Assessment and comparison of CO2 corrosion predicition models

Abstract

A number of different CO2 corrosion rate prediction models have been developed for application in oil and gas production systems and are readily available in the literature or as industry standards. These models vary from the empirical models to the more recently developed mechanistic models. In this paper a number of CO2 corrosion rate prediction models are compared. In addition to the obvious criteria of comparing the predicted corrosion rates, the models were also compared for other characteristics such as ease of implementation and insight into the main drivers of the predicted corrosion rate. Using as input the water-chemistry of condensed water, the output from each of the CO2 corrosion prediction models was compared over a range of temperatures and pressures. A common source dataset was used as input for each of the models. For the range of input values, the suite of selected models generated results which were comparable for lower temperatures and pressures. At higher temperatures and pressures where the effect of the precipitation of iron carbonate would be expected to be influential then the outputs from the models differed. This was in part because in some cases iron carbonate precipitation was not included, but in other cases this difference was due to the alternative implementations of the effect of iron carbonate precipitation. As would be expected the mechanistic models are the more complex to implement with some being such that specialist computer code is required to numerically solve the systems of equations. However, for all the complexity of implementation, the mechanistic models have the particular strength of providing greater insight into the critical variables driving the overall corrosion mechanism and hence suggest strategies for minimizing the CO2 corrosion risk during design and operation and are therefore to be recommended. Introduction The corrosion of oil and gas equipment due to the presence of CO2 in the production fluids has been the subject of extensive study. A number of CO2 predictive models have been developed in the literature with deferring levels of complexity and different theoretical bases leading to potentially inconsistent assessment of the CO2 corrosion risks. Moreover, the basis and applicability limits of these prediction models are frequently misunderstood. This paper addresses these issues by comparing some commonly used prediction models under a common dataset and providing insight about the strengths and weaknesses of each.

Understanding Passive Layer Formation for Further Corrosion Management in Gas Production Pipes

“…Using the input of corrosion rate and environmental condition, the proposed model can be executed. It can be recognized that there are three boundaries, so that there are three regions [52]: no-scale, mackinawite and siderite region. The arrangement of the boundaries does not change at various conditions, only the value of the boundaries will be different at different conditions.…”

“…The model is used to predict the type of passive layer in gas pipes through hydrogen sulfide partial pressure vs carbon dioxide partial pressure in log-log plot [52]. The execution step for the model follows the algorithm in Figure 1.…”

Deterministic Prediction of Localized Corrosion Damage in Oil and Gas Pipelines

“…where j k and R k are the flux density and the bulk source (sink) of the "particles" k, respectively. The system of Equations [11] can be solved by using any standard numerical method with the appropriate boundary and initial conditions:…”

Empirical equation of CO2corrosion with presence of low concentrations of acetic acid under turbulent flow conditions

“…The predicted data are used as the basis of material selection in the design stage and then provide inputs for maintenance and inspection strategies in the operation stage. On the other hand, most of CO 2 corrosion predictive models cost are quite expensive and several publicly available CO 2 corrosion prediction models such as Norsok, 12 Cassandra 13 and de Waard et al [14][15][16][17] do not deliberate on HAc effect in the corrosivity analysis although HAc is known as a hazardous species that gives unreliable corrosion as highlighted by Fatah et al 18 and Woollam et al 19 Hence, accurate prediction data are important in the sense that the correct material selection and corrosion mitigation method which is crucial to withstand the design live can be implemented. This is not only relevant to safety issues but also to cost, by avoiding both under and over designs.…”

“…The comparison of the predicted corrosion rates based on the above prediction equation [using Cassandra (DWM 93/95) as the CRb] and commercial model are shown in Figs [18][19][20]…”