scope of overall low level device increased inspection increased inspection increased inspection
project and and flow monitor frequency frequency frequency
available from in injection
Day 1 package
Category 5 > 95% >6 mm/yr. 300 ppm Should be within Should include Early inspection As Category 4 As Category 4 As Category 4
scope of overall low level device plus leak plus leak
project and and flow monitor detection detection
available from in injection
Day 1 package
Corrosion Inhibitor Corrosion Assumed System System Scheduling 1st On land, On land, Subsea
- The aim of the inhibitor availability model is to encompass the good track record of the inhibitor efficiency model at low to moderate corrosivities but to remove some of its conservatism in more corrosive systems. The two inputs to the model are the inhibited corrosion rate and the inhibitor availability and using different values for these can produce a whole array of outputs.
Figure 18 shows the corrosion allowance that would be recommended using the two approaches for a 20 year design life. A range of uninhibited corrosion rates are considered, from 0.5 to 10 mm/yr. which covers the range from mildly to highly corrosive fluids (less corrosive fluids would probably be handled without recourse to inhibition). In the inhibitor efficiency example, an efficiency of 90% has been assumed, in line with BP’s previous practice. The inhibitor availability model uses an inhibited corrosion rate of 0.1 mm/yr. and an inhibitor availability of 95%. During the remaining 5% of the time, the uninhibited corrosion rate is used (0.5 to 10.0 mm/yr. as appropriate).
Both models agree well for moderately corrosive fluids, while for mildly corrosive fluids (0.5 to 1.0 mm/yr.) the availability approach recommends a greater corrosion allowance. In practice, this may not be important as external corrosion may require a corrosion allowance of up to 2 mm and would over-ride the allowance recommended for internal corrosion.
Comparisons of the Inhibitor Availability Model with BP’s Previous Model 0.5 1 2 3 5 10 1.0 2.4 2.0 2.9 4.0 3.9 6.0 4.9 10.0 6.9 20.0 11.9 0 2 4 6 8 10 12 14 16 18 20
Recommended Corrosion Allowance
for 20 year design life - mm
0.5 1 2 3 5 10
Predicted Corrosion Rate - mm/yr.
Corrosion allowance - efficiency method Corrosion allowance - availability method
Inhibitor availability model based on inhibited rate of 0.1 mm/yr
and availability of 95%
Efficiency method based on efficiency of 90% Figure 18: A
Comparison Between the Inhibitor Efficiency and Inhibitor
Availability Methods of Determining
For highly corrosive fluids, the availability model recommends lower corrosion allowances than the efficiency model. This agrees well with the observed ‘high efficiencies’ of corrosion inhibitor under highly corrosive conditions. This will increase the use of carbon steel as the standard practice is to specify carbon steel with corrosion allowances up to 8mm and to use corrosion resistant steels for more corrosive fluids.
Figure 19 shows the relationship between predicted corrosion rate and the recommended corrosion allowance using the inhibitor availability method. The example shown is the same as in Figure 18 with predicted corrosion rates in the range 0.5 to 10 mm/yr. In each case, the corrosion allowance for inhibited corrosion is constant at 1.9 mm due to the assumption of an inhibited corrosion rate of 0.1 mm/yr. and the required field life of 20 years. The variation in recommended corrosion allowances is due entirely to the 5% of the time where inhibition is assumed to not occur.
Figure 19 helps to illustrate how important the period of uninhibited corrosion can be. In a severe case of a predicted corrosion rate of 10 mm/yr., the uninhibited period of 5% of the time accounts for 83% of the corrosion allowance. In this case, each 1% increase in the assumed availability of corrosion would reduce the total corrosion allowance by 16.6%. Table 13 gives some more details on this point.
1.9 1.9 1.9 1.9 1.9 1.9 0.5 1 2 3 5 10 0 2 4 6 8 10 12 0.5 1 2 3 5 10
Predicted Corrosion Rate - mm/yr
Recommended Corrosion
Allowance for 20 Year design
life - mm
Corrosion allowance for uninhibited corrosion Corrosion allowance for inhibited corrosion (95% availability)
Figure 19: The Contribution to the Total Recommended Corrosion Allowance from the Inhibited and Uninhibited Portions of the Inhibitor Availability Model
0.5 2.4 2.3 3.3 % 1 2.9 2.7 6.2 % 2 3.9 3.5 9.7 % 3 4.9 4.3 11.8 % 5 6.9 5.9 14.2 % 10 11.9 9.9 16.6 %
Table 13: The Effect of the Assumed Corrosion Inhibitor Availability on the Recommended Corrosion Allowance for a 20 year Design Life
Predicted CA assuming CA assuming % reduction in
Corrosion 95% inhibitor 96% inhibitor corrosion allowance
Rate availability availability per 1% increase in
mm/yr.. mm mm inhibitor availability
It can be seen that highly corrosive systems must assume a high value for the inhibitor availability if carbon steel is to be used with a practical corrosion allowance.
The corrosion rate prediction model presented here is for use with carbon steels, i.e. predominantly iron with low levels of carbon. However, some engineering materials contain a wider range of alloying elements such as chromium and nickel to improve the mechanical properties, such as strength or toughness. Such elements are commonly found in corrosion resistant materials and chromium in particular can increase the corrosion resistance of carbon steels, if present in sufficient concentration. 13% of chromium turns a carbon steel into a stainless steel, with excellent resistance to CO2 corrosion.
Many claims have been made over the past 5 years of the affect of adding low levels of chromium (0.5 to 1.0%) to carbon steel. Some steel suppliers claim that 0.5%Cr can halve the CO2 corrosion rate and certainly in some tests there does appear to be a benefit. The most consistent benefit seems to be an improved resistance to ‘mesa’ corrosion where large, square edged and flat bottomed pits can form. However, in other tests no benefits have been observed and it seems that the benefits may be related to m i c ro s t r u c t u re rather than composition. Other re s e a rchers and oil companies have reported that inhibitors perform worse on low alloy steels than on carbon steel and therefore, in inhibited systems, there is no benefit from the addition of low levels of chromium.
On balance, BP believe there are no proven advantages or disadvantages in terms of CO2 corrosion resistance from the presence of chromium at concentrations up to 1% in steels. It is therefore recommended that no account is taken of the presence of alloying elements at low levels and no premium should be paid for such steels. However, if the steel supplier uses low levels of chromium in the standard product, that is acceptable.
Preferential weld corrosion is a problem in most systems and production systems containing CO2 are no exception. Efforts have been made to eliminate preferential weld corrosion by alloying welding consumables with various elements such as chromium, nickel and copper at low levels (circa 1%). No universal solution has been found and there are examples of either weld metal or heat affected zone (HAZ) suffering preferential attack with most welding consumables and welding procedures. The problem is not made easier by the fact that the mechanism for preferential weld corrosion is not fully understood in CO2 service. The speed of such corrosion suggests there could be a galvanic driving force.
Even in ‘benign’ systems where predicted rates of general corrosion are low, rates of attack at welds can be unacceptably high. This causes a problem when deciding whether a corrosion inhibitor is required for a particular application. The traditional approach has been to calculate cumulative wall losses over the life of the field using corrosion models and if the predicted wall loss is less than the available corrosion allowance, inhibitors have not been specified. However, preferential weld corrosion can proceed at rates far higher than predicted and inhibitors offer the only proven method of improving the reliability of carbon steel in such cases. There have recently been cases of preferential weld corrosion causing rapid failures in systems believed to be only mildly corrosive.
Unfortunately, there can be no clear guidance for such systems but inspection programmes should recognise the risk of preferential weld attack and, if detected, corrosion inhibition should be initiated immediately.
CO2 models are basically ‘bare surface’ models with moderation factors applied to anything that affects this, such as surface scales and corrosion inhibitors. Moderation factors are used to reduce the predicted corrosion rate due to the presence of protective or semi-protective species at the surface. In other words, all such factors predict that the surface will corrode at a lower rate than would be expected if it was fully exposed to the bulk solution. Pits are one case where local corrosion rates may be higher than if the surface was exposed to the bulk solution. The environment at a corroding steel surface is different from that in the bulk due to the continual transport of reactants to the surface and products from the surface and this is reflected in the CO2models and associated factors. These effects are generally beneficial where the corrosion process is transport controlled but can be detrimental where it is the transport of inhibitor that is limited. This can be the case in a corrosion pit where galvanic affects also play an important role. The result is that the growth rate of deep pits may accelerate. This can be seen as a loss of control by the inhibitor and may place a practical limit on the size of the corrosion allowance. For example, if an inhibitor is incapable of protecting pits deeper than 8mm, once pitting has reached this depth the corrosion rate in the pit will proceed at the uninhibited rate, i.e. 10 or 20 times faster than the bare surface rate. The increase in life due to the provision of corrosion allowance beyond 8 mm would therefore be minor. In practice, the relationship between pit depth and inhibitor efficiency is not fully understood. Field experience indicates that pits below 5 mm behave normally while pits deeper than this may corrode at a higher rate. Pitting rates up to 3 times faster than predicted have been quoted in a variety of systems. Certainly, if corrosion has reached 8 mm it is likely that the local environment within a pit will be significantly divorced from the bulk environment and hence transportation of inhibitor may be unreliable. Moreover, if corrosion has caused such metal loss, the corrosion control of the system must be poor and providing extra steel is unlikely to provide a satisfactory answer.
As corrosion allowance is often consumed via pitting or localised corrosion the importance of pits should be considered when selecting the optimum corrosion allowance.
The term corrosion allowance creates the impression of a uniform wastage over time leading to the gradual and controlled reduction in wall thickness. In practice, this is unlikely to be the case and the role of the corrosion allowance is to provide protection against the periods when corrosion control is poor and short term corrosion rates are high, i.e. poor inhibitor availability in the case of inhibited systems. As there is always uncertainty in the rate of corrosion (and therefore time to failure), specifying a corrosion allowance is a compromise between capital costs and reliability. Greater corrosion allowances incur greater costs but confer greater reliability. For mildly corrosive systems, low corrosion allowances of 1.5 to 3 mm are justified as they are protecting against the possibility of internal and external corrosion. In highly corrosive systems, active corrosion is almost certain to occur and therefore greater corrosion allowances should be specified to increase the mean time to failure.
Some Operators specify maximum corrosion allowances and BP has tended to use the figure of 8 mm for some years. The reasons for this are:
1. Corrosion tends to be localised pitting attack and corrosion inhibitors perform poorly in deep pits. Therefore, extra corrosion allowance provides little benefit beyond approximately 8mm.
2. Carbon steel will not provide a long term solution for highly corrosive systems and if several millimetres of corrosion allowance have been lost, corrosion control of the system has not been achieved.
3. Intelligent pigs are sensitive to corrosion damage of circa 10% of wall thickness. This makes it difficult to detect the onset of corrosion in thick walled pipe which in turn means that corrosion may continue for some time before detection. It is preferable to detect corrosion early and remedy the situation and therefore thin walled pipe is preferable for detection of corrosion.
4. Welding and handling thick walled pipe is difficult and thick sections may require post weld heat treatment. Cost increases are therefore greater than the incremental increase in wall thickness.
The figure of 8mm should not be seen as fixed. Each project may have different drivers in terms of the optimum balance between opex and capex costs and in certain cases, replacement of flowlines may be more
economically attractive than high capital costs in Year 1. For one recent BPX project it was decided that localised corrosion was the main concern for the flowlines and therefore the definition of corrosion allowance should reflect this. BP’s first pass defect assessment criterion for pipelines allows 20% of the pressure containing wall to be lost due to localised corrosion and the design of the corrosion allowance took this into account. This approach reduced the corrosion allowance by circa 1.5 mm and saved US$1.16 million from the cost of the flowline network. In effect, the ‘traditional’ corrosion allowance was reduced from 8 mm to 6.5 mm but as the corrosion was expected to be localised, there would be 8mm of pipewall available for localised corrosion before raising any concern over integrity.
In other cases, a corrosion allowance greater than 8mm may be justified but it should be recognised that the additional costs may not be reflected in the incremental increase in reliability.
Use of Common Sense
In specifying a corrosion allowance, the Materials Engineer should not be too pedantic. Projects often define three or more nominal corrosion allowances such as 1.5 mm, 3 mm and 8 mm. Process streams are categorised as mildly corrosive, corrosive or highly corrosive using models or experience and the appropriate corrosion allowance added to the pressure containing wall thickness defined using the appropriate code. The total required wall thickness is then reviewed against the available wall thicknesses with the next greater thickness being selected. It may be the case that the corrosion allowance just takes the total wall thickness out of one wall thickness range and into another, increasing significantly the wall thickness and the effective corrosion allowance.
Example
The linepipe specification API 5L lists wall thicknesses (WT) in 1.6 mm increments for 16” linepipe in the range 12.7mm to 14.3mm. If the total required WT including 6 mm corrosion allowance is 12.8mm, standard practice would be to select the 14.3 mm size. The ‘excess’ 1.5 mm would add circaUS$11,500/km to the cost of the 16” flowline i.e. in excess of US$1 million for a 100km line. As the selection of the nominal corrosion
allowance is based on imprecise models, the Materials and Pipeline Engineers should use their judgement in the selection of the final wall thickness. They may decide that a corrosion allowance of 5.9 mm is acceptable, allowing the 12.7 mm WT linepipe to be specified.
CO2 predictive models - such as the one in this report - are based on laboratory studies, typically developed in water only systems. Various moderation factors have been applied over the years, reducing the predicted rates as experience showed them to be too conservative in their basic forms. In the approach covered here, the water cut is ignored thereby treating the pipeline or process equipment as if it was transporting 100% water. It may appear a large step to apply a model developed using laboratory data in water only systems to the field where hydrocarbons account for the majority of the throughput.
However, this is not the vast over-simplification it may seem. Water wetting of the pipewall can occur at both high and low water cuts. This is despite the widely shown plot, reproduced in Figure 21 in which a relationship is proposed between water cut and corrosion rate based on water wetting.
Applying Models to Different Flow Regimes
Effect of Water Cut
Water only... Gas / Water Oil / Water Multiphase 0.1 - 13 m/s 20 - 90oC 0.3 - 20 bara CO2 (0.1 m/s, 90oC, >6.5 bara CO2 excluded!!) Figure 20: The Application of a Model Developed in Water- Only Systems to Other Water-Containing Systems
the average corrosion rate in a system is rarely of interest: it is the maximum rate that determines time to failure. If at a water cut of 1%, 1% of the equipment is water wet 100% of the time then clearly there will be no effect of water cut on the maximum potential corrosion rate and hence time to failure.
Hilly terrain, changes in elevation or changes in flow direction can induce water hold-up in wells, flowlines and process equipment. Local water cuts can exceed 50% despite input water cuts of 1% or less. The water in dips may remain for weeks or months until an increase in throughput sweeps some of it out and a temporary increase in water production is seen at the outlet of the system. It is therefore unwise to rely on the formation of emulsions or similar dispersions to provide fully oil wet surfaces. It is for this reason that BP ignores the water cut in determining system corrosivity.
CO2corrosion rates are dependent on flow regime and flow velocity, hence the attempt to incorporate the effects of flow into the 1995 de Waard and Milliams model. In uninhibited corrosion, flow effects are of secondary importance, after the important controlling factors such as temperature, pressure, CO2 concentration and pH and for this reason BP have retained the earlier de Waard and Milliams model as the basis for their CO2 modelling. The 1995 model is included if the sensitivity to flow velocity changes are considered important.
0 1 0 100 Water Cut, % Corrosion Rate
I
II
III
Figure 21: An Often Presented Relationship Between Water Cut and Corrosion RateEffect of Flow Regime
Each flow regime will cause different rates of corrosion under otherwise identical conditions and the 1995 de Waard and Milliams model offers the best method of assessing this.
When considering inhibited corrosion rates under multiphase flow, the approach proposed on pp76 should be followed. In summary, velocities corresponding to C factors below 100 require no special consideration. Velocities corresponding to C factors between 100 and 135 raise the Category of the corrosion risk, e.g. from 3 to 4. Velocities corresponding to C factors greater than 135 should not be considered unless there is significant operating experience to justify this.
Liquid Flowrate
Bubble
Stratified Stratified Wavy
Slug Annular Gas Flowrate Figure 22: Different Flow Regimes Experienced at Various Combinations of Gas Flowrate and Liquid Flowrate
Crude oil transport pipelines or main oil lines (MOL) fall into two categories: 1. The fully stabilised type such as the Trans Alaskan Pipeline System
and OCENSA in Colombia.
2. The partially stabilised type, such as Forties and Beatrice MOLs. The corrosivity of the fluids is different in each case and pipelines should be designed and operated accordingly.
In the case of fully stabilised lines, the crude oil is processed down to