OTC 4485 Evaluation of API RP 14E Erosional Velocity Limitations for Offshore Gas Wells by M.M. Salama and E.S. Venkatesh, Conoco /nc.
Copyright
1983 Offshore
Technology
Conference
This paper was presented at the 15th Annual copy is restricted to an abstract of not more
OTC in Houston, than 300 words.
Texas.
May 2-5,
1983. The material
Abstract
la subject
to correction
by the author.
Permission
to
disturbed due to a local change in direction, a velocity component normal to the pipe wall will be introduced, resulting in repeated impacts on the pipe wall. Erosion damage of the pipe is caused by the repeated bombardment of liquid and solid particles. The erosion damage is enhanced by increasing the production capacity of a given flow systern (i.e., increasing flow velocity). In order to avoid potential erosion problems, most oil companies have been limiting their production rate by reducing the flow velocity to a level below which it is believed that erosion does not occur. This limiting flow velocity is calculated using the API RPI 4E recommended empirical equation:l
In order to avoid erosion damage and associated problems in two-phase flow systems, API RP14E reconmends limiting the maximum production velocity to a value defined by the following empirical equation:
Ve= c/* where V. = the maximum allowable erosional velocity in ft/sec P = the density of fluid in lb/cu ft at flowing conditions of temperature and pressure C = a constant generally known as the C factor, is in the range of 100 to 125
Ve=clfi
. . . . . . . . . . . . ...(1)
where
Evaluation of the above equation has shown that in cases where the form of the equation can be rationalized, the value of C, as recommended by API, is extremely conservative. These cases include erosion due to liquid particle impingement and corrosion-assisted erosion due to the stripping of corrosion inhibitor films. For the case of erosion due to sand particles entrained in the fluid, the form of the equation appears to be incorrect. The API RP14E recommendations to reduce the value of C to account for sand in the produced fluid is, therefore, improper.
V. = the maximum allowable erosional velocity (fVsec) = the fluid density (lbs/cu ft) ~ = an empirical constant generally known as the C factor For a sand-free, two-phase flow situation, the C factor is limited to 100 for continuous flow and 125 for intermittent flow. The API RP14E recommends the use of a lower unspecified C factor for fluids containing sand. Previously, production rates were generally restricted to lower values than those specified by Equation (1) due to fear of formation damage or excessive sand production. However, the recent advances in well completion procedures, sand control techniques, and our understanding of reservoirs’ flow behavior permit higher production rates. These developments, in addition to the current economic incentives, are motivating the oil industry to increase production rates, particularly for high-capacity gas wells. Under this condition, the API RPI 4E erosional velocity equation represents a major obstacle. Although the API equation has been widely accepted, the authors were un-
A method for calculating erosion damage as a function of fluid and flow characteristics is proposed. This approach can be used to calculate a limiting flow velocity for any specified allowable erosion rate. Introduction The production of hydrocarbons from underground reservoirs is associated mainly with the flow of a liquid (oil and water), gas (natural gas), and/or solid (sand). This flow situation is essentially one of a liquid-gas, two-phase flow with entrained solid particles. When the fluid flow in a pipe is
successful in all attempts to determinethe basis of this References and illustrations at end of paper.
equation. These were the main reasons for initiating this -- .
3{1
—
EVALUATIONOF THE API RP14EEROSIONAL
2
\/Fl .GLWW,
CWITV
,
!
I-00.,lMITATlnN.~ ,0 . . ..-
..-
Ff)R ..
study. This paper summarizes the initial phase of this study. Several areas which require further evaluation are identified.
.
nFFSHORF -. . . . . .
.
.-
WFI-LS ..——
nr~ “ ,
-
“
A
-r-
drop range for high capacity wells is 3,000 to 5,000 ps These numbers correspond to a value for the constant C the range 80 to 100. Although there is a very close similarity between the Bernoulli relationship (Equation (2)) and th API empirical criterion (Equation (1)), they should have n correlation because they represent two completely differen phenomena.
Erosion Mechanisms Erosion is defined as the physical removal of material from the surface. This is different from corrosion, which involves material removal by chemical or electrochemical reaction. The material removal by erosion is caused by one or more of the following:
For the case of erosion due to liquid impingement on surface, the relationship between flow velocity, V (ft/see) and erosion rate, h (roils per year), can be written as follows (Appendix 1):
1. Cavitation (bubble collapse) 2. Liquid particle impingement
v
=
B h1”6 7“”””’”””””””””””(3)
3. Solid particle impingement where Erosion damage occurs as a result of one of three mechanisms:
P B
1. Fatigue due to repeated loads induced by a bubble collapse or particle impingement.
For most practical cases and allowing for a 10-mii-per-year erosion rate, Equation (3) reduces to:
2. Abrasion due to repeated impingement of hard particles on ductile material.
V=%””””””””
3. Corrosion-assisted erosion due to the breakage of the protective surface layer either by fatigue or abrasion. Possible Rationalization Velocity Equation
The value of C under these conditions is much higher than that recommended by the empirical equation of AP RP14E.
of API Erosional
Examination of the API erosional velocity equation suggests that it may have been derived using one of the following approaches:
The velocity limitation imposed by Equation (4) is ver stringent when compared with experimental data obtained to date for liquid impingement erosion. For example, Equa tion (4) gives a limiting velocity of 38 ft/sec for water im pingement which if far lower than experimentally deter mined threshold velocities shown in Table 1 during wate impingement erosion tests. The threshold velocity is de fined as the velocity below which no measurable erosion damage occurs after a large number of impacts (106- 10 impact). In most of these experiments the erosion is caused by the multiple impacts of a water jet on specimens mounted on a rotating disk. Since erosion is generally considered as analogous to fatigue,7 the threshold velocity corresponds to the endurance limit. All these experimentally determined threshold velocites*7 are higher than thos predicted by Equation (4). Therefore for a sand-free pro duction system, an increase in the C factor from 100 (a recommended by API) to a value around 300 should no pose any operational problems.
1. Constant pressure drop limitation using Bernoulli relationship. 2. Limitation on erosion rate due to liquid impingement. 3. Limitation on velocity inhibiting layers.
to avoid removing
corrosion-
The Bernoulli relationship can be written (for the case where the gravity effect and initial fluid velocity are ignored) as follows:
“Jzm_
~-~
c
= the fluid density (lbs/cu ft) = a constant which depends on the target materia hardness and critical strain to failure.
. . . . . . . . . . ...(2)
where V
= the maximum flow velocity = the fluid density ~P = the total pressure drop along the flow path
If we consider the case where the limiting velocity governed by the stripping of the protective inhibitor film from the surface of the steel tubulars, the limiting velocity can b expressed by:
The total pressure drop along the flow path (AP) is composed of four components. These are the pressure drops in the reservoir, across the completion, along the production tubing, and across any restriction. A typical total pressure 372
OTC 4485 where T f
3
MamdouhM. Salamaand EswarahalliS. Venkatesh velocity near the tip of a bubble is about equal to the local relative velocity between the two phases. Similar velocities must occur in the roll waves of annular flows. It is very difficult without appropriate experiments to identify which of the above three possibilities is most significant and, hence, what is the value of the radial velocity.
= the shear strength of the inhibitor interface = the friction factor
This equation is derived by equating the flow induced shear stress at the pipe wall with the shear strength of the inhibitor. For most practical cases, ~ equals 8,000 psis and f equals 0.0015,9 Equation (5) can be written as:
‘=
Rabinowiczl 1 has shown that experimental results of erosive damage of ductile metals due to solid particle impingement agree reasonably well with the following erosion
F””””””””””””””””(E)
‘ate ‘quation: “_KWV2~ gP
Equation (6) has the same form as the API empirical Equation (1). The value of C based on this criterion is far higher than that proposed by the API equation. It is clear from the above discussion, therefore, that the current API erosional velocity limitation seems to be extremely conservative.
““””””””””””””
. (7)
where
Proposed Erosional Velocity Equation It appears that erosion will occur in a solid-free fluid flow system only at very high velocities, which would not be allowed in a properly designed system because of severe pressure drops. In process piping, a velocity limitation of about 100 feet per second is used. Above this velocity, it becomes more economical to increase the pipe diameter
U W V P 13
= = = = =
the volume of metal eroded the total weight of impinging solid particles the particle velocity the penetration hardness of the target material a coefficient which depends on the impingement angle. It equals 1.0 for angles between 10 and 60 degrees and 0.50 for other angles. 12 For the fully developed turbulent flow system, as in the case of a producing well system, p is appropriately chosen as
K
= a nondimensional erosive wear coefficient. Rabinowicz12 has shown by statistical analysis of the experimentally determined K values that the mean value is 0.0103. A reasonable value for K in the case of the producing well system can be based on mean plus two standard deviation. This value for K is 0.071. = the gravitational constant (32.2 ft/sec2)
0.75.
than to increase the pumping capacity. It is therefore the authors’ belief that erosion in the oil industry is mainly due to sand particles entrained in the produced fluid. Although estimating the quantity of sand produced from a gas or oil well appears to be straightforward, the quantification process is somewhat more complicated in practice. Typically, sand production is extremely erratic. Most producers seek to limit sand production to onl a few pounds per day per well, perhaps 5 to 10 pounds.l r Wells that produce this level of sand are sometimes characterized as sand-free wells.
g
For the case of flow in pipes, the maximum erosion is expected to occur in elbows. As a conservative first approximation, the eroded area is considered to be equal to the projected area of the pipe and the particle velocity equals the average flow velocity (V). The amount of impinging solid particles (W) on the surface of the elbow is a percentage of the total particles in the flow. Griffith13 has estimated that for a low-density gas system, this percentage is about 100 percent, and as density increases, the percentage is reduced to a limiting value of 30 percent for liquids. This is rationalized because in high-density fluids, most of the particles will be carried in the stream in the center of the flow without impacting the surface of the elbow. This observation is interesting because it indicates that as the density increases, the amount of impinging particles decreases, thus decreasing erosion rate and, therefore, increasing the allowable velocity. This is opposite to what is implied by API erosional velocity equation. For a two-phase flow system, the ratio of the weight of impinging particles to the weight of all the particles in the flow is, therefore, between 0.3 and 1.0. A reasonable ratio is 0.65 for a medium-density twophase fluid, as in the case of a gas well system.
The presence of sand particles in the produced fluid results in erosive damage by abrasive wear mechanism. However, in order for the sand being conveyed by the mixture of gas and oil to cause damage, it is necessary for the sand to acquire a velocity normal to the pipe wall. This velocity can be attributed to three sources: 1. Turbulent fluctuation in the flow. 2. Secondary flows in the vicinity of bends and fittings. 3. Radial two-phase velocity fluctuations. Radial transport-turbulent fluctuations range up to about 10 percent of the main flow. The velocity of particles depends on the size, but it is generally less than the fluctuation velocity. Radial velocity due to secondary flows is important. The location of the maximum wear in bends and sometimes the peculiar wear patterns in the wake of protuberances and orifices can only be accounted for by
Using the above analysis in Equation(7), rate due to flow in elbows can be given as:
secondary flow. Due to the structure of the flow, two-phase flows have strong radial velocities. Fora slug flow, the radial 373
the erosion
EVALUATIONOF THE API RP14E EROSIONAL
4
VELOCITY
~=
K(0.65W)V2 gP(m/4dz)
(~)
LIMITATIONS
FOR OFFSHORE
●A.
mpy) barrel of sand = 945
ft/sec2) units
Conclusions 1. Erosion damage in gas-producing wells occurs primarily due to solid particle impingement.
The value of A in the above units is calculated to be: A=
I.36
x106
2. API RP14E erosional velocity equation is extremely conservative for sand-free production conditions.
. . . . . . . . . . . . ...(9)
for p = 0.75 and K = 0.071, Equation (8) reduces to: h=l.86x105~
OTC 4
Although Equation (7) offers a sound theoretical basis to assess the problems of erosion in pipes, it is clear that several assumptions have to be made to derive design equations, such as Equation (12). An experimental program should be undertaken to establish the rationalization of these assumptions and verify the values of their corresponding parameters. Such an experimental program should also address the effect of sand erosion on the formation of a protective inhibitor film which is necessary to avoid corrosion problems in corrosive wells.
““”””””””(8)
where h = erosion rate (roils per year, W = sand flow rate (bbl/month; pounds) V = fluid flow velocity (ft/see) P = hardness (psi) d = pipe diameter (inches) g = gravitational constant (32.2 A = correction factor for proper
WELLS
3. Appropriate equations have been proposed for use in design against erosion for both sand-free and sandproducing wells.
-. . . . . . . . ..(10)
Acknowledgements Comparison between the predicted erosion rate using Equation (1O) and experimental data developed as part of API -OSAPR- project 2 on sand erosion by Texas A&M University Research Centeri4 is shown in Table 2. The results illustrate the validity of Equation (1O) in predicting sand erosion rates. On the average, Equation (1O)overestimates the erosion rates by a factor of 1.44.
The authors would like to thank management of Conoco Inc. for permission to publish this paper. The authors sincerely express their appreciation to Professors Ernest Rabinowicz and Peter Griffith of Massachusetts Institute of Technology for their valuable suggestions and for giving permission to reference some of their unpublished work. The authors would also like to thank Messers John Wolfe and Fred Gipson and Dr. Richard Vennett for their assistance and advise.
The result of AP1-OSAPR-project 214 also showed that for flow infield ells and tees, the erosion rates are about 50 percent that in elbows and, therefore, Equation (1O) can be written as follows: h=93,000$$
..:.;
References ‘API RPI 4E, “Recommended Practice for Design and Installation of Offshore Production Platform Piping System,” Third Ed., December 1981.
. . . . . ..(11)
Substituting the value of P for steel (P = 1.55 x 105 psi) and assuming that erosion velocity is based on an erosion rate of 10 roils per year, Equation (11) can be written as:
‘+$
2Thiruvengadam, A.; Rudy, S. L.; and Gunasekam, M. “Experimental and Analytical Investigation on Liquid impact Erosion,” Characterization and Determination o Erosion Resistance, ASTM STP474, p. 249, (1970).
““”””””””””””””(’2)
3Hancox, N. L., and Brunton, J. H., “The Erosion of Solids by the Repeated Impact of Liquid Drops,” Phil. Trans. Roy. Sot., London, Vol. 260A, p. 129, (1966).
When W approaches zero, the value of V is limited by Equation (4), which is for sand-free system. For a pipe with 3-inch diameter, the erosional velocity V (ft/see) can be given as a function of the rate of sand production W (barrels per month) as follows:
4Baker, D. W.; Jolliffe, K. H.; and Pearson, D., “The Resistance of Materials to Impact Erosion Damage,” Phil Trans. Roy. Sot., London, Vol. 260A, p. 168, (1966).
v=%““””””””’””””””(’3)
5Hobbs, J. M., “Factors Affecting Damage Caused by Liquid Impact, National Engineering Laboratory Report No. 266, December 1966.
Equation (13) is reasonably conservative and, therefore, could be used as a design criterion. Allowable velocity, a function of sand production as calculated by Equation (13), is shown in Table 3 for sand containing fluid.
6Vater, M., “Prufung und Verhalten Metallischer Werk Staffe Gegen Tropfenschlagund Cavitation, “Korrosion and Metallschutz, Vol. 20, No. 6, p. 171, (1944).
a-,
s/q
.
ITC
4485
5
—.Venkatesh ..—------ M.. Salama ——. —...—and —.. —Eswarahalli S.
Mamrkmh
where
7Heymann, F. J., “A Survey of Clues to the Relationship Between Erosion Rate and Impact Parameters,” Proc. of the 2nd Meersburg Conference on Rain Erosion and Allied Phenomena, Royal Aircraft Establishment, U.K., p. 683, (1968).
U v
wear volume rate impacting fluid volume rate fluid density (lb/ft3) 1’ impact velocity (ft/see) P tar et material hardness, psi (for steel, P = 1.55 x 10 .9psi) cc = Critical strain to failure (0.1 O for steel) g = gt13Vih5ttiOrtal constant (32.2 ft/sec2) K = high-speed erosion coefficient (= 0.01)
8Kemball, C., “intermolecular Forces and the Strength of Adhesive Joints,” in the Proceedings of the Symposium on Adhesion and Adhesives Fundamentals and Practice, Cleveland, Ohio, p. 69, (1954). ‘Fox, R. W., and McDonald, A. T., “introduction to Fluid Mechanics,” John Wiley and Sons Inc., New York, 1973.
= = = = =
Considering
the case where v is given as V=AV
10Estimating Sand Production Handbook, O. 1. Corporation, Houston, Texas, 1982.
. . . . . . . . . . . . (Al -2)
and erosion depth h is given as
11Rabinowicz, E., “The Wear Equation for Erosion of Metals by Abrasive Particles,” Proc. Fifth Int. Conf. on Erosion by Liquid and Solid Impact, Cambridge, England, p. 38-1, (1979).
h=:
. . . . . . . . . . . . . . . (Al -3)
where A is the cross-sectional
12Rabinowicz, E., “Factors Modi~ing the Erosive Wear Equations for Metals,” Israel J. Tech., Vol. 18, p. 193, (1980).
By substitution,
Equation (Al-1)
area of the pipe. becomes:
h=~~~2
V“’7=%+
E. (1982), Private Com-
APPENDIX
”””(A’-5)
1 V=v
Erosion rate due to a liquid impingement can be calculated as follows:i 5
2Pg
““”””
This equation can be simplified by substituting the value of P for steel and by accounting for V1’6 into the numerical constant as follows:
EROSION DUE TO LIQUID IMPINGEMENT
U=W2
)
By substituting the above values for K, g, and EC,Equation (Al -4) can be given as:
14Weiner, P. D., and Tone, G. C., “Detection and Prevention of Sand Erosion of Production Equipment,” API OSAPR Project No. 2, Research Report, Texas A&M University, College Station, Texas, 1976. 15Griffith, P., and Rabinowicz, munication.
. . . . . ..(A1-4) (
13Griffith, P. (1982), Private Communication.
2PV2 2 (27 gP ~2C)
““””” ”””’” ””(A1-6)
Assuming that the allowable erosion rate h is 10 roils per year, the above equation becomes:
. . . . .. (Al-l) V=*
_— -
375
““”””
”””””
””””
(A’-7)
s3i?i?ei?l?i?i2i?i?8s8. . . 8 .
mlcucNc9d-lnlnmul
