General stochastic algorithm for calculating CO2-related corrosion damage

Anti-Corrosion Methods and Materials

ISSN: 0003-5599

Article publication date: 2 March 2015

Citation

, (2015), "General stochastic algorithm for calculating CO2-related corrosion damage", Anti-Corrosion Methods and Materials, Vol. 62 No. 2. https://doi.org/10.1108/ACMM.12862baa.002

Publisher

:

Emerald Group Publishing Limited


General stochastic algorithm for calculating CO2-related corrosion damage

Article Type: Methods From: Anti-Corrosion Methods and Materials, Volume 62, Issue 2

Corrosion is a distributed phenomenon – meaning there is no such thing as “uniform” attack. Although this simple fact has been known since the early 1900s, modelling this aspect today requires a paradigm shift.

Typically, corrosion rates are estimated based on measurements from corrosion coupons – small pieces of metal – and the depth of gouges left behind on the metal’s exposed surface under specific environmental conditions.

In the November 2014 issue of CORROSION, Raymundo Case, staff scientist for ConocoPhillips, introduces a general stochastic algorithm for calculating the CO2-related distribution of corrosion damage at a given time interval, coupled with corrosion rates predicted by conventional deterministic models:

This model describes the distribution of corrosion damage across a hypothetical surface–made from carbon steel–exposed to an environment in which the presence of CO2 controls the corrosion mechanisms, says Case.

Corrosion of carbon due to the presence of CO2 within produced oilfield fluids has been studied for many years – at least since the early 1960s:

And the consensus is that the various forms of attach depend upon the interaction between the actual metal dissolution rate (kinetics) and the surface interaction with the protective capability of the iron carbonate (FeCO3) scale formed from the oversaturation at the immediate reaction interface, Case explains.

Damage distribution is assessed using “stochastic” or random-valued functions. “The output of these functions was developed from the accumulated knowledge of how CO2 corrosion mechanisms occur”, says Case:

A function was developed for the metal dissolution and others for the FeCO3 scale protective effect, so the model takes the result of the “competition” between both effects and can quantify the difference.

The results of the output can then be evaluated using statistical methods and probability algebra to provide information that describe whether, under the conditions considered, corrosion’s attack is more uniform or pitting can occur:

In each case, the statistical evaluation of the model output allows us to evaluate the mean and maximum corrosion rates expected – providing useful information for establishing risk assessments needed in inspections, he says.

Why is this work so significant? It attempts to rationalize the corrosion damage distribution – and, consequently, the morphology of the corrosion attack – by mathematical methods. The stochastic model description couples the corrosion rate assessment provided by the mechanistic modelling, based on electrode kinetics, to its physical appearance on the metal surface:

Complimentary to this aspect, the evaluation of the stochastic model output enables quantitative establishment of the likelihood of failure–at least by CO2 corrosion, notes Case.

What are the immediate applications for the model? It can be used to assess the likelihood of failure due to CO2 corrosion for down-hole equipment and facilities, as well as enabling production and corrosion engineers working in oilfields to understand the severity of the corrosive environment in terms of what kind of attack they can expect and what will be the risk of failure with service time – being able to evaluate the need of chemical corrosion inhibition treatment or schedule preventive repairs:

From a corrosion science point of view, this model shows a mathematical way to evaluate when localized forms of attack will be present, which can then be compared with actual pit propagation studies, says Case.

Perhaps, most importantly, it demonstrates a new paradigm in corrosion modelling that can be extended in many ways:

These proposed solutions are not by any means definitive, and I expect that in the future other researchers will find more sophisticated solutions applicable to many aspects such as flow-assisted corrosion inhibition, notes Case.

What’s next? “In terms of stochastic modelling, we’re researching how to model hydrogen sulfide/carbon dioxide corrosion – a.k.a. as ‘sour environment’ corrosion of carbon steels”, he explains. “This area is challenging because we need to carefully understand the interaction between both acid gases dissolved in water, produced or otherwise, and the kinetically different scales they produce”.

ConocoPhilips is collaborating with several universities, which are providing basic research and results to be evaluated for incorporation into the stochastic model:

We’re working to extend the principles of stochastic evaluation of corrosion mechanisms to pitting in stainless steels so we can predict the likelihood of stress corrosion cracking under any kind of environment, says Case.

More information is available from: http://www.nace.org