INTRODUCTION
In the framework of a research project aimed at the application of corrosion modelling in field operation, a few cases of internal corrosion observed in pipelines were studied.
While it is reasonably easy to understand a corrosion event 'retrospectively" with failure analysis methods, a large degree of uncertainty is associated with the attempt of quantifying a prediction for future damages.
Corrosion models should be implemented taking into account that conditions do change during time (fluid composition and properties, temperature, pressure ...) and in space (change of flow regime and physical parameters along the line).
When a new pipeline is designed. there is a need for predicting corrosion rate, in order to select a suitable material; however, during the operating life of the component, the corrosion rate is influenced by several parameters, values of which cannot be predicted in the design phase.
Another important goal of predictive models should be to understand how the production conditions affect corrosion rate calculation during the life of a component.
lncreasing severity of the environment for new installations (deep water, permafrost ...)
Increasing aggressiveness of produced fluids (water cut, sulphide content ... )
lncreasing age of plants and components.
The future challenges for users of metallic structural materials in the Oil&Gas business are mainly due to:It is moreover necessary to take into account the strategic importance of acting as operator of plants, and the new aptitude of regulations - especially in Europe and US - toward the responsibility of users in Health, Safety and Environmental issues. For transmission pipelines. involved costs are important and a correct decision making can result in important savings. The impact of corrosion allowance for instance can be estimated as 24-28 k$ per km of gas transmission pipe1. Corrosion allowance is typically estimated using corrosion rate models. However, the corrosion rate estimate obtained using different models can be different from the measured value2.
Predictive models are mainly based on empirical correlations with laboratory or field data. Some other models are based on a strict mechanistic analysis of the various processes involved in CO2 corrosion of carbon steel. Nevertheless, CO2corrosion is still to be considered an open issue since models can produce different estimates, markedly reflecting the different philosophies used in their development.
Most models are primarily CO2corrosion predictive models and are unsuitable for situations with appreciable H2S contamination, the last representing the actual well situation. The number of parameters involved in the Corrosion Rate (CR) calculation is quite different for the various models. All of them require - as input - temperature, pressure and CO2content, while some of them require gas, oil and water production rates and others ask only for liquid flow velocity 2.
Models can be conservative and non conservative. De Waard and Norsok3,4,5 for instance are conservative models. They usually predict higher corrosion rates (sometimes also considerably higher corrosion rates) than found in the field.
