Vous êtes sur la page 1sur 16

USMS

022057 Are Improved Surge Models Needed?


M.S. Bizanti, Louisiana Tech U.; R.F. Mitchell, 5847 San
Felipe; R.E. Leturno, Areo Oil & Gas Co.

Copyright I 'I'll Society of Petroleum Engineers


This manuscript was provided to the Society of Petroleum Engineers for distribution
and possible publication In an SPE journal. The material is SUbject to correction
by the author(s). Permission to copy is restricted to .an abstract of n01 more than
300 words. Write SPE Book Order Dept., Library Technician, P.O. Box 833836,
Richardson, TX 75083-3836 U.S.A. Telex 730989 SPEDAL.
AREIMPROVED SURGE MODELS NEEDED?

BY
M. BlZANTI, LOUISIANA TECH UNIVERSITY
R. LETURNO, ARCO OIL AND GAS COMPANY
R. MITCHELL, ENERTECH ENGINEERING RESEARCH
SPE 22057

The Authors are thankfull to· :Sherman Bradley, Lindsey Scott, Steve
Gardner, Yih-Min Jan, Bob Siegfried, and Mark Stracke all from Arco
Oil & Gas Company for their review of our paper.
SPE 22057

. ARE IMPROVED SURGE MODELS NEEDED?

ABSTRACT

It is generally accepted that pulling and running pipe causes pressure surges. The
pressure surge prediction is of economic importance in wells where the pressure must be
maintained within narrow limits to prevent lost circulation and formation fluid influx. In
these situations, the drilling engineer needs an accurate method of calculating surge
pressure.

Surge/swab pressures are commonly calculated using steady state flow models.
Recently, a fully dynamic surge model became available. In this paper, sensitivity studies
are used to quantify the differences between the two models for variations in fluid plastic
viscosity, fluid yield point, fluid density, and pipe depth.

A comparison of a conventional steady-state surge model with a dynamic surge


model identified two cases where improved surge calculations are needed:

1. Steady-state surge models over-predict surge pressures for high yield point muds,
and

2. Steady-state surge models under-predict surge pressure for high mud weights in
deep wells.

The comparison cases were swabs run with relatively high viscosity muds to
minimize dynamic effects. For most of the cases studied, the two models were reasonably
consistent. However, for high yield points, pipe elasticity reduced swab pressures to an
important degree. For high mud weights, fluid inertia resulted in higher swab pressures.

INTRODUCTION

Pressure surges have long been known to cause well control problems, e.g., Cannon
in 19341 and Goins, 1951. 2 In most wells, casing design and mud programs leave large
surge pressure safety margins. In critical wells, however, pressure surges must be
maintained within narrow limits. Also, some operations are particularly prone to large
pressure surges, such as running low clearance liners in deep wells.

Pressure surges in critical wells are commonly estimated using steady flow surge
models, such as those proposed by Burkhardt,3 Fontenot and Clark,4 and Schuh.5 In these
models, the drilling mud is perfectly displaced by the pipe motion. Fluid pressures are
calculated to be consistent with non-Newtonian frictional pressure drops due to this fluid
motion. These models neglect fluid inertia (Burkhardt includes an approximate inertia

1
SPE 2205 7

effect), the compressibility of the fluid and wellbore, and the axial elasticity of the pipe.
The lack of fluid compressibility is a conservative assumption because it predicts a higher
flow rate which generates a higher frictional pressure drop. The neglect of fluid inertia
is not a conservative assumption because, for example, negative pressure surges may occur
due to fluid backflow when the pipe is brought to rest. Axial elasticity of the pipe
reduces the rate of fluid displacement, so rigid pipe displacement is considered
conservative.

Fully dynamic surge models have been developed because of the uncertainty in the
steady state model assumptions. The first fully dynamic surge pressure model was
developed by Lubinski et al. 6 This model emphasized the importance of compressibility
in pressure calculations. Lal has corrected a number of deficiencies in the Lubinski model
and begun an investigation of parameters affecting surge pressures.7,8 Both Lubinski and
Lal assumed rigid pipe displacement. Mitchell added the effect of pipe axial elasticity to
dynamic surge analysis. 9

There are three principal differences between a steady state surge calculation and
a dynamic surge calculation:

a. Compressibility - The wellbore fluid, tubulars, and formation are deformable in the
dynamic surge model. This allows the possibility of storage effects. For instance,
where the viscous forces are extremely high, the surge pressure would be more related
to the water compression and wellbore expansion than to the steady state frictional
pressure drop.

b Tubular Elasticity and Inertia - The moving pipe in the dynamic surge model is treated
as a fully dynamic elastic system. In general, pipe displacement at the surface does
not equal pipe displacement down hole. This allows the possibility of reduced surge
pressures because surface movement is stretched over time downhole by the pipe
elasticity, giving lower pipe velocities. Pipe dynamics can increase surge pressures
when downhole pipe speed oscillates about the surface pipe speed.

c. Fluid Inertia - The force needed to accelerate the fluid may be comparable to the
viscous forces and increase surge pressures temporarily. At the end of the pipe trip,
forces are needed to decelerate the fluid, which may generate swab pressures at the
end of the surge. The effects of inertia can be seen in the pressure measurements of
Burkharde and Clark and Fontenot. lO •

Fully dynamic surge models are more complex and require more computer
resources than steady state surge models. Is this additional complexity justified? On the
other hand, are steady state models sufficiently accurate? In this paper, sensitivity studies
are used to quantify the effects of fully dynamic surge predictions.

2
SPE 22057
SURGE FIELD DATA

Limited dynamic surge field data is available. Burkhardt presents surge data from a
specially instrumented 2100 ft test we1l3, which provides a good test for analytical
models, but does not represent a real well situation. Useful data was gathered by Clark
and Fontenot10, who ran surge and circulation tests on two wells. The first was an 18,500
ft well plugged prior to abandonment in Mississippi. This well was completed with a 7"
liner, and the drilling fluid was a 17.5 ppg oil base mud. The second well was·a 15,270
ft well in Utah. This well was completed with a 5" liner and the drilling fluid was a 14.2
ppg water base mud. Clark provides complete information on drilling fluid properties
throughout the tests, and full information on pipe motion and resulting surge pressures.
Increased use of MWD tools may provide additional surge pressure data in the future.
Ramsey et al., provide an example of surge pressures recorded by MWD in an 11,954 ft
well with 9.45 ppg mud. l l

MODEL FORMULATION OVERVIEW

The dynamic surge model in this study is based on the work of Mitche1l9, and has
the following features:

a. The full balance of mass and balance of momentum for pipe flow and annulus flow
are solved.

b. Pipe and annulus pressures are coupled through pipe elasticity. Annulus pressures
caused by pipe pressures may be significant.

c. Longitudinal pipe elasticity, pipe inertia, and fluid viscous forces determine pipe
displacement. The velocity of the pipe end is not necessarily equal to the velocity
imposed at the surface.

d. Frictional pressure drop is solved for laminar flow in an annulus with a moving pipe
for power law fluids. 12Turbulent flow frictional pressure drop uses the Dodge and
Metzner friction factor for power law fluids. 13

e. Fluid properties vary as a function of pressure and temperature. Plastic viscosity and
yield point can vary significantly with temperature.

f. Formation elasticity, pipe elasticity, and cement elasticity determinine the composite
elastic response of the wellbore. For a pipe cemented to the formation, use of only
the pipe elasticity will not give conservative surge pressures.

3
SPE 2205 7

The dynamic model is solved subject to a comprehensive set of force and


displacement compatibility relations:

a. The elastic force in the moving pipe is equal to the pressure below the pipe times the
pipe end area. This means that a sufficiently high pressure below pipe could retard
the pipe motion.

b. Mass flow balances are calculated for flow through the pipe nozzle, flow through the
annulus return area, and flow into the pipe-to-bottomhole region.

c. Pressure drops are calculated through the pipe nozzle and annulus return area based
on cross sectional area changes with appropriate discharge coefficients.

d. Boundary conditions for floats were chosen to allow one-way flow through the float.
F1uid is allowed to flow out of the float, otherwise the float is treated as a closed pipe.

Surface boundary conditions set the fluid pressures in the tube and the annulus to
atmospheric pressure. The bottom hole boundary condition assumes a rigid floor, requiring
a zero fluid velocity.

The interpolated method of characteristics is used to solve the dynamic fluid model
and the elastic pipe model. 14 A detailed discussion of the model formulation is given in
Reference 9.

SWAB PRESSURE CALCULATIONS COMPARISON, STEADY AND UNSTEADY


STATE MODELS

A swab pressure problem from Bourgoyne et a1. 15 with the following data, was used
for comparing the results of steady and unsteady state models.

Problem Input Data

Bingham plastic viscosity = 60 cp


Yield point = 10 Ib/100 ft2
Hole Size = 7.875 in.
Drilling string = 15,000 ft.(14,300 ft drill pipe - 4.5"x3.826")
= 700 ft drill collar - 6.25"x 2.75" )
Bit nozzles = 3 x 11/32 in (7.875 in bit)
Mud Weight = 10 ppg
Maximum pipe velocity = 4 ft/sec

4
SPE 2205 7

Effect of Plastic Viscosity on Swab Pressure

Figure 1 shows the results of varying the plastic mud viscosity from 20 to 50 cp
on swab pressures. The calculations were conducted using 2, 4, and 6 ft/sec pipe pulling
velocity.

The results indicate higher swab pressure as plastic viscosity increases. The two
models compare reasonably well, especially at higher velocities. The differences may be
attributed to the use of different fluid models of the laminar flow pressure drop,
particularly flow in the annulus with the moving pipe. The turbulent flow correlations
appear to be nearly the same, giving good correspondence at higher velocities.

Due to the high plastic viscosity range for a 10 ppg mud, the unsteady state surge
model results probably are overdamped, i.e., inertial effects are dominated by viscous
effects, and peak pressures should approximate steady state pressures.

Effect of Yield Point on Swab Pressure

Figures 2 shows the results of varying the yield point from 10 to 50 Ib/l00 fe as
the pipe pulling velocity changes from 2 to 4 ft/sec. The results indicate that the higher
the yield point, the higher the swab pressures generated. The results of the two models
correlate well at low yield point values. Lower swab pressure is obtained with the
unsteady state model at yield point values greater than 30 Ib/100 ft2 • The yield point
range appears high for a 10 ppg mud. The dynamic swab predictions fall below the
steady state results for yield points above 20 Ib/lOO £12. This may be due to pipe
elasticity. Pipe downhole velocity is reduced due to the high pipe drag forces associated
with the high yield point values.

Effect of Mud Density on Swab Pressure

Figures 3 shows the results of varying mud density from 10 to 18 ppg while
changing pipe pulling velocity from 3 to 5 ft/sec. The two models yielded close results.
At lower mud densities, the steady state model yielded higher swab pressures. At higher
densities the unsteady state model was more sensitive and resulted in higher swab
pressures. Also, lower mud densities have a minor effect on swab pressures, especially
at lower pipe velocities. The dynamic surge model shows increased swab pressures with
increased density and velocities. The pressure increase is due to the increased inertia.
However, most of the density effects are contributed to pressure drop through the bit
nozzles, which is proportional to mud density.

5
SPE 2205 7

Effect of Depth on Swab Pressure

Figure 4 shows swab pressures obtained by steady and unsteady state models at
various well depths as a function of pipe velocities. At shallow depths, the steady state
model yielded slightly higher swab pressures than the unsteady state model. As depth
increases, the unsteady state model tends to yield higher swab pressures, especially at· o

higher pipe velocities. The difference in swab pressure predictions are due to different
fluid friction models incorporated in the steady and the unsteady state models.

CONCLUSIONS

For the special case of high viscous forces relative to inertial forces, the dynamic
surge model and the steady flow surge model are generally consistent (the models will
disagree where inertial forces are important, but dynamic effects are not considered in this
paper). It is reasonable to assume that for cases where the model results do not agree, the
differences are due to special assumptions used to develop the steady flow surge model.

For wells with high viscous forces, the dynamic model predicts lower surge·
pressures than the steady flow surge model. The lower pressures in the dynamic model·
are due to the elastic response of the pipe. Because the pipe can elongate in response to
the viscous forces, the pipe velocity at the bit will be lower than the pipe's velocity at the
surface, resulting in lower surge pressures. The rigid pipe assumption used in the steady
flow model is conservative in the sense that predicted surge pressures will always exceed
actual surge pressures; however, this may place unnecessary restrictions on trip speeds.

The dependance of density, plastic viscosity, and yield point on pressure and
temperature are included in the dynamic model but not considered in the steady flow
model. Improved measurement of these properties in high temperature-high pressure
viscometers and improved pressure-temperature correlations 16- 18 make these property
variations desirable in any surge model, because of the strong effect they have on surge
pressures.

In deep wells with high mud weights, the influence of the inertia of the mud
column cannot be neglected, even in high viscous forces. Not only is the surge pressure
increased by the force needed to accelerate the mud column, but opposite pressure is
needed to stop the mud column when the pipe stops. Thus, surge pressures may be
generated after a swab and swab pressures after a surge due to fluid inertia. While
attempts have been made to include inertial forces in steady flow models [e.g., Ref. 3],
a fully dynamic model is necessary to properly account for inertial effects.

The need for improvements in steady flow surge models has been determined by
this comparison between a standard steady flow surge formulation and a fully dynamic
surge model. Many of the novel features of the dynamic model could.be included in an

6
SPE 2205 7

enhanced steady flow model; however, the full dynamic model is recommended for critical
wells because of its greater overall accuracy.

ACKNOWLEDGEMENTS

The authors are grateful to ARCO Oil and Gas Company and Enertech Engineering
and Research Company for granting permission to publish the article.

REFERENCES
1. Cannon, G.E., "Changes in Hydrostatic Pressure Due to Withdrawing Drill Pipe
from the Hole", Drill. and Prod. Prac., API (1934), pA2.

2. Goins, W.e., Jr., Weishert, J.P., Burba, J.L., Jr., Dawson, D.D., Jr., and Teplitz, A.
J., "Down-the-Hole Pressure Surges and Their Effect on Loss of Circulation", Drill.
and Prod. Prac., API (1951), p. 125.

3. Burkhardt, J.A. "Wellbore Pressure Surges Produced by Pipe Movement," J.Pet.


Tech (June 1961), p. 595-605.

4. Fontenot, J.E. and Clark, R.K., "An Improved Method for Calculating Swab and
Surge Pressures and Calculating Pressures in a Drilling Well", Soc. Pet. Eng.
J.(October 1974), p. 451-462.

5. Schuh, F.J., "Computer Makes Surge-Pressure Calculations Useful", Oil and Gas
Journal, (August 1964), p. 96.

6. Lubinski, A., Hsu, F.H., and Nolte, K.G., "Transient Pressure Surges Due to Pipe
Movement in an Oil Well,"Revue de LInst Franc du Pet (May-June 1977), p.307-
347.

7. Lal, Manohar, "Surge and Swab Modeling for Dynamic Pressures and Safe Trip
Velocities," Proc. 1983 IADC/SPE DrIng Conference, (New Orleans 1983), p 427-
433.

8. Lal, Manohar, "Analysis of Factors Affecting Surge and swab Pressures", presented
at the 1984 IADC/SPE Drilling Technology Conference, (Houston 1984), p 231-
244.

9. Mitchell, R.F., "Dynamic Surge/Swab Pressure Predictions", SPE Drilling


Engineering, (September 1988), p. 325-333.

7
SPE 22057

10. Clark, R.K. and Fontenot, J.E., "Field Measurements of the Effect of Drillstring
Velocity, Pump Speed, and Lost Circulation Material on Downhole Pressures",
presented at the 49th Annual Fall Meeting of the SPE of AIME,(Houston 1974).

11. Ramsey, M.S., Robinson, L.H., Miller, J.F., and Morrison, M.E.,"Bit Hydraulics:
Net Pressure Drops are Lower Than You Think, "World Oil,(October 1983), p65-67.

12. Savins, F.J., "Generalized Newtonian (Pseudo-plastic) Flow in Stationary Pipes and
Annuli", Pet. Trans. AIME, (1958).

13. Dodge, D.W. and Metzner, A.B., "Turbulent Flow of Non-Newtonian Systems",
AIChE J., (June 1959).

14. Wylie, E.B. and Streeter, V.L., Fluid Transients, (Ann Arbor: FEB Press 1982,
Corrected Edition 1983).

15. Bourgoyne, A.T., Jr., et aI, Applied Drilling Engineering, Soc. Pet. Eng.,
(Richardson 1986).

16. Peters, E.J., Chenevert, M.E., and Chunhai Zhang.,"A Model for Predicting the
Density of oil Muds at High Pressures and Temperatures," SPE 18036, 63rd
Annual Technical Conference and Exhibition of the SPE., Houston,(1988) pp 197-
207

17. Alderman, N.J., Gavignet, A., Guillot, D., and Maitland, G.C.,"High Temperature,
High Pressure Rheology of Water-Based Muds," SPE 18035, 63rd Annual
Technical Conference and Exhibition of the SPE., Houston, (1988) pp 187-196.

18. Houwen, O.H. and Geehan, T.,"Rheology of Oil-Based Muds," SPE 15416, 61st
Annual Technical Conference and Exhibition of the SPE., New Orleans (1986).

8
SPE 2205 7

81 Metric Conversion Factors

cp x 1.0* E-03 = Pa.s


ft x 3.048* E-01 = m
Ibm/gal x 1.198264 E+02 = kglm 3
In x 2.54* E+OO = cm
psi x 6.894757 E+OO = kPa

* Conversion factor is exact

9
"
seE 2205 7

s
W 1000
A
B (')
-I
p BOO t-

R
E 600 f--
S ')(

S ~
..1.

u 400 c-
I

R .
E
200 - - -
1"1
,
p
S o I I

I
G 20 30 40 50
P~ASTIP VlaCQSITY, cp
ST.STATE(2FpS) -r- ~T,&TATF(4FrS) -r ~T.STATJ:(6FPS)

-8- DYN.(2FPS) --7r PYN,(4fPS) ~ PYN.(QfPS)

Fig. 1- Swap pressure V~, plastic


viscosity (cp)
~

SPE 220 Ij 7 .,'

SWAa PRESSURE, PS'~


1400 I I

1200

1000

800

600

400~ s- -

200
0 1 I I , I

10 20 ~O 40 50
YIELD POINT - ~~/100 fT sq
-.- ST .ST EAT E(2F PS) -+- ST·~TATF(~fra) -r ~T.aTATt=(4Fra)
-8- DYN·(2FPS) -x- PYN.(3fPS) -+- PYN·(4FPS)

Fig. 2- SW&t:l pre~aur~ V~~ yi~lq point


,
SPE 22057
'. .

SWAB PRESSURE, pSIa


1000, I

800

BOOJo ==" ~ *
400~ =====;;====~W~=l' ,

200

0
10 12 14 1ft 18
PFN~'TY, rr~

ST .STAT E(3f pS) -t- ST.~TATt=(ifr&) -r ~T.STATt:(pFr~)


-8- DYN.(3FPS) -?f- PYN.(ifPS) ~ PYN·(5FPS)

Fig 3- Swab pr(:)~~~r~ V~, m~~ q~nsity


,.
SfE 22057'
f ,,~

SWAa PRJ:SSURE, PSI~


1000

800 -

v
- ~

~
600 f--
'r
,
400 ~~ ...,
f-

lfI

200 -

o I I

10000 12000 14QQQ 1tlPOQ


Pt=fTH I FT

-e-
ST.STATE(3F PS) -+- ST·STATF(4fpa) -r AT.STATFHiFPS)
-8- DYN.(3FPS) --*- PYti·(4fPS) -+- PYN·(5fPS)

Fig. 4- SW&p prefls~re V~, qapth

Vous aimerez peut-être aussi