Beruflich Dokumente
Kultur Dokumente
Abstract
This paper describes an integrated approach, based on corrosion modeling and laboratory testing, to
optimize the use of carbon steel in corrosive service for applications such as downhole tubulars, pipelines, and
facilities. This approach presents economic advantages, such as reducing the use of expensive corrosion resistant
alloys, while ensuring the operational integrity of equipment and facilities. A key part of this integrated approach is
to apply reliable corrosion models underpinned with laboratory data. To be most effective, the models should
account for the relevant chemistry and physics of the corrosion process, including the effects of detailed water
chemistry, liquid hydrocarbons, and the degree of protection from iron carbonate or iron sulfide scales. Ideally,
models should account for variations in conditions and flow characteristics along the length of a wellbore or
pipeline. Case studies are presented that demonstrate how corrosion modeling in conjunction with laboratory
testing may be used to the selection of validate carbon steel for challenging applications.
Introduction
Although high cost corrosion resistance alloys (CRAs) were developed to resist internal corrosion, carbon
steel is still the most cost effective material used in oil and gas production. It is very important to develop an
integrated corrosion prediction approach for optimizing the use of carbon steel in corrosive service while ensuring
the operational integrity of equipment and facilities.
Both corrosion models and laboratory testing are frequently used in this industry to make lifetime predictions of
facilities using carbon steel and further to make decisions on materials selection. Corrosion models, including
empirical, semi-empirical, and mechanistic ones, have been developed over the past several decades to predict
corrosion of carbon steel1-6. These corrosion models can provide engineers quick and economical corrosion
predictions. Most of the models were validated by laboratory data and/or field data. Empirical and semi-empirical
models usually provide reasonable predictions inside of their validation range but poor predictions outside of their
range. Mechanistic models generally can extrapolate to conditions outside of their validation range and remain
accurate to a certain degree. Consequently, one should always understand the validation range and limitations of
the models to apply these correctly. Moreover, although part of corrosion mechanisms are well understood in lab
investigations, due to the complexity in production operations, it is still challenging to apply lab short-term testing
results and corrosion models to predict corrosion of facilities for twenty to thirty years of service. An integrated
approach was therefore developed by ExxonMobil1 to apply a reliable corrosion model in conjunction with
laboratory testing for predicting corrosion in oil and gas production.
ExxonMobil herein refers to capabilities within ExxonMobil Development Co., ExxonMobil Upstream Reseach Co.
and ExxonMobil Production Co.
IPTC 13785
This paper describes ExxonMobil's corrosion prediction model and the application of this model as part of an
integrated corrosion prediction approach.
Model Overview
ExxonMobil's corrosion prediction model is one of the key components of ExxonMobil's integrated
approach for corrosion prediction. This integrated approach enables the optimal use of carbon steel in corrosive
service for downhole, pipeline, and facilities applications. This integrated approach is increasingly important, due
to increasing production from corrosive resources, such as highly sour gas fields, and the need to effectively
manage the integrity of existing resources. The software implementation of ExxonMobil's corrosion prediction
model, CorrCast, provides a window based, user-friendly tool for performing corrosion rate predictions in sweet
(CO2) and sour (H2S) services. Furthermore, CorrCast serves as a platform for delivering new corrosion prediction
capabilities, such as wet gas top-of-line corrosion prediction, to ExxonMobil operating companies.
Model Description
CorrCast is based on a mechanistic corrosion model that accounts for the relevant chemistry and physics
of the corrosion process in a mixed aqueous-liquid hydrocarbon environment, which is depicted schematically in
Figure 1. The model has a number of advanced capabilities including:
The model has been validated against a wide range of field and laboratory data.
IPTC 13785
for an injected corrosion inhibitor. A sensitivity tool is included in the software implementation to test the variation
of results with changes to selected input parameters.
Corrosion rates can be predicted in CorrCast at selected points in pipes for field application or for a metal coupon
in a test condition-controlled vessel for laboratory application. Rates can also be predicted as a function of length
along a pipeline for field application, and as a function of time for a metal coupon in an autoclave for laboratory
application. A detailed description of the model and calculation methodology is described below.
Electrochemistry
model
(unscaled rate)
Water chemistry
model
Gas composition
(pCO2, pH2S)
Scaled
corrosion rate
Water analysis
(salts, organic acids)
Input
Flow parameters
(pipe geometry,
velocity)
Hydrocarbon
properties
(water cut, inhibitive)
Hydrocarbon effects
Water chemistry
The water chemistry model requires test/field conditions (temperature, pressure), gas composition (%H2S / PH2S,
and %CO2 / PCO2), and water analysis (salt concentrations, organic acids' concentrations, bicarbonate
concentration) as input parameters. Gas solubility is modeled using Henry's Law for CO2 and H2S. Fugacity
coefficients are estimated for CO2 and H2S to adjust for real gas pressure using the Peng-Robinson equation of
state.
The water chemistry model includes equilibrium reactions for: carbonic acid dissociation, hydrogen sulfide
dissociation, weak acid (HA) dissociation, and water dissociation. Salinity is represented by the ionic strength.
Equilibrium constants are taken from the literature and vary with temperature and salinity. pH and all the species
concentrations are computed from the water chemistry model and used as inputs for the electrochemistry model
which calculates the unscaled corrosion rate.
Electrochemistry model
+
CorrCast models the corrosion current by assuming direct attack by H , H2CO3 HCO3 , and HA at the surface of the
metal. The electrochemical reactions include the cathodic and anodic reactions listed below:
-
2H2CO3 + 2e = H2 + 2HCO3
2HCO3- + 2e- = H2 +2CO32+
2H + 2e = H2
2HA+2e = H2 + 2A2+
Fe = Fe + 2e
(1)
(2)
(3)
(4)
(5)
The corrosion current is equal to the anodic current which is equal to the sum of the cathodic currents:
(6)
IPTC 13785
The unscaled corrosion rate is therefore calculated using the following equation.
CR =
where
icorrMFe
= 1.155icorr
FenF
(7)
icorr is in A/m2, MFe is the molecular weight of iron in g/mol, Fe is the density of iron in kg/m3, n is number
of moles of electrons used in reducing or oxidizing a given species, F is the Faradays constant, and CR is in
mm/s, which can be conveniently converted to mm/y.
When H2S is present in the water solution, it is assumed that the adsorption of sulphur on iron affects the potential.
This effect varies with coverage and is fit by a Langmuir isotherm.
Scaling methodology
In CO2 and CO2/H2S corrosion environments, corrosion products such as iron carbonate (FeCO3) and iron sulfide
(FexS), can form and decrease the corrosion rate of carbon steel. Scale formation is therefore one of the key
factors affecting corrosion rate. CorrCast applies scaling factors for quantifying the effects of scale on corrosion at
the conditions of interest.
or
FeS
(8)
where CR scaled is the scaled corrosion rate, CR unscaled is the unscaled corrosion rate, and SFFeCO
3
or FeS
is the
scaling factor for FeCO3 or FeS, which is a function of temperature, pH, and local supersaturation.
Hydrocarbon liquids
CorrCast includes the effects of hydrocarbon liquids (crude oil or condensate) on corrosion in two ways. First, it
uses a viscosity based drainage model to estimate how much the liquid hydrocarbon contacts the metal surface
versus the aqueous phase. This is a function of the water cut and the relative viscosities of the water and liquid
hydrocarbon phases. If no data exists on inhibitive or corrosive properties of the liquid hydrocarbon, it is assumed
to be corrosion neutral (simply reduces corrosion by displacing the aqueous phase). Second, it can account for
inhibitive or corrosive properties of the liquid hydrocarbon using the empirical equation below
CR = CR actual
a
1
W
b+
(9)
where a and b are constants, and W and H are water and hydrocarbon volumes respectively. This equation can
account for species in the liquid hydrocarbon that partition into the water that further inhibit or accelerate corrosion.
The empirical constants a and b can only be obtained for a particular hydrocarbon from experimental data.
Therefore, laboratory tests are required to establish how field hydrocarbons, if present, may affect the corrosion
rate.
IPTC 13785
Validation Range
Validation ranges of key parameters in CorrCast are given below:
Gas
Impeller for
water/hydrocarbon
mixing
Test Material
Liquid
Baffles for
vortex
prevention
Impeller for
shear stress
production
Figure 3: Autoclaves at ExxonMobil's Materials and Corrosion Laboratory are specially designed to
replicate field conditions, including fluid chemistry and mass transfer
A large-diameter, multiphase flow loop can simulate flow regimes that are not attainable in a stirred autoclave,
primarily slug flow and flow with large upsets. Figure 4 is a photograph of the 102-mm (4-in) diameter multiphase,
Hastelloy C276 sour flow loop at the M&CL. This 30-m (100-ft) long flow loop pumps liquid and gas separately
IPTC 13785
and provides the ability to replicate all production flow regimes at inclinations between 0o and 90o. The maximum
gas and liquid velocities are ~9 m/s (30 f/s) and ~4 m/s (13 f/s) respectively. Liquid temperatures can be varied
from 5oC (40oF) to 120oC (250oF) with a maximum total pressure of 6.9 MPa (1,000 psi) and a maximum H2S
partial pressure of 0.2 MPa (30 psi). Both coupons and electrochemical probes can be used in the flow loop, which
provides the ability to remove coupons under pressure. Results can be extrapolated to other pipe sizes by
ensuring equivalent mass transfer between the flow loop and the service line based on the aforementioned
mechanistic relations.
Figure 4: 102-mm (4-in.) multiphase sour flow loop used for corrosion testing at the materials and
corrosion laboratory
Case Studies
Case 1
In this example, carbon steel was evaluated to determine its applicability for tubing in a new well at a
mature field. The well would initially be used as an appraisal well, producing oil for approximately three months,
and it would then be switched to a water injector for the remainder of its life. At this field, 13 Cr is used for tubing in
production wells to mitigate CO2 corrosion and carbon steel is used for tubing in water injection wells. If carbon
steel was not feasible in this case, 25 Cr Super Duplex would be required since 13 Cr is susceptible to pitting in
water injection service where there may be upsets in oxygen control. Reservoir and subsurface engineers
provided the design conditions during the initial production phase, including hydrocarbon liquid, hydrocarbon gas,
and formation water analyses for the reservoir (key component bicarbonate 160 ppm), expected range of
temperatures (55oC to 85oC) and pressures along the wellbore (PCO2 < 58 kPa, no H2S), and estimates of the flow
velocities (~1 m/s) and water cut (~10%).
Corrosion modeling was performed with a widely-used empirical model and CorrCast based on the expected upper
bound of CO2, the supplied formation water chemistry, and the range of temperatures expected along the wellbore.
The empirical model predicted general corrosion rates up to 6.9 mm/y (270 mpy). This model did not, however,
account for any beneficial effects from the formation of an iron carbonate scale or from the oil. This prediction
indicates that up to 1.7 mm of wall loss could occur during the three month production period, which corresponds
to more than 10% of the tubing thickness. In contrast, CorrCast, which includes the effect of a protective iron
carbonate scale, predicted general corrosion rates up to 0.5 mm/y (20 mpy) for 100% water cut. Corresponding
unscaled corrosion rate predictions were up to 8.0 mm/y (315 mpy), which agrees well with the empirical model,
but such high corrosion rates are unlikely because conditions should be favorable for the formation of an iron
carbonate scale. With the addition of oil (10% water cut), CorrCast predicted general corrosion rates of <0.1 mm/y
(<4 mpy). These predictions indicate that there will be less than 0.2 mm of wall loss during the three month
production period and that the prediction from the empirical model represents a conservative upper bound.
This example demonstrates how the advanced capabilities of the model enabled it to reduce the conservatism built
into the empirical model and provide a more realistic corrosion prediction. Based on this analysis, carbon steel
IPTC 13785
was used for the tubing in this well. Since the model was inside its validation range no laboratory testing was
conducted.
Case 2
In this example, inhibited carbon steel was evaluated for a wet gas flowline in highly corrosive service. The
o
flowline would transport full wellstream fluids with a CO2 partial pressure of 1,800 kPa at a temperature of ~30 C.
The reservoir is predicted to sour over time, giving rise to small amounts of H2S (up to ~0.7 kPa). Reservoir and
facilities engineers provided detailed conditions including gas and water analyses, flow rates, and temperature.
Corrosion modeling was conducted with CorrCast to confirm the need for inhibition and to compare against
uninhibited corrosion tests conducted at M&CL during the corrosion inhibitor qualification program. For the initial
conditions (sweet), CorrCast predicted general corrosion rates of 1.6-3.8 mm/y (64-155 mpy), which confirmed the
need for inhibition. A range of corrosion rates was reported because a poorly protective iron carbonate scale was
predicted to form (1.6 mm/y corresponds to the scaled rate, 3.8 mm/y corresponds to the unscaled rate). For the
end of life conditions (sour), CorrCast predicted a general corrosion rate of 0.2 mm/y (8 mpy). The reduction in
corrosion rate versus the initial conditions was due to the formation of a protective iron sulfide scale. CorrCast
predictions agreed well with uninhibited corrosion tests conducted on X52 and X65 pipeline steels. For the initial
conditions (sweet), the measured corrosion rates of the X52 and X65 coupons were 3.9 mm/y (154 mpy) and
1.4 mm/y (54 mpy), respectively. These rates agree well with the range predicted by CorrCast, and the measured
difference in rates between the two steels was attributed to differences in microstructure affecting the formation of
the iron carbonate scale. For the end of life conditions (sour), the measured corrosion rates of both steels was
0.2 mm/y, which agrees with the CorrCast prediction. Additional corrosion tests were conducted with inhibitor, and
the measured corrosion rates were 0.01 mm/y (0.4 mpy), which confirmed the inhibitor would mitigate corrosion at
the dosage tested.
This example demonstrates how an integrated approach of modeling and testing supported the use of inhibited
carbon steel for a challenging application. It also shows the ability of the model to predict the impact of corrosion
scales in both sweet and sour conditions. Based on the modeling and corrosion inhibitor qualification testing,
inhibited carbon steel was selected for the flowline.
Conclusions
This paper provides an overview of ExxonMobil's corrosion prediction model and its implementation in the
CorrCast software. It describes how to apply the model in conjunction with laboratory testing to evaluate the use of
carbon steel in oil and gas production environments. An integrated approach to corrosion modeling and laboratory
testing is important to ensure long-term reliability of carbon steel production equipment with minimal life-cycle cost.
Acknowledgement
The authors would like to acknowledge R. V. Reddy from ExxonMobil Upstream Research Company, and
J. C. Bondos and J. L. Nelson from ExxonMobil Production Company for early-stage corrosion model
development.
References
1. C. de Waard, D. E. Milliams, "Prediction of Carbonic Acid Corrosion in Natural Gas Pipelines", First
International Conference on the Internal and External Protection of Pipes, Paper F1, Cranfield, UK: BHRA
Fluid Engineering, 1975.
2. C. de Waard, U. Lotz, D. E. Milliams, "Predictive Model for CO2 Corrosion Engineering in Wet Natural Gas
Pipelines", Corrosion, Vol. 47, No. 12, p. 976, 1991.
3. C. de Waard and U. Lotz, Prediction of Corrosion of Carbon Steel, Corrosion/93, paper no. 69, Houston, TX:
NACE International, 1993.
IPTC 13785
4. CO2 Corrosion Rate Calculation Model, NORSOK standard No. M-506, http://www.nts.no/norsok, Oslo:
Norwegian Technology Standards Institution, 1998.
5. E. Dayalan, G. Vani, J. R. Shadley, S. A. Shirazi, and E. F. Rybicki, Modeling CO2 corrosion of carbon steels in
pipe flow, Corrosion/95, paper no. 118, Houston, TX: NACE International, 1995.
6. S. Nesic, J. Postlethwaite and S. Olsen, An electrochemical model for prediction of CO2 corrosion,
Corrosion/95, paper no. 131, Houston, TX: NACE International, 1995.
7. J. C. Bondos, R. V. Reddy, D. V. Pugh, D. A. Norman, J. L. Pacheco, and J. L. Nelson, Accurate corrosion
prediction through an integrated approach, SPE Production & Operations, SPE 111430, May 2007.