naturally occurring fluctuations in fluid temperature (T) and pH of the internal pipe environment can be transmitted and impact corrosion. A logical and stepwise approach was implemented as a research strategy. The initial model of Smith et al. (2011) was modified to simulate MIC causing micro-organisms such as sulphate-reducing bacteria (SRB) on widely used ASTM A105 carbon-steel pipe that is corroded under steady-state, abiotic and synthetic conditions. This was solved using traditional, deterministic simulations to give a predicted, underlying corrosion rate (CR) of 0.5 mm yr-1 as impacted by internal pipe-fluid T and pH. Importantly, findings underscored the controlling importance of low pH on CR. This initial model was then simulated, for the first time, using the probabilistic Fr 13 framework (Collins et al., 2016) 4 in which distributions to mimic fluctuations in T (K) and 3 This research was a Finalist, IChemE Global Awards 2016, Innovative Product, Manchester, UK, Nov. 4 Collins, S.D., Davey, K.R., Chu, J.Y.G., O’Neill, B.K., 2016. A new quantitative risk assessment of Microbiologically Influenced Corrosion (MIC) of carbon steel pipes used in chemical engineering. In: pH in the pipe were (reasonably) assumed as truncated Normal, and; a new corrosion risk factor (p) was synthesized such that all p > 0 characterized a CR failure i.e. a corrosion rate greater than 0.5 mm yr-1. Normal distributions that were truncated were used because these permitted T and pH to fluctuate randomly during process operations but limited these to values that could occur only practically. Predictions showed that 28.1 % of all corrosion of ASTM A105 pipe, averaged over the long term for a range of fluctuations 290.15 ≤ T ≤ 298.15 K, and 4.64 ≤ pH ≤ 5.67, would in fact be greater than the underlying value despite a design margin of safety (tolerance) of 50 % CR, and were therefore process failures (p > 0). Findings highlighted that corrosion was a combination of ‘successful’ and ‘failed’ operations. This insight is not available from traditional risk approaches, with or without sensitivity analyses. It was concluded that the Fr 13 framework was an advance over the traditional, deterministic methods because all corrosion scenarios that can practically exist are simulated. It was concluded also that if each simulation was (reasonably) thought of as one operational day, there would be (28.1/100 days × 365.25 days / year) ~103 corrosion failures in ASTM A105 pipe per year. However it was acknowledged that to enhance corrosion simulation, the free corrosion potential (Ecorr, V vs SCE), a key parameter in this initial model formulation, should more realistically be considered a combined function of the internal pipe-fluid T and pH, and; that this assumption should be tested, and, that this would necessitate a trial-and-error simulation for corrosion rate (CR). It was also determined that the truncations that were used for T and pH were too restrictive for off-shore oil processing (Arnold and Stewart, 1999; J. Y. G. Chu, Upstream Production Services Pty Ltd., Australia, pers. comm.). To address this, the initial model was extended mathematically for the first time, and; Fr 13 risk simulations carried out using spread-sheeting techniques utilizing the Solver CHEMECA 2016: Chemical Engineering – Regeneration, Recovery and Reinvention, Sept. 25-28, Adelaide, Australia, paper 3386601. ISBN: 9781922107831 function (Microsoft Excel™). A significant advantage was that the distributions defining the naturally occurring fluctuations in T and pH could be entered, viewed, copied, pasted and manipulated as Excel formulae. Predictions showed (Collins and Davey, 2018) 5 an underlying corrosion rate CR = 0.45 mm yr-1 – a change of approximately 10 % when the design margin of safety (tolerance) was reduced from 50 % to a more realistic 20 % for the improved model. This is significant because the tolerance of a model should be as low as can be accepted, as higher tolerances can infer that the process is safer than it actually is. Fr 13 simulations showed that 43.6 % of all corrosion of internal ASTM A105 pipe, averaged over the long term for a range of realistic fluctuations 282.55 ≤ T ≤ 423.75 K, and 4.12 ≤ pH ≤ 6.18 would be deemed to be process pipe-failures (p > 0). This translates to a corrosion failure in ASTM A105 pipe every 160 days, averaged over the long term. It is not expected that these would be equally spaced however. Findings were used in investigative ‘second-tier’ studies to explore possible intervention strategies to reduce vulnerability to corrosion and to improve plant design and safety. For example, repeat Fr 13 simulations revealed that, for a fixed mean-value of T = 353.15 K a decrease in pH from 5.15 to 4.5 resulted in an increase in carbon-steel pipe corrosion of ~1.55 mm yr-1 i.e. ~347 % increase. This implied that the pipe vulnerability to Fr 13 corrosion failure could be practically minimised by adding bases, such as potassium hydroxide or sodium carbonate (Kemmer, 1988). However, if the pH is too high, anions in the pipe-fluid could precipitate and form insoluble mineral scales, leading to fouling (Pichtel, 2016). It is acknowledged that the present research is limited to an abiotic system i.e. one without micro-organism kinetics. A justification is that the models presented in this research should be seen as a ‘starting point’, which could be expanded in later iterations to include: biotic model components such as the simple bacterial kinetics in the predictive MIC model of Maxwell and Campbell (2006); other species that are involved in MIC such 5 Collins, S.D., Davey, K.R., 2018. A novel Fr 13 risk assessment of corrosion of carbon-steel pipe in de-aerated water. Chemical Engineering Science – submitted CES-D-18-00449, Feb. as sulphates, chlorides and hydrogen sulphide (H2S); different metals/alloys that are used in pipe equipment where MIC can be found e.g. copper or zinc (Roberge, 2000), or; a ‘global’ model i.e. two or more connected unit-operations (Chandrakash and Davey, 2017). (A global model however, might not be applicable because MIC can be initiated in localized sites (Roberge, 2000)). It is concluded that these thesis findings nevertheless significantly enhance understanding of factors that lead to excessive corrosion rates in ASTM A105 pipes. It is concluded also that the Fr 13 risk framework appears generalizable to a range of micro-organism-metal systems and is an advancement over current existing risk and hazard assessments. If properly developed, it is thought that this risk technique could be adopted as a new design tool for steady-state unit-operations in both the design and synthesis stages and to increase understanding of MIC behaviour and outcomes. This research is original and not incremental work. Results and findings will be of immediate benefit and interest to a range of risk analysts, and to a broad range of practical operations involving carbon-steel pipe flows