Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
CORROSION – IT’S ABOUT TIME
MetCorr-SES
Bruce Craig
MetCorr-Stress Eng Services
February 7, 2017
ASM Houston Seminar – Materials Selection and
Corrosion Management in Oil Field Environments
MONTAIGNE’S AXIOM
Nothing is so firmly believed
as what we least know.
MetCorr-SES
CORROSION RATE
We measure corrosion rates and predict them:
Using values like mpy which denote uniform thinning.
Crack thresholds and crack growth rates for
environmental cracking.
We then take such numbers to design equipment
or predict remaining life but is this adequate?
Is it correct?
MetCorr-SES
WHAT WE REALLY MEAN BY CORROSION RATE IS –
HOW LONG WILL IT LAST?
IT’S ABOUT TIME NOT RATE AND MOST OFTEN WE
DON’T HAVE A GOOD ANSWER.
MetCorr-SES
GENERAL/UNIFORM CORROSION
We have this down pretty good.
For example, CO2 corrosion models generally
assume uniform corrosion but as stated in a
review of models, “some CO2 models include
a “qualitative” risk for localized corrosion
attack” (Nyborg, Corrosion 2010, Paper 10371).
MetCorr-SES
CO2 CORROSION MODELS ARE QUITE GOOD
Results for tubing corrosion
prediction in the Haynesville
Shale.
Predicted TTF ~ 5 months.
Actual TTF = 6 months
MetCorr-SES
BUT WE CAN’T ALWAYS GET THE
MESSAGE ACROSS!
Email from Pipe Supplier
We sold a customer 13Cr tubing until they switched to 2-3/8”,L-80 to save money.
The pipe was in these wells for one year, then failed from corrosion.
They now have more than 1,000,000 ft. of steel tubing in Haynesville Shale wells.
MetCorr-SES
MetCorr-SES
WALL LOSS AND CORROSION ALLOWANCE
Modelling CO2 corrosion typically assumes uniform
wall loss then we adjust the CA to suit the desired life
time.
However, pitting (localized attack) is not linear and
there are no adequate models at present to evaluate
the usefulness of CA for pitting.
So how much time do we have (TTF) when pitting is
the mechanism?
MetCorr-SES
CRAs
Fail by Localized attack – pitting, EC
Currently no means to predict expected life of
CRAs so we are overly conservative.
PREN is frequently misapplied to downhole
oxygen free environments.
MetCorr-SES
FUNDAMENTAL ISSUES WITH PITTING AND EC
PREDICTION FROM A TIME PERSPECTIVE
MetCorr-SES
PITTING
The common practice in the oilfield is to use a multiplier such as 10x
the uniform metal loss.
However, the situation is far more complex and a simple rate is not
adequate to define the life of a component in service.
Pitting is a stochastic (probabilistic) process that involves;
Pit initiation (this includes an incubation time)
Metastable pitting with possible repassivation
Stable pitting and growth/ or death and arrest.
MetCorr-SES
SHALLOW PITS AND THE INITIATION OF
SUBCRITICAL EC
MetCorr-SES
Critical pitting potential and repassivation potential can be measured in the lab
but are not easily translated into actual field process conditions that can be
monitored for pitting prediction.
Extreme value methods (ie Gumbel) can be used for direct measurement of
existing pits to predict pit growth rates but not pit initiation which for CRAs is more
important.
Since there are no ready accurate models for pitting of CRAs we make very
conservative choices for their application to avoid any pitting and EC.
MetCorr-SES
PITTING AND EC ALWAYS HAVE AN INCUBATION TIME
This is what we need to focus on – time to initiate
attack.
Complete resistance would = ∞ incubation time.
Poor resistance would = seconds/minutes/hours/days
MetCorr-SES
ENVIRONMENTAL CRACKING
HSC (HE) – Fast
(A common question is how long will Q125 last
when exposed to H2S in my well?)
SCC – Slow
(How slow is slow? Depends on many factors including
inspection interval)
MetCorr-SES
SCC – WHEN WILL CRACKS ALL CONNECT?
MetCorr-SES
HOW LONG DO WE HAVE BEFORE LINKAGE?
18HIC in Plate Steel Exposed to Sour Gas
MetCorr-SES
“One of the toughest thresholds to dispel is KISCC”.
“The incorrect concept of SCC thresholds may be
related to the desire of designers and engineers to
identify conditions that will permit the implications
of SCC to simply be ignored.”
Peter Andresen (recognized authority on SCC) in Proc. Of
Symposium on Chemistry and Electrochemistry of Corrosion and
Stress Corrosion Cracking.
MetCorr-SES
BUT STILL WE PERSIST-API 17TR8
MetCorr-SES
• H controlled Kth and [da/dt]II can be predicted for some alloys from mechanism-based understandings and limited experiments. However, this is contingent on knowledge of:Mechanics of crack tips and material & environment dependent fracture
mechanismsMaterial-dependent hydrogen-metal interactions and H transport ratesMaterial, geometry and chemistry dependent crack tip hydrogen
concentrations
• Many parameters remain elusive or difficult to determine at the correct length scale but new techniques may enable such determination. (Prof. John Scully, UVA).
• In other words you can’t simply take KISSC from FM tests and stick it into LEFM equations!!!
MetCorr-SES
HYDROGEN STRESS CRACKING (aka HE)
For thin wall components (and even many thicker ones) ie. casing
steels, pipeline steels, etc. the idea of crack growth rate is
meaningless.
MetCorr-SES
THE IMPORTANT FACTOR IS CRACK INITIATION NOT
PROPAGATION.
Most well tubulars are < 1” wall as are
pipelines.
In H2S at even 10 inch/min that means the
crack will cross the entire wall in 6 minutes!
MetCorr-SES
HYDROGEN DIFFUSION AND TRAPPING
H
H
H
H
H+
H+
H+
H+
H+
H+
H+
H+
• Casing ID• Riser ID• PL ID
Hydrogen gradient
Stress
Pipe OD
It takes time to
accumulate Hcr
MetCorr-SES
DSM WELDS (NO PREEXISTING FLAWS)
MetCorr-SES
DSM WELDS (NO PREEXISTING FLAWS)(BUT FOLKS DRIVE THEMSELVES CRAZY LOOKING FOR A PREEXISTING DEFECT!)
MetCorr-SES
API FULL SCALE CASING SSC TEST
W/PRE-EXISTING FLAWS
MetCorr-SES
CRACK INITIATION, SMOOTH SURFACES AND
INCUBATION TIME
MetCorr-SES
INCUBATION TIMEMetCorr-SES
Elsea and FletcherDMIC 1964
CRACK INITIATION
Crack initiation at the bottom
of a notch in 304 SS at surface
instability.
sinusoidal surface
profile due to wave
growth induced by an
applied stress, σ.
(Asaro-Tiller)
MetCorr-SES
NACE MR0175/ISO 15156
P110 Casing in Sour Shale Wells
NACE says one factor to consider is time of exposure but how?
• Millions of feet of P110 casing in sour Eagle Ford and Haynesville fields with very few failures due to SSC.
• NACE MR0175/ISO 15156 prohibits P110 since it is prone to failure from SSC.
MetCorr-SES
FAILURE OF AN EARLY PRODUCTION RISER
(HE IN CO2 w/o H2S)
Sufficient hydrogen from the CO2 corrosion reaction plus zinc
thread dope.
MetCorr-SES
STRESS CORROSION CRACKING
Fast or slow cracking
SCC of pipeline steels – slow
HIC – slow
But the important factor for both of these is
branched crack linkage which can’t be
predicted.
Therefore, crack initiation is the key.
MetCorr-SES
THE FUTURE – MULTISCALE MODELING (MSM)
MetCorr-SES
Overly simplistic LEFM and pitting factors
will not and do not work.
AtomsCrystal
StructureMicrostructure Component
Thermo-mechanical
Treatments
Phases Defects+
• Vacancies
• Dislocations
• Twins
• Stacking Faults
• Grain Boundaries
• Voids
• Cracks
& their distribution
Travel across length scales to reach the scale of the component
MetCorr-SES
YS
KI
Etc.
MetCorr-SES
Unit Cell* Crystalline Defects Microstructure Component
*Simple Unit Cells
AngstromsDislocation Stress fields
→ NanometersMicrons Centimeters
Grain Size
Lengthscales in Materials Science
Start with a cursory look at the length scales involved in Materials Science
MetCorr-SES
MSM OF PIT INITIATION
MetCorr-SES
Herbert et. al.,
Electrochimica Acta, 2014
MetCorr-SES
HE modeling using
Quasi-Continuum
Density Function
Theory (QCDFT),
quantum mechanics
(QM) and molecular
dynamics (MD).
Zhang and Lu
MetCorr-SES
CONCLUSIONS
Simply modeling uniform corrosion (wall loss), pit propagation and crack growth rates are not sufficient for future oil and gas projects.
In fact these approaches are non-conservative and may be dangerously misleading.
In thin wall components and for HSC, initiation of pitting and cracking are of paramount importance -not propagation.
In other words its about time – incubation time.
MetCorr-SES
DEDICATION
Dedicated to two good friends and corrosion engineers that passed away last year.
Don Drake and Tom Bednarowicz