Hajianmaleki 2012
Hajianmaleki 2012
Hajianmaleki 2012
IMECE2012
November 9-15, 2012, Houston, Texas, USA
IMECE2012-89638
ABSTRACT
INTRODUCTION
Understanding drill string buckling behavior is a
MOTIVATION
significant challenge to the petroleum industry. In this paper,
the explicit finite element method implemented in Abaqus Buckling analysis of drill strings is a serious problem
software is employed to study the buckling of drill strings for in the petroleum industry. Buckling can intensify the bending
inclined straight wellbores. Classic solutions for the critical stress and lead to fatigue failure over time. More importantly,
buckling length of self-weighted columns as well as critical buckled shapes exert larger side forces than unbuckled which
buckling load for drill pipe inside inclined wellbores are increases friction losses and can lead to the lockup of the string
compared to explicit FEA and accurate results are provided by and potential loss of equipment and section of the well.
the finite element based predictions. The effect of different Increases in the depth and deviation of wells have intensified
inclination angles and string effective weight due to the this problem and led to the need for comprehensive models
buoyancy effect has been studied and the results for sinusoidal capable of analyzing and predicting different aspects of drill
and helical buckling are compared to analytical results and string buckling.
experimental data in the literature. The theoretical predictions
for different inclination angles agree with the simulations. BACKGROUND AND LITERATURE REVIEW
Theoretical buckling load of inclined drill strings approaches
zero by decreasing the effective weight of a floating drill string. In order to increase the rate of drilling (rate of
However, the results of finite element simulations show that penetration) optimum weight should be put and maintained on
significant buckling load would still exist for very low drill the drilling bit. The weight on bit (WOB) is limited by a critical
string effective weight. These results are confirmed by buckling load where the pipe loses elastic stability. Therefore, it
experimental results provided by other researchers. Overall, the is very important to determine the maximum load permissible
efficacy of using explicit finite element methods to model drill on the drill string. Critical buckling load depends on the pipe
string buckling behavior is demonstrated. geometry, drill pipe elastic modulus and wellbore geometry. It
Since then, different formulations for the onset of Traditional implicit finite element models have been
sinusoidal and helical buckling in different wellbores have been problematic in solving buckling problems for constrained
proposed by other researchers [3-15]. geometry because buckling behavior causes local instabilities
and sharp changes in contact behavior, which leads to
Gao and Miska [16] studied the effect of boundary numerical instabilities. Techniques using non-linear springs and
conditions and friction on buckling in horizontal wellbores and penalty methods have been used with varying degrees of
w is more success, but an alternative based on the explicit finite element
showed that if the dimensionless length µ = 4 L method is sought. This method does not require well defined
EI
boundary conditions; instead, it will solve for them by
than 5π, the assumption of a long drill pipe (free from boundary determining where the drill string contacts the wellbore. Also,
conditions effect) can be applied. the explicit method does not use a direct, banded, linear
equation solver, thus the computational cost of a solution
Theoretical critical buckling load is usually
changes linearly (compared to quadratic) with the problem size.
determined by an eigenvalue analysis. Eigenvalues estimate the
Large deformations, sliding and contact constraints are
upper bound of critical buckling load (bifurcation load). The
relatively easy to implement in the explicit procedure [18].
more critical lower bound of the critical buckling load can be
predicted with large deflection analysis using the finite element The implementation of the explicit method in the
method. In general, load-displacement curves for a buckling Abaqus software uses a time-forward integration scheme based
problem are nonlinear and the curvature of the current on the central difference operator. Internal time increment
wellbores makes it more nonlinear [17]. Explicit non-linear controls are used to keep the solution stable over long
finite element modeling that considers large displacements and durations [19]. Since the results and analyses used for
contact is a good candidate for solving these problems. comparison are static, the inherently dynamic solution from
Abaqus/Explicit is run for a sufficient duration to produce
PURPOSE AND SCOPE quasi-static results. The solution is checked by observing that
the total kinetic energy becomes asymptotic to zero with
The purpose of this paper is to demonstrate the
respect to time. The general procedure for constructing and
efficacy of using the explicit finite element method to model
executing an Abaqus/Explicit model has the following steps:
the buckling behavior of drill strings. Both analytical solutions
and experimental data are compared with the numerical 1. Define the part and environment geometry.
simulations. Where appropriate, the results are presented in 2. Incorporate the material and section properties.
normalized form. When experimental descriptions are 3. Model the wellbore using Connector elements with
available, the explicit FEA models were constructed to match clearance, positive stops, friction, and damping.
the experimental setup. Similarly, most simulation assumptions 4. Define the loading steps and output requests.
in the literature are implemented in the FEA models. Since 5. Mesh the part geometry.
Experimental Data
Fig. 1: Lateral deflection of the center of a vertical column
under its own weight from a perturbation for various heights In 1994, Salies et al. [20] performed vertical buckling
experiments on a stainless steel pipe having the following
dimensions: OD = 0.25”, ID = 0.21”, and length = 643”. The
Lubinski’s critical column length formula, Equation 4,
test was performed inside a 2” ID Plexiglass tube that simulated
predicts that for pipe with a 3.826-inch ID and a 4.5-inch OD,
the wellbore. The experimental sinusoidal buckling load is
the critical length is 133.6 feet. This length occurs as the
compared to explicit FEM and theory [1] in Table 2.
deflections plotted in Figure 1 are beginning to diverge, thus
the FEA results agree with the prediction based on Equation 4. Table 2: Comparison of the parameters γ1 (Equation 2) in
The criterion for buckling is a condition of equilibrium. If the experiment and explicit FEM
column is in a stable equilibrium, then once a lateral
perturbation force is removed, the lateral deflection will return Method Experiment Reference [1] Explicit FEM
to the previous state. Clearly, this is the case for any drill string
under tension. As the drill string transitions from tension to γ1 1.5 1.94 1.81
compression, the strength of the equilibrium stability decreases
as restoration force is decreased. Since the base of longer The result for buckling load from explicit FEM is
vertical columns experience higher compressive forces, it is between the experimental data and the theoretical formulation.
expected that the strength of the stability is decreased. This is The difference between the experiment and the theory is
shown in Figure 1 where the perturbation returns to equilibrium explained by the imperfections present in the test pipe.
fastest for the shortest column.
The critical buckling length produces a condition of Inclined Wells
neutral equilibrium. This suggests that the deflection would not
change once the perturbation force is removed. This is seen as Analytical Solutions
the column with a length of 133.6 ft, the critical column length, Dawson and Paslay [7] suggested the following
is approaching a steady state deflection. The condition of formula for the critical load that predicts the onset of sinusoidal
neutral equilibrium is difficult to maintain in practice. buckling of drill strings in long inclined wells.
Once the critical length is exceeded, thus increasing
the compressive force on the base of the drill string, a condition EIw sin α
of dynamic equilibrium exists where the sum of the forces F=2 (6)
results in acceleration. This is demonstrated in Figure 1 by the r
divergence of the lateral deflection for column lengths greater
Incl. angle 30 60 90
Experimental Data
Arslan [21] performed experiments to study the effect
of pipe weight (floated pipe buckling) on critical buckling load Fig. 7: Bottom load vs. axial displacement curve for Arslan’s
of tubulars in horizontal wells. Two experiments of his are experiment in air and water
simulated in this paper to show the capability of the model in
simulating buoyant drill string buckling. The simulated case is a
Table 6: Critical buckling load (lb) for the pipe in water with Fig. 8: Effect of effective weight on sinusoidal buckling force
load imperfection The results are made dimensionless by dividing to the
value at normal weight conditions. The results show that the
Experiment FEM theoretical model loses its accuracy when the effective weight
Equations decreases to very low levels. At an effective weight around
Mode 1.5% of normal weight the actual buckling load is 3.5 times the
6 and 7
Top Bottom Top Bottom load predicted by equation 6, according to the FEM results.
DISCUSSION
Sinus. 110 92 111.3 109.2 92
The theoretical formulations presented for buckling
are predicted in this paper using explicit FEM. The
Helical 180 160 156.6 153.6 152.2 formulations for column buckling under self-weight, sinusoidal
and helical buckling in inclined wells are successfully
The trends seen in the force-deflection curves for predicted. However, the buckling load predictions in simulating
buckling from explicit FEM are close to experimental results. experiments were higher that actual experiment data. This was
The onset of buckling from the explicit FEM results predict observed both in vertical and horizontal wells. The authors
higher values than the analytical and experimental results believe that this problem is due to uncertainties present in
because they include the effect of friction and do not have the experiments. These uncertainties include imperfections both in
imperfections present in the experiments. loading and geometries, residual stresses in the structures,
imperfections in experimental devices, and errors in
For pipe in water the FEM results are closer to the
determining the onset of buckling, especially the onset of
experimental results than theoretical predictions using
helical buckling. As shown in Table 5, imperfections can have a
Equations 6 and 7. The simulation results from buckling in air
considerable effect on prediction of buckling. It was also shown
with imperfections show that introducing the load imperfection
that by increasing the length, the effect of boundary condition
is closer to predicting helical buckling and the geometric
vanishes; but the effect of length becomes important and lowers
imperfection is closer in predicting sinusoidal buckling.
the onset of sinusoidal and helical buckling.
However, more research is needed on the effects of geometrical
and loading imperfections on buckling.
CONCLUSIONS AND FUTURE WORKS
Continuing with Arslan’s experiments, the experiment
The explicit finite element method has the ability to
is simulated for different effective weights of the drill pipe with
simulate the buckling behavior of drill strings inside well bores.
the results presented in Figure 8.
The model has been created in Abaqus FEM software and the
results were accurate in all cases when compared to the
theoretical values. The differences in the numerical predictions
and experimental results are most likely due to model and
experimental uncertainties, especially the imperfections present
in the experiments. The effect of inclination angle, length, and
effective weight has been studied and it was shown that the
theoretical prediction of sinusoidal buckling is accurate at
different inclination angle but losses its accuracy by lowering