Beruflich Dokumente
Kultur Dokumente
Guidelines
Revision 2.1
(1999)
J W Martin
Main CD
Contents
EROSION GUIDELINES REVISION 2.1 (1999)
By
J W Martin
Summary
Erosion can be defined as the mechanical loss of material by the impact of liquid
droplets and/or solid particles.
Under aggressive operating conditions velocity limits, and hence production limits, are
set to avoid erosion. If these limits are overly conservative then BP AMOCO loses
production; if they are too optimistic then BP AMOCO risks erosion damage and the
loss of system integrity.
This document updates the knowledge on the erosion of piping and tubing in
production and injection service (Ref. 1). The two 'Flow Charts' for the assessment of
erosion risk have also been updated:
The 'Velocity Limits for Avoiding Erosion' flow chart lays down rule-of-
thumb velocity limits for the avoidance of erosion damage in non solids-
containing environments, i.e. ‘totally solids free’ or ‘nominally solids free’
conditions. ‘Nominally solids free conditions’ are defined as up to 1 pound of
solids per thousand barrels of liquid for liquid systems or up to 0.1 pounds of
solids per million standard cubic feet of gas for gas systems.
Different velocity limits will apply in different situations, depending on the flow (gas,
liquid or multiphase gas/liquid), the environment (corrosive or non-corrosive) and
whether or not solids are present.
The models used for the calculation of erosion wastage rates are based, in the main, on
laboratory test programmes. Hence they are likely to be at their most reliable for
simple flow conditions in non-corrosive environments. There is less confidence in the
models for multiphase solids erosion and guidance for erosion-corrosion (solids plus
corrosive environment), as these are based on a very limited data set.
All of the predictive models suffer from limited comparison with field experience.
Contents
Erosion Guidelines....................................................................................................... 1
Summary.......................................................................................................... 1
Contents .......................................................................................................... 2
Summary Guidelines - Flow Charts and General Comments ......................................... 3
Figure 1 - First Pass Velocity Limits................................................................. 4
Figure 2 - Calculation of Erosion Rates ............................................................ 5
Notes on Flow Charts. ..................................................................................... 6
Figure 1 - First Pass Velocity Limits ..................................................... 6
Figure 2 - Calculation of Erosion Rates................................................. 8
General Comments and Conclusions................................................................. 10
Erosion Guidelines - Discussions ................................................................................. 12
Introduction ..................................................................................................... 12
Discussion of the Guidelines............................................................................. 17
1. Non-corrosive fluid flow, no solid particles ...................................... 17
2. Corrosive fluid flow, no solid particles ............................................. 17
3. Non-corrosive fluid, with solid particles ........................................... 19
References: ...................................................................................................... 30
2
Summary Guidelines - Flow Charts and General Comments
A flow chart for determining 'first pass' erosional velocity limits (entitled "Velocity
Limits for Avoiding Erosion") is given in Figure 1. The recommendations in this
flow chart are generally based on an allowable erosion rate of 0.1 mm/yr.
For ‘totally solids free’ or ‘nominally solids free’ conditions, if production is required
outside these limits then advice can be sought from the relevant specialists in the
Upstream Technology Group (UTG).
A flow chart for the assessment of erosion rates (entitled "Calculation of Erosion
Rates") is given in Figure 2. It is recommended that this is used with care. There are
many areas of uncertainty and the models recommended in the flow chart are relatively
unproven and many are still being developed. Under conditions of erosion-corrosion
the guidelines in Figure 1 are applicable for estimating the erosion-corrosion rates.
3
VELOCITY LIMITS FOR AVOIDING EROSION
Note 1: Data Collection
Gas/Liquid ratio. Production rates. Tubing or piping internal bore. Solids present or
absent. Gas and liquid densities at temperature and pressure (if these are not known
then a rough assessment can be made on the basis of an oil density of 800 kg/m3, a Data Collection
water density of 1000 kg/m3 and a gas density of 1 kg/m3 at STP and then adjusting the
density for pressure and temperature.)
For the 'full' Harwell and Tulsa Models: where E is the erosion rate in mm/yr, W is the sand flow rate in kg/day, V is the mixture velocity in m/s, D is the sand size in microns, d is the
The data indicated above plus; Gas Viscosity (under pipe internal diameter in mm, ρm is the fluid mixture density in kg/m3.
operating conditions). Liquid Viscosity (under operating
conditions). Solids density and 'shape' (e.g. sharp, From the assessment of the Salama Model undertaken within UTG, it best equates to a 5D bend situation in comparison with the 'full'
semi-rounded). CO2 and H2S partial pressures. Tubing Tulsa/Harwell models. It is therefore recommended that it is not used for systems where geometrical features other than 5D bends may be
or piping geometry and configuration. Steel hardness present (e.g. 1.5D elbows, tees, severe constrictions). The model is most probably suitable for application to downhole completions, although
(if material of construction is a carbon/low alloy steel). in this instance care needs to be taken regards regions of significant flow constriction (e.g. insert valves).
A very simplified version of the Salama model (developed by Salama & Venkatesh), applicable to gas systems with carbon steel bends (including
1.5D elbows, tees, etc.) is:
E = 604 x MV2/d2
where E is the erosion rate in mm/yr, M is the solids production rate in g/s, V the mixed velocity in m/s and d the pipe diameter in mm.
Note 2: Salama,
Gas, No Liquid?
Yes (Salama and Venkatesh)
Pure gas streams. No significant liquid loading.
Gas, no liquid?
or
Full Tulsa
Note 5:
No
1st Pass: Salama, RCS and/or API model, 2nd Pass: Full Tulsa model.
In liquid systems particle impact velocities are reduced by the flow regime and the presence of a liquid buffer layer at the metal surface. The
RCS and API models are based on empirical tests in liquid piping and bends and have built-in allowances for such effects. This does mean,
however, that there can be scaling problems in different geometries or with different solid particle sizes. The Salama model is still a 'simplified'
model, but will take some account of solid particle sizes.
where E is the erosion rate in mm/yr, W is the sand flow rate in kg/day, V is the mixture velocity in m/s, D is the sand size in microns, d is the
pipe internal diameter in mm, ρm is the fluid mixture density in kg/m3.
From the assessment of the Salama Model undertaken within UTG, it best equates to a 5D bend situation in comparison with the 'full'
Tulsa/Harwell models. It is therefore recommended that it is not used for systems where geometrical features other than 5D bends may be
present (e.g. 1.5D elbows, tees, severe constrictions). The model is most probably suitable for application to downhole completions, although
in this instance care needs to be taken regards regions of significant flow constriction (e.g. insert valves).
Simplified versions of the RCS and the API models, applicable to carbon steel bends, are:
where E is the erosion rate in mm/yr, M is the solids production rate in g/s, V the mixed velocity in m/s and d the pipe diameter in mm.
Note 4:
Note 6: No
Slug Flow?
The Harwell model for multiphase erosion is
based on vertical flow. Under such conditions
slug flow, which leads to liquid being thrown No No No
down onto the bottom of a pipe, is not produced.
Slug flow? Stratified flow? Bubble/Churn Annular flow?
Thus the standard Harwell models for annular
mist, churn and bubble flow are not applicable.
Flow?
In slug flow the 'liquid slug' will be thrown against
the pipe wall at velocities approaching the net
mixed velocity. In addition, at the slug front there
will be considerable mixing and hence entrained
gas, such that the slug front will approach the
homogenous mixture. Therefore, it is Yes Yes Yes Yes
recommended that the pure liquid models be
used (see Note 5) but that the mixed fluid velocity
and mixture properties should be used rather
than the liquid velocity and density.
Note 8:
1st Pass: Harwell
Slug Flow? 1st Pass Salama,
Full Tulsa Model Harwell
RCS and/or API Model Salama,RCS and/or
Model
Use the mixed (averaged) fluid density and and/or API Model Full Tulsa Model
velocity
Stratified Flow? Full Tulsa Model Bubble/Churn Flow? Harwell Annular Flow? Harwell Model.
Use the liquid velocity calculated for and/or Full Tulsa Model For comparison, check using the Full
the hydraulic diameter Do not use the Tulsa Model for Tulsa Model with the mixed velocity and
Churn flow. For bubbly flow with with:
the Full Tulsa Model use the mixed (i) Mixed (averaged) fluid properties
(averaged) velocity and liquid (ii) Liquid properties
properties The actual erosion rate should be
somewhere between the two values.
Note 9:
GQS38294/1
Notes on Flow Charts.
General Comments:
Velocities refer to net mixed velocities (nominal gas velocity plus nominal liquid
velocity). Units are in ft/s (1 m/s = 3.281 ft/s).
ρm refers to mixed fluid density in lbs/ft3 (1 kg/m3 = 0.06242 lbs/ft3)
C factors relating Vmax to √ρm are in ft/s(lbs/ft3)1/2. Multiply by 1.22 to convert to C
factors in m/s(kg/m3)1/2
pptb - pounds of solids per thousand barrels of liquid.
lb/mmscf - pounds of solids per million standard cubic feet of gas.
Advice on erosion-corrosion is best available at time of publication. The situation is
uncertain and the guidelines are subject to change.
Further advice can be obtained from the relevant specialists in UTG.
6
the solids will ‘settle’ out of the flow stream creating a stationary bed or more slowly
moving bed of solids.
7
FCRCS - ‘Filmed’ corrosion rate for carbon steel
CR13Cr -Corrosion rate for 13%Cr steel
or
• A system where the materials of construction are fully corrosion resistant to the
anticipated conditions.
Note 14: Non-corrosive; Gas no liquid; No Velocity Limits for the Avoidance of
Erosion
There are other flow related phenomena that need to be considered for high velocities,
e.g. noise and vibration.
Note 15: Non-corrosive; Liquid no gas; No Velocity Limits for the Avoidance of
Erosion
It is important to take necessary steps (including possibly limiting the fluid velocity) to
avoid other possible problems, such as cavitation; plant noise/vibration; water hammer;
etc.
Note 17: Corrosive; Liquid no gas; No Velocity Limits for the Avoidance of
Erosion
8
Note 18: Vmax=200/√ √ ρ m or 20m/s whichever is less
Corrosion inhibition selection will need to take account of the fact that the inhibitor will have
to ‘work’ under flowing conditions and it may be possible to select an inhibitor that will
‘work’ at velocities above the limits defined here.
11
Figure 2 - Calculation of Erosion Rates
General Comments:
Advice is best available at time of publication.
Most of the models used assume sharp sand particles with a diameter of 150 µm. The
Salama model (used for single phase gas or liquid conditions only) and more detailed
Tulsa and Harwell models can make allowances for solids particle size (all three
models), plus density and shape (Tulsa model only).
The erosion calculations are generally for bends and conditions of turbulence (e.g.
constrictions) only. The exception to this is the Tulsa model that has a (as yet
untested) module for evaluating the erosion rate in straight pipe. In general, erosion in
straight sections is at least an order of magnitude less than at bends. The only
exception to this will be horizontal slug flow where liquid is thrown against the pipe
wall.
The Full Tulsa Model is available as a computer software package (‘Sand Production
Pipe Saver’; SPPS v. 4.1.)
The Harwell Model is available as a computer software package (‘Design Procedure
for Erosion-Corrosion in Multi-phase Flow’; Sandman v. 3.9.).
Further advice can be obtained the relevant specialists in UTG.
where E is the erosion rate in mm/yr, W is the sand flow rate in kg/day, V is the
mixture velocity in m/s, D is the sand size in microns, d is the pipe internal diameter in
mm, ρm is the fluid mixture density in kg/m3.
From the assessment of the Salama Model undertaken within UTG, it best equates to a
5D bend situation in comparison with the ‘full’ Tulsa/Harwell models. It is therefore
10
recommended that it is not used for systems where geometrical features other than 5D
bends may be present (e.g. 1.5D elbows, tees, severe constrictions). The model is most
probably suitable for application to downhole completions, although in this instance
care needs to be taken regards regions of significant flow constriction (e.g. insert
valves).
A very simplified version of the Salama model (developed by Salama & Venkatesh),
applicable to gas systems with carbon steel bends (including 1.5D elbows, tees, etc.)
is:
E = 604 x MV2/d2
where E is the erosion rate in mm/yr, M is the solids production rate in g/s, V the
mixed velocity in m/s and d the pipe diameter in mm.
Note 5: 1st Pass: Salama, RCS and/or API model, 2nd Pass: Full Tulsa model.
In liquid systems particle impact velocities are reduced by the flow regime and the
presence of a liquid buffer layer at the metal surface. The RCS and API models are
based on empirical tests in liquid piping and bends and have built-in allowances for
such effects. This does mean, however, that there can be scaling problems in different
geometries or with different solid particle sizes. The Salama model is still a
‘simplified’ model, but will take some account of solid particle sizes.
where E is the erosion rate in mm/yr, W is the sand flow rate in kg/day, V is the
mixture velocity in m/s, D is the sand size in microns, d is the pipe internal diameter in
mm, ρm is the fluid mixture density in kg/m3.
From the assessment of the Salama Model undertaken within UTG, it best equates to a
5D bend situation in comparison with the ‘full’ Tulsa/Harwell models. It is therefore
recommended that it is not used for systems where geometrical features other than 5D
bends may be present (e.g. 1.5D elbows, tees, severe constrictions). The model is most
probably suitable for application to downhole completions, although in this instance
care needs to be taken regards regions of significant flow constriction (e.g. insert
valves).
Simplified versions of the RCS and the API models, applicable to carbon steel bends,
are:
RCS:
E = 4.1 x MV2.5/d2
API:
E = 5.33 x MV2/d2
11
where E is the erosion rate in mm/yr, M is the solids production rate in g/s, V the
mixed velocity in m/s and d the pipe diameter in mm.
Note 8: Slug Flow? 1st Pass Salama, RCS and/or API Model
Use the mixed (averaged) fluid density and velocity
12
General Comments and Conclusions
4. Erosion depends critically on the fluid flow regime. Solid particles carried in
gas flow may hit pipe walls at the full gas velocity, although it is likely that in many
cases the solid particles will drop out of the gas stream and either form a static bed or a
moving bed (moving dunes, scouring). Under full liquid flow, solid particles will
frequently travel at the liquid velocity, but will be significantly slowed by a liquid
barrier layer on the pipe wall before striking the pipe material surface. Under
multiphase flow, some solid particles may be carried at/near the gas velocity (if the
flow is annular mist) and may or may not be slowed down by a liquid barrier layer -
depending on the thickness of the annular liquid film. Careful assessment and a
knowledge of flow regimes is required in such cases.
5. Empirical and field data suggest that there is a threshold solid particle size
below which erosion will not occur. This threshold is unclear and probably relates to
whether, at the net fluid velocity, a given particle has the momentum to carry it
through the barrier fluid at the pipe or tubing surface. It should be noted that such
thresholds are only applicable to ‘normal flow’ conditions. For example, is has been
found that for downhole sand screens through which very small particles can ‘pass’
even these very small particles can result in erosion due to the very high energy flow
13
and high probability of impacting the metal surface. Most of the work reported is based
on sand particles of 150 µm diameter. However, the full Tulsa model can make
allowance for different particle sizes, densities, shapes and sharpness. The full Harwell
model and the Salama model can make allowance for different particle sizes.
6. Most erosion damage will occur at bends and flow disruptions and is likely to
be at least an order of magnitude greater than erosion in straight pipe or tubing. The
possible exception to this is slug flow where flow can impact on the pipe or tubing wall
on straight sections. The full Tulsa model now contains a module (as yet not validated)
for erosion in straight pipe. Presently this only covers single phase flow (e.g. slug flow
is not covered).
All of the above are referred to at times as erosion or erosion-corrosion. In this report
the first is referred to as flow-enhanced corrosion. The second and third are forms of
enhanced corrosion resulting from erosion-corrosion.
14
11. In solids-containing environments, the situation for erosion-corrosion is
unclear. If either the expected erosion or expected corrosion are an order of magnitude
less than the other then synergistic effects are likely to be small. Laboratory data
suggests that solids erosion can lead to severe localised attack in carbon steel if the
erosivity is below a certain value or totally destroy a region of protective layer at
higher values (leading to general corrosion but not penetrating the wall so quickly).
There is evidence to suggest that, in anaerobic CO2 containing environments, solids
can damage protective layers on 13 % Cr materials leading to erosion-corrosion at
temperatures up to 80°c. Above this the 13%Cr steel has been found to re-film very
quickly, i.e. no synergy between erosion and corrosion is expected. Results on duplex
stainless steel suggest that there is no corrosion-erosion synergy - implying that the
wastage is only through erosion.
15
Erosion Guidelines - Detailed Discussions
Introduction
Erosion can be defined in a variety of ways, but is essentially the wastage of material
due to the mechanical removal of material surfaces by flowing environments. Such
wastage is most extreme when solids are present in the environment.
Erosion problems are likely to increase in BP AMOCO in the future because of:
• increased water cuts putting pressure on total fluid production rates to maintain
oil production,
• increased use of multiphase flow in the transport of production fluids,
• increased sand and solids production rates due to a number of factors, such as
increased water cut, use of proppant and reservoir fracturing techniques.
Many flow dependent wastage mechanisms are termed "erosion". For produced fluids
there are four main mechanisms to be considered:
The third of these is sometimes confused with flow-enhanced corrosion, where the
flow regime leads to enhanced mass transport of corrosion products and reactants. In
these Guidelines erosion-corrosion in the absence of solids is taken to refer to
enhanced wastage due to the physical rupture of the protective, corrosion-product
layer by energetic fluid flow regimes and the consequential corrosion. The mechanical
removal of inhibitor might be defined as a form of erosion-corrosion but is not
discussed in detail in these Guidelines.
16
Discussion of the Guidelines
1. NO solid particles
The Guidelines in this Section are only applicable to ‘totally solids free’ conditions, i.e.
where there is no risk of solids particles being transported in the flowstream. It should
be recognised that even very low levels of solids (below the detection levels of even
‘state of the art’ solids monitoring techniques) can cause significant wastage (erosion
or erosion/corrosion) rates. Hence it is encumbent on the user of these Guidelines to
ensure that there is no risk of solids entrainment before using the guidance in this
Section.
For single phase non-corrosive liquid flow (i.e. totally solids free and with no entrained
gas bubbles) there are no velocity limit requirements to avoid erosion damage.
However, it is important to take necessary steps (including possibly limiting the fluid
velocity) to avoid other possible problems, such as cavitation1; plant noise/vibration;
water hammer2; etc.
Liquid droplet erosion (e.g. in annular mist flow) of metals under non-corrosive
conditions in the total absence of solids will only be a concern at velocities above
70 m/s (230 ft/sec). This is the maximum velocity limit defined to avoid the possibility
of droplet erosion for gas-condensate wells in the DNV Recommended Practice (Ref.
13). Totally solids-free, non-corrosive fluids at such high velocities are relatively
uncommon in oil/gas field service, with the possible exception of flow through choke
valves, which are covered in separate Guidelines.
1
Where liquid pressures are at or near the vapour pressure/gas bubble point pressure then bubbles can
form at regions of localised pressure drop these can then implode abruptly at points where the local
pressure rises again above the saturation/bubble point pressure. These implosions can cause removal
of material [cavitation] and/or noise problems.
2
Water hammer results from the shock pressure due to the sudden stopping of a liquid (e.g. when
closing a valve or where reciprocating pumps or compressors are used). The magnitude of this shock
pressure is a function of the fluid velocity, the stoppage time and the elasticity of the pipe. The
accompanying mechanical vibrations can result in fatigue failure if corrective actions are not taken.
17
suffer accelerated attack as the flow rate increases, as a result of increased mass
transport.
The situation is made much worse if the flow rate increases enough to cause erosive or
mechanical breakdown of protective layers. There are two circumstances to be
considered - multiphase gas-liquid flow and single phase liquid flow. The former is
generally much more energetic than the latter and thus more likely to lead to
mechanical disruption of protective product layers.
For single phase liquid flow (i.e. totally solids free and with no entrained gas bubbles)
there are no velocity limit requirements to avoid erosion damage. However, as note
above, it is important to take necessary steps (including possibly limiting the fluid
velocity) to avoid other possible problems, such as enhanced corrosion under flowing
conditions; cavitation1; plant noise/vibration; water hammer2; etc.
For wet (i.e. potentially corrosive) gas and multi-phase flow conditions, in the specific
case of inhibited carbon steel it is recommended that the maximum velocity for design
considerations should be taken as C=200 or 20m/s (whichever is lower). However,
corrosion inhibition selection will need to take account of the fact that the inhibitor will
have to ‘work’ under flowing conditions and it may be possible to select an inhibitor
that will ‘work’ at velocities above the limits defined here. For other
materials/conditions it is recommended to consider the limits for ‘nominally sand-free’
conditions as an interim measure, as there is little/no information available on how the
limits for these materials/conditions may differ for totally solids free conditions (i.e.
where the only erosion damage mechanisms are the result of liquid droplet or gas
bubble impingement).
For the purpose of these Guidelines ‘nominally solids-free’ conditions are defined as
less than one pound of solids per thousand barrels of liquids (<1pptb) for ‘liquid’ (e.g.
oil/water) systems and less than 0.1 pounds of solids per million standard cubic feet of
gas (<0.1lb/mmscf) for gas systems.
The origin of the 1pptb limit is that this was determined to be the minimum level of
solids that could be detected using ‘state of the art’ sand detection tools. The
0.1lb/mmscf was determined to be the equivalent quantity of solids for a gas system.
Therefore these limits should be applied to systems where there is the possibility of
solids being present, but where these are likely to be (or actually are) below the limits
of detection when using ‘state of the art’ sand detection monitors3.
3
Note the limit of detection of less rigorous sand detection methods is significantly less than these
limits. For example in the case of the ‘shake out’ centrifuge test the limit of detection is only 275pptb
and the limit of detection for the ‘Leutart Sampler’ is 5pptb. This must be taken into account when
determining whether a system can be considered ‘nominally solids free’ or not.
18
There are no specific issues for ‘nominally solids free, non-corrosive fluid flow’
conditions. This can best be dealt with in the same way as for ‘Solids containing - non-
corrosive fluid flow’ (Section 3.1.) with the solids content being set to 1pptb or
0.1lb/mmscf, as appropriate for the particular application.
There are two circumstances to be considered - multiphase gas-liquid flow and single
phase liquid flow (NB guidance for wet gas is included under multi-phase gas-liquid
flow, as liquid water is required for corrosive conditions). The former is generally
much more energetic than the latter and thus more likely to lead to mechanical
disruption of protective product layers.
In the case of 13%Cr steel the C-factor of 300 was determined from previous testing at
AEA Harwell (Ref. 14) and field experience. Rather than defining a true 'velocity limit'
above which unacceptable erosion/erosion-corrosion will occur, this represented the
maximum C-factor for which data was available and for which there was no evidence
of unacceptable erosion/erosion-corrosion. Therefore, it represents a limit of
understanding rather than an actual acceptance limit. A number of E&P Business Units
have identified a need to exceed the present maximum allowable velocity to maximise
production. There is therefore a clear business driver to understand the maximum flow
rates that could be allowed for 13%Cr steel. As a result if this a Project has been set up
within the ‘No Corrosion R&D’ programme for 1999/2000 to evaluate the maximum
allowable velocity for 13%Cr steel via ‘Field Tests’ on gas flowlines in the Tuscaloosa
(Louisiana) Field.
Similarly for duplex stainless steel, the C-factor of 350 was established by examining
the limits of data available from previous testing at AEA Harwell, published
information and field experience. The test work in 1999/2000 to evaluate the maximum
19
allowable velocity for 13%Cr steel may well be extended to duplex stainless steel to
examine if this C-factor can be increased, if there is sufficient Business Unit interest.
In stratified and annular mist flow direct impingement on the pipe wall will be most
severe at bends. The situation with multiphase slug flow is more uncertain. In slug
flow the churning and breaking wave at the leading edge of a slug can give rise to
perpendicular impacts on the bottom of straight horizontal pipe as well as at bends.
There is currently no well defined limit for the initiation of such damage, especially as
the situation is complicated by the presence of significant mixing and entrained gas
bubbles in the slug front. If it is assumed that the liquid slug impacting on the wall
needs to have the same impact velocity as above and that the liquid slug impact
velocity is, at worst, equal to the mixed fluid velocity, then the API limit with C=135
ft/s(lbs/ft3)0.5 could be applicable in the case of carbon steel. Thus for carbon steel if
slug flow is established and if the mixed fluid velocity is above the API limit with
C=135 ft/s(lbs/ft ft3)0.5 then pitting damage could be expected at any location all along
the bottom of a pipe. The situation might be mitigated somewhat if the protective
layer on carbon steel can reform between slugs This is not possible in continuous
annular flow and not likely at bends in slug flow. (NB apply the same principle but use
C=300 and 350 ft/s(lbs/ft3)0.5 for 13 % Cr steel and duplex stainless steel respectively)
The situation is further complicated in multiphase annular mist flow and multiphase
slug flow when corrosion inhibitors are added. There is some suggestion that
corrosion inhibitors might be effective up to the same velocity as protective corrosion-
product layers (Refs. 4 & 5). If this is the case, then once the thresholds for physical
damage to protective corrosion-product layers have been reached, corrosion inhibition
is unlikely to be effective. However, the strength of the bond between the corrosion
inhibitor and the metal surface may be greater than that of the precipitated corrosion
product layer. The latter is only physically bonded to the metal surface whereas the
corrosion inhibitor will be chemically bonded and perhaps more able to resist
displacement. If the corrosion inhibitor is bonded to the corrosion product layer then
the layer/metal bond may be the weak link. In such a case the erosion may clean the
surface of weakly bonded corrosion product layers and the corrosion inhibitor can then
bond directly to the bare metal surface, providing far greater resistance to corrosion
even under erosive conditions.
Flowing sand particles do eventually remove a corrosion inhibitor film from a steel
surface in experiments using an impinging liquid jet containing sand. However, work
at the University of Tulsa showed that a suitable corrosion inhibitor chemical was still
beneficial, by significantly increasing the safe operating velocity of the fluids by as
much as a factor of 4 or 5. These are still preliminary findings for a particular product
and set of conditions. It is not yet possible to derive a semi-quantitative rule of thumb.
For wet (i.e. potentially corrosive) gas and multi-phase flow conditions, in the specific
case of inhibited carbon steel it is recommended that the maximum velocity for design
considerations should be taken as C=200 or 20m/s (whichever is lower). However,
corrosion inhibition selection will need to take account of the fact that the inhibitor will
have to ‘work’ under flowing conditions and it may be possible to select an inhibitor
that will ‘work’ at velocities above the limits defined here.
20
Loss of corrosion inhibitor from bulk fluids by adsorption onto the surface of sand
particles can be a significant effect under certain circumstances ,such as high inhibitor
concentrations (>150 ppm) and high sand concentrations (>35 pptb). The adsorption
losses are normally insignificant for low corrosion inhibitor concentrations (<50 ppm)
and low sand concentrations (<35 pptb).
However, the situation in the field is often aggressive enough to require the use of
corrosion inhibitors. If this is the case, highly turbulent flow will increase corrosion
rates further. Some corrosion inhibitors perform poorly under highly turbulent flow
conditions whilst others can perform acceptably under extremely aggressive flow. In
general, the more turbulent the flow regime, the higher concentration of inhibitor that
will be required to achieve acceptable corrosion rates and therefore operating costs
will increase. Under such circumstances corrosion inhibition selection (and dosage
levels) will need to take account of the fact that the inhibitor will have to ‘work’ under
flowing conditions upto the maximum liquid velocity expected. In addition, flow
velocities in excess of 10 m/s should be viewed as high and extra thought given to
corrosion control and monitoring. UTG have issued guidelines on the prediction and
monitoring of CO2 corrosion (Refs. 6 & 7).
For 13%Cr steel it is recommended that the C-factor developed for ‘multi-phase’ flow
of 300 is used in the absence of any better information (this is likely to err on the
conservative side).
For duplex stainless steel a series of laboratory based flow loop tests were carried out
on behalf of BP Amoco by Det Norske Veritas Industry AS (DNV), Norway using
treated sea water. Interpretation of the test results demonstrated that for single phase
liquid flow a C-factor of 450 ft/s(lbs/ft3)0.5 could be applied for the ‘nominally solids
free’ condition of up to 1pptb (Ref. 15).
3. Solids-containing flow
3.1.1. Introduction
Although the specific erosion models produced by the different R&D programmes are
not always in good agreement, there are several areas of general agreement.
The basic mechanism of erosion of most metals (i.e. ductile materials) is ductile
ploughing of the surface by impacting solid particles. The material lost per impact is
greatest at angles of impact between 15° and 60° and is proportional to m(Vi)n where n
is between 2 and 2.5, m is the particle mass and Vi the actual particle impact velocity.
The overall wastage rate is then the mass loss per impact times the impact rate. In the
simplest case, the rate of impact is equal to the mass flow rate of the particles divided
by the mass per particle and if it is assumed that area of impact is the projection of the
cross-sectional area onto a bend (or a projected area in the path of the flow, such as a
21
restriction) then the overall wastage rate per unit area (i.e. the penetration rate) will be
a function of m(Vi)n times M/m divided by the pipe cross-sectional area A, where M is
the solids production rate. However, M will be proportional to the product of the
solids concentration, S, and the mixed fluid velocity, V. Thus:
E = K x m(Vi)n x M/(m x A)
or
E = K' x (Vi)n x S x V/d2
where d is the pipe diameter, K and K' constants and E the erosion rate.
If a further simplification is made that the particle impact velocity, Vi, equals the mixed
fluid velocity, V (or is a constant proportion of the mixed fluid velocity) and that n=2
then:
E = K' x V3 x S/d2
or
E = K' x V2 x M/d2
This, in essence, is the core form of all of the ‘simple’ erosion models produced by
RCS, API, Tulsa, and Salama & Venkatesh (but not the Harwell model for multiphase
flow), i.e.:
RCS:
E = 4.1 x MV2.5/d2
API:
E = 22.4 x MV2/d2
Salama & Venkatesh:
E = 604 x MV2/d2
Tulsa:
E = 4280 x MV1.73/d2
where M is the solids production rate in g/s, V the mixed velocity in m/s, d the pipe
diameter in mm and E the erosion rate in mm/yr.
As can be seen, the difference between these models lies in the different values of the
constant K' and some variation in the exponent of V.
Although the Salama & Venkatesh, simplified Tulsa, RCS and API approaches are
simple to use, a full understanding of the effect of various parameters such as flow
regime, pipe size and fluid viscosity is only possible by utilising either the full Tulsa
model (SPPS v. 3.0) or the AEA Harwell Model (Sandman version 3.9).
For single phase gas flow the Salama & Venkatesh approach can be used to give an
‘order of magnitude’ indication of the likely wastage rate. This will give the worst
case erosion rates in the absence of liquid buffering at the metal surface and assuming
that the solids remain within the gas stream, i.e. that they do not 'drop out'.
22
Alternatively the more recent Salama model (Ref. 16) can be used to give an indication
of the likely wastage rate. However, in this case it should be noted that an assessment
of the Salama Model undertaken within UTG indicated that it best equates to a 5D
bend situation in comparison with the ‘full’ Tulsa/Harwell models. It is therefore
recommended that it is not used for systems where geometrical features than 5D bends
may be present (e.g. 1.5D elbows, tees, severe constrictions). The model is most
probably suitable for application to downhole completions, although in this instance
care needs to be taken regards regions of significant flow constriction (e.g. insert
valves). For a more detailed consideration of the likely erosion rate the full Tulsa
model (SPPS v. 3.0) should be used.
For single phase liquid flow the full Tulsa Model (SPPS v. 3.0) should be used where
possible. However, given that this is a computer software package that will not be
universally available, the API and/or RCS models can be used for initial assessments
(the latter giving rapid assessment and the former a more accurate assessment based on
bend geometry). These models are based on simple slurry impingement tests and lab-
scale flow loops and may suffer a problem with scale-up to field conditions. However,
they should give rates of the correct order of magnitude. Alternatively the more recent
Salama model (Ref. 16) can be used to give an indication of the likely wastage rate.
However, in this case it should be noted that an assessment of the Salama Model
undertaken within UTG indicated that it best equates to a 5D bend situation in
comparison with the ‘full’ Tulsa/Harwell models. It is therefore recommended that it is
not used for systems where geometrical features than 5D bends may be present (e.g.
1.5D elbows, tees, severe constrictions). The model is most probably suitable for
application to downhole completions, although in this instance care needs to be taken
regards regions of significant flow constriction (e.g. insert valves).
For multiphase flow regimes the situation is more complicated. For regimes very close
to pure gas flow the Salama & Venkatesh or Salama models can be used for ‘order of
magnitude’ estimates (see restrictions on use of the Salama Model in Section 3.1.2.).
However, when there is any appreciable liquid present then this rate will be mitigated,
although the degree of mitigation will depend very strongly on the flow regime
characteristics. For regimes very close to pure liquid flow the API, RCS and/or Salama
models can be used (see restrictions on use of the Salama Model in Section 3.1.2.).
However, where there is any appreciable gas present this will not be appropriate as it is
likely to be non-conservative. For multi-phase flow erosion rates below those for pure
gas but above those for pure liquid flow would normally be expected.
The Harwell programme complemented the Tulsa programme; the latter is based on
fluid flow and modelling and has started with single phase flow conditions while the
Harwell programme was an empirical programme based on multiphase flow conditions.
The major concern with the Harwell programme is that it was based almost entirely on
a 2" test loop and scale-up complications are likely to be present in multiphase flow.
The Harwell programme showed that, even for the same mixed velocities, the erosion
rate depends on the flow regime. The dependence was so strong that the proposed
erosion model was a function of SxV rather than SxV3:
23
E = S x (C1 + C2 x V x √ρm)
where E is the erosion rate, C1 and C2 constants which depend on flow regime, S the
solids concentration, V the mixed fluid velocity and ρm the mixed phase density.
Harwell have developed a computer software program “Design Procedure for Erosion-
Corrosion in Multi-phase Flow, Release 3”. As with the Tulsa software package this
program is not available commercially, but is only available to participants in the Joint
Industry Programme (JIP). BPX was a member of this JIP. The Program enables the
user to determine the flow regime, it then calculates the likely erosion wastage rate
based on the appropriate C1 and C2 values.
For ‘first pass’ assessments of the likely erosion wastage rate the following procedures
can be used:
It is recommended that the flow regime for the intended multiphase duty is firstly
assessed. The following criteria can then be applied:
Annular Flow:
Use the Harwell Release 3 software package to assess the likely erosion wastage rate.
An ‘order of magnitude’ assessment can be achieved using the Tulsa SPPS v. 3.0
software package using the mixed velocity together with (i) the averaged fluid
properties and (ii) the liquid properties. The actual erosion rate should then fall
between these two values.
Bubble/Churn Flow:
Use the Harwell Release 3 software package to assess the likely erosion wastage rate.
Additionally the Tulsa SPPS v. 3.0 software package with mixed velocity and liquid
properties can be used for comparison purposes.
Stratified:
Use the Tulsa SPPS v. 3.0 software package with the liquid velocity calculated for the
hydraulic diameter and the liquid properties.
One flow regime that has not been covered by either the AEA Harwell or Tulsa JIPs to
date is horizontal slug flow. Slug flow is of interest to BP Amoco at a number of
locations, e.g. in Alaska slug flow in large diameter flow-lines is often encountered,
where solids are often present and, indeed, failures have been experienced.
Unfortunately, there is no available data from either the JIP programmes or the
literature in this area and BP AMOCO's own experience is complicated by CO2
corrosion. If erosion is a problem in such regimes then there are two possible solid
impingement mechanisms:
24
• solids on the bottom of a line are picked up and thrown down by a
passing slug but do not get carried forward a significant distance.
• solids are entrained in the slug carried forward and thrown against the
pipe wall by the breaking wave at the slug front.
In both cases solids are unlikely to be carried at velocities exceeding the mixed fluid
velocity. The erosion may be mitigated to some extent as the pipe wall would be
expected to be protected by a significant liquid layer. However, the liquid slug front
will be a zone of considerable mixing and entrained gas, such that the liquid slug front
may approach the homogenous mixture. Therefore, as an interim measure until this
type of flow has been fully investigated, it is recommended that for such instances the
Tulsa SPPS v. 4.0 software package is used with the mixed fluid properties (density
and viscosity) and velocity. For an ‘order of magnitude’ assessment the API or RCS
Models can be used, again employing the mixed fluid properties (density) and velocity.
The use of these models together with the mixture properties and velocity are likely to
give a conservative estimate of the erosion under slug flow, as it assumes that any
point on the pipe wall will be subjected to impingement by a liquid slug front
continuously. Whilst this ‘in built’ conservatism needs to be recognised, it is
considered that this represents the best advice available at this time.
All of the erosion rate models show a dependence of erosion rate on the substrate
material. The programmes that consider alloyed steels show a slight increase in
erosion resistance going from carbon steel to the more highly alloyed materials. This
effect is, however, not marked. Thus, for example, the Tulsa programme has a factor
of 1.5 between carbon steel and annealed 13 Cr steel, with 22 Cr duplex stainless steel
being approximately the same as carbon steel.
The Salama & Venkatesh model has an inverse relation between erosion rate and
carbon steel hardness, and the Tulsa model has the erosion rate proportional to the
hardness to the power of -0.59. Given the range of hardness likely for carbon steel
pipework and tubing neither correction will account for much more than a factor of
two.
Thus, for pure solid particle erosion, the effect of substrate material (when comparing
steel alloys) on the erosion rate is a second order effect of much less importance than
flow regime, mixed velocity or solids content. The effect can be quantified in different
models, but a reasonable 'rule-of-thumb' would be that steel alloy composition does not
have a significant effect on erosion resistance.
In general, the erosion resulting from the impact of a single solid particle is a function
of the momentum of that particle at impact and the total erosion is a function of the
total momentum impacting on a surface. By this reasoning, there could be several
impacts from a large number of small particles or one impact from a single large
particle but, so long as the total momentum was the same, the erosion would be the
same.
25
However, the presence of a barrier layer of liquid at the surface and the bulk flow of
fluids round bends can mean that smaller particles are less likely to reach the surface
than large particles; or, at least, suffer a greater percentage loss of momentum. Thus,
in practice, erosion is likely to be less for smaller, less massive, particles than for large
particles, even if the total solids mass production rate is the same. Only three of the
Models, i.e. Tulsa’s SPPS v.4.0., Harwell’s Sandman v.3.9. and the Salama Model
take any account of the particle size in their calculation of the erosion wastage rae.
In the Harwell project, broadly speaking for the conditions tested (2bara CO2, 30oC)
the erosion-corrosion rate was found to be equal to erosion rate plus the 'unfilmed'
corrosion rate. It is worth pointing out that under the conditions tested (2bara CO2,
30oC) the formation of iron carbonate films (often termed ‘scaling’ in CO2 corrosion)
would not be expected.
The Tulsa programme tested carbon steel in CO2 and sand-containing environments
with 50 psig CO2 at 200 oF (93.3 oC) and at pH 5.0, 5.5 or 6.0 (i.e. conditions under
which the formation of iron carbonate films is likely). Three regimes in the erosion-
corrosion wastage of carbon steel were identified. These were as follows:
(i) 'Scaling Regime'. In this regime the semi-protective corrosion product layer is
retained on the metal surface, affording some protection. This is the normal situation
for solids free conditions, or more benign erosion-corrosion conditions.
(ii) 'General Wastage Regime'. In this regime any scales/surface films are removed
from the metal surface by solids erosion and/or do not have the time to form. Hence
metal wastage as a result of both erosion and corrosion can go on unabated. This is the
normal situation for very aggressive erosion conditions.
(iii) 'Pitting Regime'. In this regime the solid particles prevent scales/surface films
forming at impingement points on the metal surface, whilst scale/surface films form on
the rest of the surface. This leads to pitting damage. Corrosion in the 'bare'
impingement areas can be significantly more aggressive in terms of metal penetration
rate than for general wastage. Some scales/surface films can act as cathodic areas,
significantly accelerating the corrosion rate in the relatively small anodic 'bare'
impingement areas. This occurs at conditions intermediate between 'scaling' or 'general
wastage'. Corrosion rates up to twice that anticipated for ‘un-filmed’ conditions have
been observed.
ECRC have developed a software program (SPPS-EC), which can predict the
threshold velocities for these three regimes. However, at present the model can not
26
predict the likely wastage rate under erosion-corrosion conditions. The Tulsa work has
also indicated that some corrosion inhibitors may be able to increase the threshold
velocities for these three regimes (Ref. 10). However, this effect is not yet sufficiently
well established for use in design. In any event, any such increase is likely to be
corrosion inhibitor and system dependant, meaning that to apply any increase in
threshold velocity to the design would require specific testing of the candidate
corrosion inhibitors under the anticipated system conditions.
It is clear from the above that there is possible synergy between erosion and corrosion
in carbon steel systems. However, the quantification of such effects is difficult. At this
stage it is suggested that no clear velocity thresholds can be established for erosion-
corrosion. As an interim measure the following philosophy is recommended:
If the erosion rate is less than 0.1 mm/yr then there is no need to consider
erosion/corrosion interactions, i.e. the total wastage rate will be the predicted
corrosion rate plus the predicted erosion rate.
If the predicted erosion rate is greater than 0.1 mm/yr, then use the CO2 model (Ref. 2)
to determine the likelihood of iron carbonate scale formation.
For the case where no iron carbonate scale is anticipated the total wastage rate can be
taken as the erosion rate plus the un-filmed corrosion rate (i.e. in line with the
conclusions of the Harwell work).
For the case where iron carbonate scale is anticipated the total wastage rate can be
taken as the erosion rate plus twice the un-filmed corrosion rate (i.e. to reflect the
‘pitting regime’ in the Tulsa work).
In the Harwell programme the 13%Cr steel was found not to corrode at lower
temperatures (30°C) under erosion-corrosion conditions until about 2 µm of material
had been removed by erosion. Thereafter the wastage rate increased to 1 - 2 mm/yr,
remaining at this level even after the sand was removed. The 'corrosion resistant'
properties were only restored once the material had been re-exposed to air. This
observation is in agreement with studies in Sunbury (Ref. 11), which found that at 30°
C in CO2-containing solutions the protective layer never completely reformed. At
higher temperatures (50°C and 80°C) the results from Harwell indicated no synergy
between erosion and corrosion. These results were again supported by data from the
27
Sunbury experiments (Ref. 11), which found that the protective film reformed very
rapidly after damage at temperatures of 80°C and above (temperatures up to 150°C
were tested) in a CO2-containing solution.
If the erosion rate is less than 0.1 mm/yr then there is no need to consider
erosion/corrosion interactions, as it is anticipated that the protective film will not be
destroyed, i.e. the re-filming processes will be faster than the wastage rate. Therefore,
the total wastage rate will be the predicted corrosion rate (if any, see Ref. 12 for
further details) plus the predicted erosion rate.
If the erosion rate is greater than 0.1 mm/yr then the total wastage rate at temperatures
lower than 80°C should be taken as the erosion rate plus the corrosion rate for ‘un-
filmed’ carbon steel in the given chemical environment. For temperatures above 80°C,
the total wastage rate should be taken as the erosion rate plus the corrosion rate
expected on 13%Cr steel (Ref. 12).
In the Harwell work the duplex stainless steel was found not to corrode under the
conditions used, even in the presence of sand. Therefore, it is recommended that in this
case the total wastage rate is taken to equal the erosion rate, i.e. that no allowance is
made for corrosion.
28
References:
2. "A Corrosion Philosophy for the Transport of Wet Oil and Multiphase Fluids
Containing CO2", J Pattinson, ID Parker & AS Green, BP GRE Report No.
ESR.93.ER.013, March 1993.
3. "Erosional Velocity Limits for Duplex Stainless Steel", J Pattinson & J W Martin,
BP GRE Report No. ESR.95.ER.058, July 1995
4. "A Review of Erosion Corrosion in Oil and Gas Production", JS Smart, Paper 10,
NACE Corrosion Conference, 1990
8. "Salt water velocities in pipes; for continuous flow", British Standard MA18, 1976
10. "Erosion/Corrosion Research Center: Advisory Board Report May 11, 1996", E F
Rybicki, University of Tulsa, USA
12. “Guidelines for the Use of 13%Cr Stainless Steels in Chloride Containing Waters
Under Non-Sour Conditions”, DME Paisley, BP GRE Report No. ESR.95.ER.040,
April 1995.
15. “Erosion of Alloy 625 and 25%Cr Duplex Stainless Steel in Water Injection
Service”, memorandum by J W Martin to S Whitehead dated 22nd April 1997.
16. “An Alternative to API RP14e Erosional Velocity Limits for Sand Laden Fluids”,
M M Salama, OTC Proceedings 1998, Paper 8898.
29