SPE 36829

Complex Well Control Events Accurately Represented by an Advanced Kick Simulator

w 4s’ .

SocietyIXPstroleumEnginwrs

l.M.Rezmer-Cooper, SPE, and J, James, Anadrill, P. Fitzgerald, Schlumberger Wireline and Testing, A.B.Johnson, SPE,
D.H.Davies, SPE, and l.A.Frigaard, Schlumberger Cambridge Research, S. Cooper, Schlumberger Retail Petroleum
Services, Y.Luo, SPE, and P. Bern, SPE, BP Exploration,

c~ywht 1S98 .S.XIetY of Petroieum Enginears, Inc

peak choke pressures, arrival times of gas at surface, and
Thm paper was pmpamd !W pmswdattcm at the 1SSd SPE European Petroleum Conference
held (n M!Ian, Italy, 22-24 OcfOber 19SU
early kick detectors,
Thm pww was se&tad for IMewntatbn by an SPE Program Canmdlae followng raww c+
Information cordmned m an abtmct wbmitted by the authcf(s) Contents cd the paper, as Modelling multiple zone inftuxes and different intlux typx
presmled, hava nd bwn revwwad by ttw %~ C$ Patroloum EngInoers and are subject to
correct-x by the author(s) The material, as presented, doss not necessarily reflect any
allows for the versatility required for the dynamic simulation
powuon of the society d Petmloum EngInwm, ds ofhcmx, w members Papem presented at of controlled underbalanced drilling and well control arising
SPE nwetmgs ●e Subteci tn publwafmn reww by Edftonal Commlteas of 0?9 SW@ cd
Petroleum Engmwm Permismon to copy is rastmted to an abstract 01 not more than 302 from such practices.
words Illuatrtdms may not ta copwd Tha abstracl should calm conspicuous
acknowledgment 01 where and by WI!CX!Ithe paper was presantad Wrde L!brarian, SPE, P O
80X 833S36. Richardson TX 7=3S36. U S A fan 01.214-P52-9435 The results from the simulation of a field kick in a highly
deviated well (that included an extended shut-in phase and
fragmented kill procedure) have been used to demonstrate the
Abstract importance of accurately modelling such non-standard kills,
New functionality has been added to an advanced kick and the consequences of not adhering to standard well control
simulator to provide a more versatile and robust tool in well techniques,
planning and well control. New physical models are described
that enable the simulator to represent accurately, for a range
drilling and formation fluids, the dynamic conditions in Introduction
slimhole wells, multiple producing zones in highly Drilling environments and the techniques employed to cope
deviated/horizontal sections, underbalanced drilling, losses to with them are becoming increasingly complex, Elaborate
permeable zones, and compliant wellbores. tools are also king used to understand the physics involved
in these complex scenarios. Practical planning for well
The slimhole pressure drop model highlights the importance control events and related problems (post-event analysis and
of accurate prediction of the annular flow regime when day-to-day kick tolerance calculations) has been carried out
assessing the effects of high speed pipe rotation, Current with advanced simulation tools [1,2].
industg misconceptions concerning slimhole annular
pressure drop are discussed, along with the implications for These models were initially developed as planning and
early kick detection in slimholes. training tools to improve drilling safety. Such simulators are
the results of extensive theoretical and experimental research.
A new gas migration algorithm, validated for kicks in wells The first phase of the simulator described here was developed
deviated up to 90° is now included. We show that gas can with the collaboration of the UK Department of Energy (then
move significantly faster in deviated wells than in an the UK Health and Safety Executive), and BP International,
equivalent vertical wellbore, and highlight how the simulator [3]. The second phase, and the focus of this paper, was
can & used to give improved estimates of surface gas funded in part by the European Union Thermie programme
ftowrates for worst case kicks for use in mud-gas separator and BP Exploration. The objective of such tools is to develop
sizing. a deep understanding of the physical processes that occur
when gas enters the welltmre during drilling. Although the
The inclusion of models for wellbore compliance, fluid loss prima~ applications have been in improved safety of
and gas suspension characteristics enable a more realistic operations, the simulator described in [3] has also been used
representation of shut-in pressure. We also discuss the to reduce overall well costs. However, whilst the simulator
important differences between gas kicks and mixed oil/gas has been demonstrated to be accurate, and has been widely
kicks in terms of important welt control parameters, such as used, it was identified as being limiled in its breadth of

141
2 I.M. REZMER-COOPER ET AL. SPE 36829

application, originally being designed to model only flows in carried as large bubbles. However, in shedding a trail of small
vertical wells with diameters in excess of 6in, whilst bubbles, the gas cloud becomes smaller. These small bubbles
considering the wellbore to be a rigid, leaktight vessel. With remain stationary relative to the fluid (typically they are
the drive in drilling in the 90’s towards more extended reach smaller than 2mm), and are held in suspension by the
and horizontal wells, slimmer geometries, and wells drilled effective yield stress of the mud. If the well is deep enough,
underbalanced, there is a need for an enhanced kick simulator all of the migrating influx can b suspended, and gas
that can accurately model the above scenarios. In this paper migration can cease completely. Field evidenix of this is also
we describe the theoretical and experimental work that has given in [7] along with an experimental study of gas
produced such a simulator, and give examples of its suspension characteristics. Between O.S-S”/O of the influx void
application in a wide range of drilling and planning fraction can be held by the yield stress of the mud. This is
operations. We also attempt to address the link between significant in deep wells or in large annular geometries (such
surface pressure rise and gas migration by the incorporation as deep water risers), Once in suspension, the gas will remain
of more physical factors than have been previously included there until either the mud rheology breaks down, or the mud
within a dynamic simulator. is circulated out of the well.
The effect of a variable suspended gas fraction has been
included in the simulator, and it is shown in the example
Gas Migration and Shut-in Pressures below that the nature of the shut-in pressure rise can be quite
When surface pressures take a considerable amount of time to sensitive to the amount of gas held as small bubbles in
stabilise, it is ofien difflcuh to determine the drillpipe suspension. Here we consider a 10 bbl methane influx as it
pressure that truly reflects the actual formation pressure. migrates up a 10000 ft vertical well filled with water-base
Industry standard kick control and kill procedures must be mud. Within the simulation we allow for the compressibility
straightfonvard, and rely on simple calculations that can be of the mud-gas mixture, but do not allow any volubilitywithin
performed by hand under conditions of extreme stress. the mud, so that we can isolate the suspension characteristics.
Billingham et al, [4], highlight some of the difficulties that We consider three differing gas suspension fractions as the
can arise by using the simple calculations. Furthermore, these influx migrates in a rigid, leak-tight wellbore (O, 2 and 5’%o,
methods can be flawed, and are not appropriate in highly Figure 1). If there is no gas going into suspension, the surfaa
deviatedlhorizontal wells, [4,5]. As the SIDPP is used to pressure would rise by nearly 2750 psi. With 2’% gas
determine the kick zone pressure, and hence the kill mud suspension, the surface pressure rises by 2250 psi before the
weight, an artificially high drillpipe pressure will result in the influx reaches the top of the well, However, for a gas
well being overbilled. An example of this misinterpretation suspension void fraction of 5’XO, the maximum pressure
can be seen in the field kill described later in the paper. increase is 1000 psi, and gas only just reaches the surface.
Thus, for the case of a trail of suspended gas remaining in the
well, the rate of pressure rise would give an inaccurate (low)
Gas Migration and Suspension estimate of the gas migration rate if suspension is ignored,
After a kick has been taken, there is usually (but not
always) a tendency for the in!lux to migrate in the wellbore if
it has moved from any horizontal or undulating sections. A Comprehensive Gas Migration Algorithm
Influx migration in a closed well can cause excessive From experimental data, the migration of gas in a
pressures within the wellbore if suitable pressure controls are wellbore can be seen to b affected by both the flow geometry
not implemented. Many authors have focused on the link and the inclination. The experimental tests described here are
between gas migration and surface pressure rise. A recent made with theologically accurate fluids that have been
paper by Grace [6] has indicated that gas migration rates can described in previous papers, [8,9].
be estimated directly from the surface pressure rise by the Johnson and co-workers have reviewed in detail the
traditional field method. An alternative view is offered by literature of experimental and theoretical studies of single
Johnson ef al. [7] in which the conflicting literature on gas bubble dynamics. These studies have investigated the rise
migration is reviewed. Laboratory, large-scale test data and velocity of gas in typical drilling geometries and with realistic
field data are all reviewed to illustrate that gas can migrate at drilling fluids. Earlier experiments have concentrated on
a wide range of rates, from the very fast (-19800fVhr) to the deviations up to and including 60”. This early work
very slow ( 180ft/hr), and may become completely suspended highlighted that it is impxtant to characterise the effects of
in certain circumstances For gas void fractions greater than deviation and geometry separately. The correlation within the
10”A,gas migration rates in realistic drilling fluids have keen simulator has been based upon the Zuber-Findlay model, [10],
shown to be approximately 6000 tVhr, [7], These are that relates the gas velocity, Vg, to the homogeneous velocity,
significantly larger than the equivalent migration rates in
Vh, via a distribution factor, Co, and a slip velocity V~l,p,
water. Initially, one can envisage the bulk of the influx being

142
SPE 36829 COMPLEX WELL CONTROL EVENTS ACCURATELY REPRESENTED BY AN ADVANCED KICK SIMUIATOR 3

which represents the rate at which the gas bubbles slip past ● In flowing muds, increasing deviation from the vertical

the liquid phase, to approximately 45°-500 increases the bubble vehxity,

The inclinable 15m flowloop described in [9] was used for beyond 50° the bubble velocity decreases.
all of the tests described below, with a variety of different ● Gas can migrate/move ve~ fast (as high as 13000 ft/hr),

geometries listed in Table 1, Figure 2 shows typical Zuber- even in highly deviated sections.
Findlay plots for some of the experiments [three geometries at ● Slip velocity falls with increasing radius ratio
75° deviation). This figure shows a good degree of linearity (increasingly narrow annular gap).
and levels of scatter not too dissimilar from those found in ● The distribution factor rises with radius ratio for all

previous experiments, We also note that Ihe Taylor bubble flows, and increases markedly with deviation in narrow
velocity, plotted at zero homogeneous velocity, falls on the annuli.
same line as all of the data taken with co-current flows, and
gives cotildence both in the applicability of the Zuber-
Findlay model at high deviations and the method used to Gas Dissolution
measure the gas velocity. The results of the experiments are when an intlux enters the wellbore, some (or all) of it will
summarised in Figures 3 and 4, which show the gas be dissolved in the mud at the ambient temperature and
distribution factor and the slip velocity as a function of pressure. In addition to gas dissolution in oil-based muds,
deviation and geometry. Using a model based on gas there are also significant dissolution effects in synthetic muds
migration in a pi~ with a simple modification for the and even in water-base fluids. For a methane influx into a
centrebody in terms of an eikctive diameter will predict well drilled with oil-base (or certain synthetic) muds, above
inaccurate gas migration behaviour in an annular geometry, about 8000 psi all of the influx can be dissolved, leaving no
as both slip and distribution factor are functions of geometry free gas. If this dissolved gas becomes dispersed, and is held
and deviation. by the yield stress of the mud below the bubble point in the
The distribution factor represents the gas velocity well, there will be no free gas, and no pressure rise during
enhancement due to the concentration of the gas in those shut-in. Rezmer-Cooper er al., [13], discuss in detail the
regions of the flow where the liquid phase velocity is faster consequences of a gas intlux into an HFHT well drilled with
than the mean flow. In the annular geometry, the gas is OBM. They show, using the simulator described here, that
forced into the region of the faster flow, increasing the most of the influx will be dissolved in the mud, thus
distribution factor. An increase in distribution factor can also removing one of the processes for increasing the wellbore
be seen with deviation in narrow geometries. pressure, The same may be true (for smaller volumes of gas)
Rommetveit et al. [11] descritw full-scale gas-kick in a WBM. Solubility effects are complex and are strong
experiments performed in a horizontal pipe with air and two functions of pressure and temperature. A detailed simulator is
different liquids. They concluded that a Zuber-Findlay type really the only way to accurately allow for volubility effects in
model does not represent their data well, and developed a well planning, [1].
more general correlation. However, it is not clear that their
correlation is a better fit to their data than a Zu&r-Findlay tit.
They simply show that the data has a large degree of scatter Wellbore Compliance and Fluid Compressibility
around a singie best-tit line with fixed parameters. No A shut-in wellbore is equivalent to a large vessel
allowance has been made for the dependence of the Zuber- containing compressible fluids of possibly more than one
Findlay parameters on the geometry, deviation, etc. No phase. When the pressure in the wellbore changes, the
attempt has been made to parametrize the gas-slip velocity wellbore wall starts to deform, For small variations in
and distribution factor in a manner similar to the analysis pressure, the deformation is linearly elastic. For larger
described in [9] and also utilised here. Luo ef aI, [12], have changes, the deformation is plastic. We assume that the
also conducted gas migration experiments in a full-scale pressure changes are such that we are below the formation
flowloop with hole angles from the vertical to 5“ beyond the yield stress. Elastic deformation reduces the rate of shut-in
horizontal. Their results showed that the effects of hoIe angle, pressure rise; changes in the wellbore pressure lead to small,
mud rheology and drillpipe eccentricity were all complex and but significant, changes in the system volume. The wellbore
mutually interdependent. Gas migration rates in excess of compliance can be divided into two components: openhole
10,000fUhr were also found, depending upon the hole size and cased sections. The bulk system compressibilityy is the
and angle, etc. With their limited datasets they could not sum of these compliance effects coupled with the
separate this complex interdependence and reach a definite compressibility of the welltmre fluids. The effects of wellbore
quantitative conclusion regarding each of the parameters compliance can be of the same order as the compressibility of
varied within their experiments. Nevertheless, the general the fluids, and thus have important consequences for any gas
agreement between the datasets and the results presented here migration velocities inferred from the shut-in pressure rise
is good, and similar conclusions can be reached: rates, [14]. We show in the following example how a

143
4 I.M, REZMER-COOPER ET AL. SPE 36829

potentially compliant wellbore can be modelled within the fluid loss, which is controlled by two parameters: one is the
framework of plaming for well control operations. Figure 5 mean invasion speed, and the other is a ‘timescde’, which is
shows a range of simulations based on a 20 bbl gas kick taken used to simulate tiltercake build-up, amongst other effects.
in 12.5 ppg WBM in a 15000 tl vertical well with 9 5/8” The ‘zero-time’ for the timedependence is assumed to be the
casing to 10000 ft. We consider three cases: a rigid wellbore, start of the simulation, although we realise that this is
a typically compliant wellbore, and a very compliant perhaps not realistic if any additional flowing zones were
wellbore. We note that in these simulations, the wellbore exposed long before the currently, drilled section, and the
fluids are also compressible. In each case, the actual gas tiltercake is undisturbed, When there is a multiphase mixture
migration rate is -5000 fVhr. Using the simple field model of fluids present in the wellbore, it is virtually impossible to
relating gas migration directly to surface pressure rise, we model the losses of each individual phase as a distinct entity.
obtain a severe underestimate for the rigid wellbore of 2584 Thus, we model the loss of each phase in proportion to the
Whr, due to the effect of the mud compressibility. This is volume fraction of the phase present at that particular time
further reduced by the wellbore compliance. The typical and position.
compliance reduces this to 2040 Whr, and a very compliant The presence of fluid loss complicates the possible
wellbore reduces it even further to 1643 Whr. Therefore, in a sequence of events that can occur during shut-in, [13]. The
compressible system, and with a compliant wellbore, there is rate of fluid loss maybe so great that the pressure rise due to
a danger of significantly underestimating the gas migration the influx is compensated by the leak-off. When there are
rate, the gas will & at surface earlier than expected, and significant amounts of fluid loss, it takes longer for the flow
inappropriate action may be taken. from the formation to shut-off. However, the shape of the
pressure build-up curve is similar to that with no losses.
Again, gas migration rates derived from the pressure rise
Fluid Loss rates using the field method will be underpredicted when
Fluid leaking from a shut-in wellbore can either reduce there is fluid loss, and gas will be at surface earlier than
the wellbore pressure, or reduce the rate of rise of the expected.
pressure. It is the net flowrate (influx minus fluid 10SS)that In the following example, we show, with the aid of the
determines the sign of the time rate of change of the shut-in simulator, how fluid loss can initially mask an influx, lead to
pressure rise (in the absence of gas volubility). Therefore, it is a larger overall influx, and yet give rise to the same type (or
important that fluid loss rates (of whatever severity) are shape) of pressure build-up curve (we neglect the decrease in
accurately characterised, and included in any planning where pit-level due to the increased hole volume while drilling)..
it is likely that drilling fluid may leak off to the formation in Furthermore, the peak choke pressure while circulating OU(
significant quantities, the kick will be higher in the case with fluid loss, due to the
Drilling muds flowing into a porous medium forma frlter- larger inftux that has been taken. We consider the effects of
cake on the surface which act to reduce the flowrate of fluid circulating out a 20 bbl kick in a 7500 ft well (nominal
through the surface of the medium. For formations with a openhole diameter 8.5 in, cased to 3000 R) with the Wait and
permeability greater than about 1rrdl: it is the mature Weight method. We compare a case with losses to an
filtercake rather than the rock permeability that limits the equivalent kick in a rigid, leak-tight wellbore. Figure 6(a)
flux of fluid into the formation. When the overpressure across shows the casing pressure and Figure 6(b) the corresponding
a filtercrrke increases, e.g., during shut-in, the cake is pit gain. In the case with losses, the initial shut-in is at a
compacted and the effective permeability is reduced. The higher pressure as there is more gas in the wellbore due to the
effect of this drop in permeability approximately balances the initial masking of the kick by the losses. However, the shut-in
increase in overpressure; the effective fluid loss rate is thus pressure rise rate is less, as the losses continue (although now
unchanged, [14]. For a mature fihercake grown in a static at a lower rate, due to the filter-cake build-up). A higher
fluid the fluid flux rate (volume flowrate per unit area) can be choke pressure is required when circulating the kick, as by
in the range O.l-l~nds. This appears to be a very small rate, the time the gas reached the surface, the losses have been
However, for a well with a long openhole section, this can reduced by the filtercake build-up, but there is a larger total
lead to significant losses. For a newly opened section of kick to circulate out. If we examine the pit level trace more
permeable formation, the flux can be greater than lpmh. closely, we see that there are indeed losses while drilling
However, the effect of the growing tiltercake is to reduce the during the early stages of the kick, evidenced by a pit-level
flux to fall in proportion to (time)-i’*. Johnson and Tarvin, decrease. However, it is unlikely that such a level of losses
[14], have shown how it is possible to obtain estimates of not could be detected with conventioml equipment. Finally, we
only the wellbore compliance, but also the fluid loss from see that there was a total of about 10 bbl lost to the formation.
careful analysis of leak-off test (or formation integrity test) In summary, the greater the fluid loss, the longer it takes
cknta. for the flow from the formation to shut-off. The more mature
Within the simulator, we model only the gross features of the the filter-cake, the lower the fluid loss rate, and hence an

144
SPE 36629 COMPLEX WELL CONTROL EVENTS ACCURATELY REPRESENTED BY AN ADVANCED KICK SIMULATOR 5

earlier formation shut-off time. In some eases, the fluid loss ● Frictional pressure losses are insignificant.

may be such that the continuing inftux is masked; shut-in With a physically-treed simulator that does not have to
pressures may still appear to stabilise, even though the influx make such basic assumptions, we ean reassess the suitability
continues. of the single bubble model to estimate the ‘worst case’ surface
gas flowrates and peak choke pressures. Perhaps the most
important effeet that is neglected in the single bubble model
Reconciling apparently contradictory observations is the effect of gas slip. In [16] the results of a simple
Much of the industry controversy concerning gas analytical model that is an extension of the single bubble
migration is due to the misinterpretation of the effects of gas model highlight that in cases where the gas slip is large
migration, gas suspension, and gas volubility. Gas migration compared to the annular mud velocity, the single bubble
is a complex issue that cannot be deseribed by a simple rule of model can underestimate the surface gas flowrate by 3-4
thumb or single slip velocity. At large eoneentrations, the gas times. One of the most important things to note is that if we
will move FAST. This rapidly moving gas cloud ean leave a are not circulating, i.e., just letting the gas migrate, then the
trail of bubbles, which may STOP, and keome suspended in single bubble model will predict no gas flowrate at surface.
the well by the yield stress of the mud. Irr field conditions, an We now show that the enhanced analytical model
accurate estimate of gas migration during a well control described in [15] is still not stilciently comprehensive to
incident is diflkult, as the gas can SLOW down as it beeomes account for the detailed sequence of events that occur as a
more dispersed. Additionally, unless we account for the gas kick is circulated to the surface. To enable a more
effects of fluid loss, mud compressibility, and wellbore realistic representation of the near-surface physical processes
compliance, we may seriously underestimate the gas we appeal to the simulator.
migration rate. In the next section we highlight the Figure 7 gives the surface gas oufflow rate from the
importance of an accurate model of gas migration rates in one simulator for a 50 bbl kick in a 1000Oftwell, having a Shut-
important area of well control contingency planning. in Drillpipe Pressure (SfDPP) of 2000 psi and an annular
volume per unit length of 0,064 bblltl. As with the analytical
model that included the effects of gas slip and initial
distribution (described in [16]), we can see that slip alone can
Use of the Simulator in Mud-Gas Separator Sizing give rise to gas flow at surface. In addition, halving the slow
The Mud-Gas Separator (MGS) is designed to enable circulation rate from 4 bbt/min to 2 bbt/min does not halve
effective separation of the mud and gas circulated from the the peak gas flowrate seen at surface. Furthermore, these peak
well as the result of a well control operation. The capacity of flowrates are also all significantly higher than the equivalent
the separator is designed according to two criteria. These are single bubble calculation
discussed in detail in [15,16]. Traditionally, for a given It is the delay lXhV&n the peak surface pressure and peak
separator design and the profile of gas expected at surface flowrate that can give rise to high surface gas flowrates.
from a hypothesised ‘worst ease kick’, a ‘safe operating Figure 8 gives in detail the sequence of events that occurs as
handling capacity’ can be identified. Thus, maximum slow the gas is circulated to the surface. We show the choke
circulation rates (SCRS) can be selected. This means that by pressure and the surface gas flowrate as the kick is circulated
careful selection of the kill rate, even under worst case out at an SCR of 4 bbl/min:
conditions, the separator gas handling capacity should not be (l). Gas is first at surface. The choke pressure is still
exceeded, [17]. This assumes that the rate of gas delivery is increasing to maintain a constant bottomhole pressure. We
directly proportional to the SCR; this is not necessarily the start to bleed off gas.
case. (2). With gas now at surface and being circulated out, the
The key parameter in the design and sizing of the mud- maximum choke pressure is reached. We now start to reduce
gas separator is the maximum gas flowrate that can be the choke pressure as more gas/mud is bled off.
expected at surface from a given kick, as it influences both (3), With the choke pressure now falling, and still
the separation and blowdown capacity calculations. maintaining a constant bottomhole pressure, the peak surface
Traditional sizing methods based on the representation of the gas flowrate eccurs.
kick as a single bubble have assumed the following: The reason for the time lag between the peak flowrate and
● The intlux enters the well and remains as a single the peak pressure can be seen by considering what is
bubble while it is circulated to the surface. happening in the tail of the gas, Between stages (2) and (3),
● The intlux does not slip relative to the mud. the pressure in the tail of the bubble is falling as the tail slips
● During the kill, bottomhole pressure equals the and expands, imparting additional momentum to the mudlgas
formation pressure. mixture ahead of it, and thus increasing the gas flowrate at
. The annular geometry is uniform surface. In the single bubble model, with 100°/0gas void in
c The mud density is constant. the whole bubble as it reaches the surface, the bubble cannot

145
6 I.M. REZMER-COOPER ET AL. SPE 36629

expand or slip, and hence this additional momentum cannot the effective viscosity of the fluid. Once the critical Taylor
ke realised. number is reached, vortices will form which lead to a
Conventional MGS design based on single bubble turbulent-like pressure drop as they extract energy from the
modelling often includes numerous areas where conservatism mean axial flow. As the axial flowrate is increased, till
is applied, For example, Low and Jansen, [18], state that for turbulence will occur. The introduction of a yield stress
‘typical modern oilfield MGS’S, a safety factor of 5 to 10 exist causes a marked increase in the stability of the kuninar
between separation and blowdown capacity’. This (and other) regime, and consequently vortices are not generated until
safety factors do not appear to be quantified in the literature. much higher rotation rates. The effects of rotation in the
However, we have shown here using the simulator that the differing flow regimes can be seen at the two extremes of the
single bubble model does not necessarily lead to the worst plots in Figure 10.
case flowrates. Of course a certain safety margin for
blowdown m~city (e.g., 100%) should be taken into account, The new model developed from these experiments has
but there appear to be shortcomings in the conventional given very good agreement at the lower rotation rates for
system design based on single bubble modelling, suggesting ‘normal’ geometries, and at high rotation rates for slimhole
that MGS design methods and calculations need to be muds and geometries. Agreement is poorer for high rotation
reviewed. More physically based methods will allow for a rates in the conventional drilling geometries.
more effective approach to MGS sizing, and less effective The simulator can be used to examine the sensitivity of
combinations of well design equipment and procedures will various important parameters on a well control operation in a
be highlighted. Ultimately, these methods will enable the slimhole. Table 2 shows some of the input parameters used in
correct degree of conservatism to & adopted, in a cost- the simulation of a propo~d slimhole well. Figure 11
effective manner. highlights the effeets of rotation for a range of circulation
rates. The differing effects of rotation in Iaminar and
turbulent flow regimes can clearly be seen as the rotation rate
‘Slimhole’ Frictional Pressure Drop is increased in steps.
Until recently, the industry has been polarised on the role of Early kick detection is essential in slimholes because of
drillpipe rotation on annular pressure drops. Some papers the reduced annular volumes. Figure 12 emphasises the
have explicitly stated that rotation acts to increase the axial importance of not allowing too large a kick into the wellbore.
pressure drop, whilst others have taken the opposing view, The differences between the maximum shut-in casing
stating that an increase in rotation rate decreases the annular pressure (SICP) for lbbl, 5bbl, and 20bbl kicks are
pressure drop, [19-21]. Experiments reported by by McCann significant. Furthermore, Figure 13 shows that the trend in
et al., [22], highlight that both of these seemingly conflicting the standpipe pressure for a kick in a slimhole (kick starts at
views are correct, and that both effects ean be observed. We 10 reins) is such that the decrease in the hydrostatic
confkn these findings here and show that accurate slimhole imbalance due to the lighter density kick fluid in the annular
pressure drop predictions can only he achieved by prediction side of the U-tube is more than compensated initially by the
of the correct flow regime. A typical flow regime map is flow of the gas into the wellbore. The opposite effeet usually
shown in Figure 9. occurs in a ‘normal’ 8.5in hole.
The model for Iaminar flow employed in the simulator has
already been shown to be a good representation of the
behaviour of the a non-Newtonian drilling fluid under Controlled Underbalanced Drilling and Multiple
conditions of moderate pipe rotation, [3]. A new empirical Producing Zones
model for turbulent flow has been developed based on It has long been recognised that a key to improving the
experiments conducted in the 15m flowkmp deseribed in [9]. recovery of reserves is to minimise the reservoir damage
These experiments were conducted in a range of geometries, created while drilling. The main objective of underbalaneed
and for three mud analogues at two eccentricities (concentric drilling (UBD) is to increase the well productivity by
and 69% eeeentered). At each of these combinations, the reducing the formation damage. In both conventional rotary
frictional pressure drop was recorded at twelve different and coiled tubing UBD it is necessary to u= a lightweight
liquid flowrates The drillpipe rotation was also varied. For drilling fluid, or if the reservoir is sub-hydrostatic, to lighten
the eeeentric case, speeds of O, 200 and 450 rpm were used. the fluid by injeeting gas. The gas may be injeeted through
Speeds of O, 100, 200, 300, 400 and 500 rpm were studied in the drillpipe, or into the annulus via a parasite string, gas-lift
the concentric geometries. Figure 10 gives a sample set of mandrel, or eccentric casing. Falk and McDonald, [23] give
experimental data for one of the muds and geometries as a an excellent review of UBD techniques.
tlmetion of the rotation rate. In lamimr flow, the annular The simulator has been enhanced so that it can represent
pressure losses deerease with increasing pipe rotation, whilst all of these injection techniques. One of the prima~ requests
in the laminar flow regime, the azimuthal stresses will reduce for UBD is the safe and accurate control of the bottomhole

146
SPE 36829 COMPLEX WELL CONTROL EVENTS ACCURATELY REPRESENTED BY AN ADVANCED KICK SJMUIATOR 7

pressure at a level bdow the formation pressure throughout flowing zones, it is diffkult to ascertain the nature of the
the drilling operation, This is not straightforward when the downhole flows based purely on the surface measurements;
well is underbalaneed as it is possible that the entire length of individual events can be very complex.
the openhole section can produce fluids. Such downhole Most underbalanced drilling operations have been
behaviour can significantly complicate a well control event. A conducted in well established formations with little or no well
new facility has been added to the simulator so that fluid ean control problems documented. However, as the precedure
flow into (and leak out of) the well from any permeable becomes more established, there is the likelihood that the
region in the openhole section. procedures will be extended to less well known formations,
In the following example we consider conventional rotary where the chance of a well control incident will increase. In
UBD carried out in a 6000fl well. In addition to the drilled this case, a tool such as the simulator can be used in the
8,5in influx, there are also two additional flowing zones 20ft planning environment so that surface equipment could be
and 100R long Ioeated at 4500tl and 5000ft measured depth properly sized, conservative easing programs designed, and
respectively. The pore pressure gradient from 3000ft is 0.46 gas injeetion methods optimised.
psi/tl, and the well is drilled with an 8.3 ppg mixture of oil
water and viseositler. Gas is injeeted at a depth of 3000f? (at
the casing shoe) at a rate of 1 MMsefMay to create the Gas and Mixed Oil-Gas Kicks
underbalanee. Figures 14(a),@) give the degree of Previous versions of the simulator have been limited to
undertxdanee and casing pressure during the event, This also simulating a gas infhrx that oeeurs while drilling, in the
highlights the ditlkulty in obtaining a steady-state when there vicinity of the bit. Such a gas influx is still thought to be the
are multiple flowing zones. The choke pressure has to be single most dangerous well control incident, although we
adjusted three times before a pseudo steady-state is reached have also just seen that multiple gas intluxes whilst drilling
after approximately 70 reins. This difficulty is due to the fact underbalanced are also potentially hazardous. Gas-saturated
that the additional flowing zones have been kicked off by the oil influxes are also particularly common in deep HPHT
injected gas and the drilled gas (we are drilling with a rotary wells; the influx appears to be oil, but as it is circulated out of
BOP), At first the gas that they produce goes into solution. the well, the gas comes out of solution. If not properly
However, as their flow accelerates, free gas is produced. Once controlled, this too could lead to a major accident. Finally, a
these zones are obviously flowing, the injected gas is turned numlxr of brine kicks have &en observed. The simulator now
off and we use only the produced gas to maintain an models the effects of an influx of gas, oil and brine, including
underbalance. One hundred or so minutes into the simulation gas dissolved in oil or brine. In the following example, we
a significantly overpressured zone is penetrated and the highlight the differences between a pure gas kick, and a kick
outflow increases more suddenly. Suppose now that an consisting of a gas-saturated oil influx for similar geometries
attempt is made to try and increase the back pressure to and drilling conditions. The kicking zone is a 500mD, 15°A
control the flow whilst circulating through the choke. Figure porosity formation, and the fluid is driven into the wellbore in
14 (a) shows that if we increase the choke pressure to 300 psi each case by a 500 psi overpressure. The Gas-Oil ratio for
the flow starts to increase, but settles down into a different the oil-gas kick is 650 sef/STB, with the stock tank oil gravity
pseudo steady state after about 150mins. If we were to being 30° API. The gas is assumed to be methane, and we
increase the pressure to 350 psi we reeover the original compare the important well control parameters. We will
steady-state However, a back pressure of 400 psi (at the same consider that the kick is detected when the pit-level indicator
circulation rate of 252 gpm) leads to the well being is at 20 bbl. Taking into consideration the accelerating flow
overbalanced afler 180 reins. We also show what happens if as the circulating frictional pressure drops to zero as the
it is deeidcd to kill the well by the Wait and weight method. pumps are turned off and the BOP is shut, we end up with a
Initially a brief period of circulation is attempted at a back shut-in pit-gain of 25 and 37 bbl for the mixed and gas-only
pressure of 350 psi, However, we deeide that the well must be kicks respectively. As the kick is circulated to surface (we use
shut-in and killed with a mud weight calculated to be 8.7 ppg. the Wait and Weight method) we highlight the differences in
This example shows that existing well control techniques can the nature of the kick as it is circulated out. We note that we
still apply, even to fairly complex downhole flows, and even are using a pseudomil base mud that has an effective
when the kick was taken while drilling underbalaneed. If the volubility approximately 25°Aless than that for an equivalent
kick is severe, there maybe no way around having to kill the OBM having an effeetive oil volume fraction of 40 ‘%..
well, This, of course, destroys some of the benefits from In Figures 15(a) and (b) we compare the choke pressure
drilling underbalanc.cd, as it may lead to formation invasion, and pit-gain during the whoIe event. We highlight the
and lower eventual productivity. However, the safety of the following features:
operation is not compromised. Similar analyses can be ● Time of arrival at sufiaee - the gas stays in solution a lot

performed with the simulator for wells underbalanccd by the longer in the oil and soluble phase of the mud. Free gas only
method of drillpipe injeetion. In either case, with multiple occurs very late during the kill for the oil-gas kick; oil is at

147
8 I.M. REZMER-COOPER ET AL. SPE 38829

surface well before significant amounts of free gas appear. the ‘undenveight’ kill mud. An intermittent kill schedule was
● Early Kick Indicators - as expected the pure gas kick followed for the next 325 reins. Periods of circulation were
was shut-in earlier. It took a fiuther 20 reins until the mixed interspersed with periods when the well was shut-in and the
kick was detected. active pit weighted to 15 ppg. Eventually, the remaining
. Peak Choke pressure - Both peaks occur during the drillpipe pressure was bled off, the choke was opened and a
early stages. The gas kick has a much higher peak, consistent flowcheck established the well was dead.
with a greater volume of gas that is less distributed as it is Figure 16 shows the comparison of the computed and
circulated to surface. measured drillpipe and casing pressures. We note that the
. Maximum Surface Gas flowrate - again this is much total pit gain computed in the simulation is 54 bbl. The
higher for the all-gas kick and has a sharper peak as the computed stabilised shut-in pressures are correct to within
influx is less distributed (not shown). 5’Yo, and depend upon having the correct amount of gas,
expanded to the right volume, at the right point in the
wellbore. The gas was only 270ft from the surface when the
An Unconventional Kill BOP’s were shut. Continuing the simulation without shutting
In this last section we demonstrate how a complex model in shows that the gas would have been at surface about two
is required to accurately simulate a real field kick that minutes later. We can see from the simulation that the time of
occurred in a highly deviated well, and that included an formation shutoff has &en missed while allowing the shut-in
extended shut-in phase and a fragmented kill procedure. We pressures to stabilise. This led to the overprediction of the kill
note that it would be virtually impossible to obtain anything mud weight. We present in Figure 17 the results of the perfect
like a realistic representation of the kick and kill behaviour by Driller’s method kill. With the formation pressure later
the use of a simple model that does not have a detailed deduced from a posferiori measurements, a kill mud weight
representation of the multiphase fluid properties. of 12.3 ppg would have been sutllcient, and the gas could
Using the simulator to analyse a real event shows how have been removed in four hours shorter time. Thus, the
quickly a fairly straightfonvard kick can become a very simulator can highlight important safety factors, For
complex incident, and highlights the dangers in not following example:
well established well control procedures, A more detailed ● Misidentification of the formation flow shut-off, and the

discussion of the field kick discussed here will be presented in importance of obtaining a correct estimate of the SIDPP to
a future paper. optimise the correct kill mud weight.
The well control event simulated here lasted a total of . The importance of adhering to established well control
nearly 9.5 hours, making it a challenging test for the procedures.
simulator. One would expect simulator accuracy to decrease ● The importance of a fast, hard shut-in

with the length of time of the operation; we show that this is ● The importance of pit discipline and early kick
not necessarily the case. detection.
The influx occurred whilst drilling a 59° deviated section
of 12.5in hole at a measured depth of3314 ft (vertical depth
2474ft), in shallow water. For some time prior to the kick, Conclusions
average rates of penetration (ROP’s) were in the range of A versatile and validated tool for planning and post-event
200+00 Whr. Just before the kick was taken, peak ROP’s as analyses of the complex events that can occur in today’s wells
high as 520 ft/hr were recorded. Pit gain measurements were has been presented. It tackles more realistic scenarios than
obscured by additional flows into the main pit from two ever before, and can be used to highlight a number of
centritlges left running at the shaker, as well as flowback important procedural issues in complex well control
from the surface lines. Circulation continued for 5 reins afler scenarios.
drilling stopped so that an MWD survey could be made. It . Gas in mud can migrate fast in deviated wells
was only afler circulation stopped to make a connection that (potentially faster than an equivalent influx size in a vertkal
the well was observed to be flowing. The well was eventually well).
shut-in 16.5 reins after the start of the kick. The total kick ● In turbulent flow the annular pressure losses increase

volume was estimated to be 56 bbl, Initial shut-in pressures with increasing pipe rotation, in laminar flow annular
recorded on the drillpipe and casing were 200 psi and 480 pressure losses decrease with increasing rotation.
psi, stabilizing to 745 psi and 943 psi respectively afler ● The behaviour of the standpipe pressure during a kick in

approximately an hour. From the stabilised shut-in pressures, a slimhole can be the opposite of that in a regular-sized
a kill mud weight of 16.8 ppg was calculated. Insufllcient wellbore, with the kick being signaled by an increase in the
barite was on board, and so the active pit was weighted to 15 pump pressure in the narrow annulus.
ppg (original drilling mud weight was 10.7 ppg) and a new ● The characteristics of a severe kick taken while drilling

final circulating pressure was calculated to compensate for underbalanced can be dependent upon the method used to

148
SPE 36829 COMPLEX WELL CONTROL EVENTS ACCURATELY REPRESENTED BY AN ADVANCED KICK SIMUtATOR 9

generate the underbalance. A kick in a well that has been gas Kicks, SPE Drilling Engineering, Deeernber 1991.
drilled underbalanced may be complex, with influxes from 9. Johnson, A.B., and Cooper, S. :Gas Migration Velocities
During Kicks in Deviated Wells, paper SPE 26331.
more than one zone. Planning for such events is essential if
10 Zuber, N., and Findlay, J.A,: Average Volumetric
dangerous incidents are to be avoided.
Concentration in Two-Phase Flow Systems, Journal of Heat
● Mixed Gas-Oil Kicks can be difficult to detect early with
Transfer, pp453Jt68, November 1%5.
current detection equipment, Gas can beat surface much later 11 Roinrnetveit, R, Bach, OF., Liaaen, N.A., Maysounabe, P.,
than a conventional pure gas kick. Bjorkevolt, KS., Time, R.W., Hansen, S.A., Wang, Z., Hy-
● Current Mud-Gas Separator safety factors are adequate Billiot, J., Maglione, R., Gie, P.: Well Control in Horizontal
for safe sizing, but can be quantified with a more accurate Wells Studied in Realistic Experiments, paper presented at
representation of the surface gas flowrate that results from a the 1995 IADC Europearr Well Control Conference, Milan.
12 Luo, Y., Browrr, N,P., Loekyear, C., Hearn, P., and Bern,
worse case kick.
P:Gas Kicks in High angle and Horizontal Wells, paper
presented at the 1995 IADC European Well Control
Complex well control problems require a complex well Conference, Milan 1995.
centrol model. With foresight and contingency planning such 13 Rezmer-Cooper, I,M., Johnson, A.B., and Fitzgerald, P.:
simulators can be a powerfid tool for distilling the relevant “Factors Affecting Shut-in Pressure Rise”, paper presented at
and important information from which prmedural decisions the 1995 IADC Well Control Conference for the AsiafPacitic
ean be made. region, Jakarta, November 29-December 1.
14 Johnson, A,B., and Tarvin, J.A.:New Model Improves Gas
Migration Velocity Eslirnates in Shut-in Wells, Oils and Gas
Acknowledgments Journal, November 15.1993
15 Rezrner-Cooper, [M. and Johnson, A. B.: “Does the ‘Single
The enhancements described here form part of the European
Bubble’ Method Predict Worst Case Flowrates for Mud Gas
Union Thermie contract OG/1 18/94 PO; their support along Separator Sizing?, paper presented at the 1995 IADC Well
with BP Exploration is greatly acknowledged. The authors Control Confmence for Europe, Milan, June.
would also like to thank Simon Neild, Nicole Humphries, 16 Dawes, S.R., Smith, P.A., and Leach, C. P.: “Circulation of a
Emma Northrop, and Ben Willcocks for their detailed and Gas Kick - What Gas Flowrates Should We Expect? - What
patient validation of the simulator algorithms and Performance can We Expect From the Mud gas Separator”,
performance, Dean Tucker of Sedco Forex for the slimhole paper presented at the 1995 IADC Well Control Conference
well data, and all operators who have given feedback during for Europe, Milan, June.
the development of the simulator. 17 Hoopingarrrer, J.B., Grief, C. V., Neme, E.E., Rodt, GM., and
Bates, TR, Jr.: “Rig Mwlitications Meet New UK High
Pressure Requirements”, lADC/SPE 19976, 1990.
18 Low, E., and Jansen, C.: “A Method for Handling Gas Kicks
References safely in High Pressure Wells”, Journal of Petroleum
1. Leach, C.P.: “Examples From Actual Wells of How to Plan Technology, June 1993.
for Well Control Events Using a Realistic Simulator”, “, 19, Cartalos, U., and Dupuis D,: An Analysis Accounting for the
paper presented at the 1996 IADC Well Control Confmnce Combined Effect of Drillstring Rotation and Eccentricity on
for Europe, Aberdeen, May 22-24, Pressure Losses in Slinrhole Drilling, SPE/lADC 25769.
2. Leach, C. P., and Querrtin~.M.: How to Desigrr for Well 20 Ribeiro, P. B., Podio, A. L., and Sepehrnoori, K.:Ttre Effect of
Control Events, World Oil, June 1995. Rotational Speed and Eccentricity on AMular Flows with
3. Tarvin, J.A., Waltcm, 1. Wand, P. and White, D. B. Analysis of Appliwstion to Shrnhole Drilling, SPE 26958.
a Gas-Kick Taken in a Deep Well Drilled with Oil-Base Mud, 21, Marken, C. C., He, X., and Saasen, A.:The Influence of
paper SPE 22560. Drilling Conditions on annular Pressure Losses, paper SPE
4. Billingham, J., Thompson,M., and White, D. B.: Advanced 24598.
Influx Analysis Gives More Infxrnation Following a Kick, 22. McCann, R., Quigley, M. S., Zarnora, M., and Slater,
papa SPEJL4DC 25710 presented at the 1993 lADC/SPE S. :Effects of High Speed Pipe Rotation on Pressures in
Drilling Conference, Amsterdam Narrow Annuli. Paper SPE 26343.
5, Jardine, S.1., White, D. B., and Billinghanr, J.: Computer- 23. Falk, K., and McDonald, CM.: “An Overview of
aided Real-Time Kick Analysis and Control, SPWIADC UnderbaIaaced Drilling applications in Canada”, SPE 30129,
25711 (1993). 1995.
6. Grace, R.D., and Shursen, J.L.: “Field Examples of Gas
Migration Rates”, paper IADC/SPE 35119 presented at the
1996 IADC/SPE Drilling Confmence, New Orleans, Mmch
12-15.
7. Johnson, A., Rezrner-Ccxrper, I., Bailey, T,, and McCaM, D.:
“Gas Migration: Fast, Slow or Stopped”, paper SPE 29342
presented at the 1995 SPIMADC Drilling Conference,
Amsterdam, 28 February-2 March.
8. Johnson, A. B,, and White, D,B.: Gas Rise Veloeity During

149
10 I.M. REZMER-COOPER ET AL. SPE 38829

TABLE 1- FLOWLOOP PARAMETERS USED IN GAS TABLE 2 - PARAMETERS FOR SLIMHOLE WELL
MIGRATION EXPERIMENTS SIMULATIONS
Pipe Bore Centrebody Eccentricity Deviation Well Depth (R) 7500
(mm) Diameter (%) (degrees) Casing Shoe det)th (ft) 3000.
Casing Inner/Outer Diameter (in) 4.56/5.0
124 89 80 60,75,85,89 Drillpipe Length (ft) 7300
124 . . 30 Drill Collar Lenpth(ft) 200
200 89 98 5,15,45,60,75, Pipe Inner/Outer Diameter (in) 2.44/2.875
80,83,85,88 Collar Inner/Outer Diameter (in) 2.44/3.346
200 . 5,15,30,45,60, Bit Diameter (in) 3.78
75,80,83,85 Farm 3 (lbf1100sq.ft) 1
Farm 6 (lbf7100sq.11) 2
Farm 100 (lbf7100sq.fi) 7
Farm 200 (lbf/100 w@) 11
Farm 300 (lbf7100sq.ft) 14
Farm 600 (lbf/100 sq,tl) 24
Formation Overpressure (psi) 100
Formation Permeability (mD) 100

3om

1
—0%
.-.-2 %
----5 %

_._--. --------
———————

I I {, ., ., ., ,1, ,., .,.,.

n
0 10 20 30 40 50 td 70 80 90 I@) 110 120 130
Time(reins)
Fig. 1. Shut-in easing pressure rise for gas migrating in a
leak-tight wellbore, allowing for the suspended gas fraction,.

150
SPE 36629 COMPLEX WELL CONTROL EVENTS ACCURATELY REPRESENTED BY AN ADVANCED KICK SIMUIATOR 11

3. I---+=9
0.7

0,6
N
M

+
●
2- 0s

“K 0.4 o
mls
0.3 0
1. 0.2 1

L_LU—4J
o 102OI34O53LW 7080$U

0.0 0.4 0.8 1.2

DeviationFromVertical(deg)
%’h mls

~ 200 mm Pipe Fig.4, Summary of slip velocity, V,,p, derived from

● 124mm Pipe with 89 mm Centre Body
experimental data for air-Xanthum solution flows in pipe and
A 200 mm Pipe with 89 mm Centre Body
annular geometries.
Fig.2. Typical Zuber-Findlay plot at 75° deviation,

3.CHI
I I I 1 I

2.75 + Wldc Pipe

2.50 M Narrow Pipe

x wide
AMulus
22s a
o Narrow Amulus
;D o
21XI 0 0

1.75
I o x
1.s0 x
x 1 x
k x x.
$ +
1.25 L +

,.~
o 1$ w

Deviat~onfro;’ Vertic; (deg)75

Fig.3. Summary of Distribution factor, CO,derived from
experimental data for air-xanthum solution flows in pipe and
annular geometries.

151
12 I.M. REZMER-COOPER ET AL. SPE 36S29

M@

— Rigidwellbore
—- 5xlo-’0Pi’
—- Ioxlo-’”
Pa-’
# *W

‘t
&-&m

o I I I
10 20 30 40 50 a

Time (reins)

Fig.5. Shut-in casing pressure for gas migrating in a leak-tight wellbore, aIlowing for the wellbore compliance.
(a)

cm, I 1 1 I

-
.- 503
&
t

,~
o 150 2CM

Time’~mins)

@)
60

45

30

Is

---------
-15 1.1. 1.>1.
0 40 160 200
TYme(mi;sO)

Fig.6.Circulating out a kick with and without fluid loss: (a) Choke pressure, and (b) pit-gain,

152
SPE 30029 COMPLEX WELL CONTROL EVENTS ACCURATELY REPRESENTED BY AN ADVANCED KICK SIMULATOR 13

6.0
‘-[ ,/
//’
/0’ ,//;,//,
......
4.0
......----””--
,/
2.0 &

0.0
0.0 1.0 2.0 3.0 4.0

Slow Circulation Rate (bbl/min)

Fig.7. Surface gas flowrates as a function of the slow circulating rate,using different models. Industry standard refers to the single
bubble model.,
2700
I 1 I I I I I I I
_-
~ 2660
a
; 2660

fit

I I I 1 I I I I I I
2640 I
ms162&254s9 S657 *5060

Time (rein)

12
I I I I I I I I I

10

&

0
60tl S2L’1 6466 W67686060

Time (rein)
Fig.8. Relative timings of (1) Gas first at surface, (2) Maximum surface casing pressure and, (3) Peak surface gas outllow rate as gas
is circulated out at 4 bbl/min,.

153
14 I.M. REZMER-COOPER ET AL. SPE 30S29

Axial Flow
(Reynolds Number)
I I I I

TURBULENT
FLOW .............. ..
-.,..#.. ----
..”.
,,..,’-
.,,,. TURBULENT
\ .!” FLOW +
VORTICES

~ LAMINAR+W)RTICES
I I 1 I
o 50 102 1s0 2(KI 250
Rotation (Taylor Number)
Time(reins)

Fig.9.Typical flow regimes in slimhole drilling as a finction

of the Reynolds number and Taylor number. Fig. 11 .The Effects of dnllpiW rotation on annular pressu~
losses in a slim annulus. Rotation is increased from 0-600
rpm in 100 rpm increments every 30 reins.

/L..,
I i! k.!u!l

I
e
❑
0

+Orpm

Mlcolpln

x200rp10
+ o
03001p01 0 50 100 150 203 250

II 04wqnn

05wlpn
Time (reins)

-0.0 0,5 1.0 I .5 2.0 2.5 3,0

Fig. 12. The effect of kick size on the maximum casing
Axial Velocity (m/s)
pressure for a distributed kick in a slim annulus.

Fig. 10. Experimental pressure drops as a function of the

rotation rate.

154
SPE 36829 COMPLEX WELL CONTROL EVENTS ACCURATELY REPRESENTED BY AN ADVANCED KICK SIMULATOR 15

3r33
I 1 I 1 I 1

1 I I I
km
o 5 10 Is m 2s

Time (reins) (a)

Fig. 13. Standpipe pressure for kick in slimhoIe (3,78in).

! I ..-
.-.
,’ 7”:
-. W”I,
.!
---------------

—Wopsl
--..350 p
—-400p61 Time (reins)
-. .Wellklu
( I 1 I 1’ I
3 50 02 I 50 200 250 Km 3s0

(b) ?0

60
1203 1 I 1 T
r I I 1
50

40

30

20

10

0
0 20 40 60 80 lCJ2 no 140 Iso 180
-.. Time (reins)
-. -----

01 I I I I I I J
0 30 {Go 230 ml 350

Ti’;e (m~s)
Fig. 15. Comparison between pure gas and mixed oil-gas
kicks in a pseudo oil-base mud, (a) Casing pressure, (b) Pit-
Fig,14. Parasite string underbalanced drilling: (a). Degree of gain,
underbalance (shown by a negative value), (b).Choke
pressure.

155
16 I.M. REZMER-COOPER ET AL. SPE 36829

(a) (a)

2oa3
I’1’l’l’ 1’
~

&
“ 18@l

g ,m :

I
~ ,~ -
a)

2 ~—----=- . .... . .... .

~ 1200 -
\
,
,? ,m
— Simulah

g — “Pufwt”C3rilla’s KiU .

I I 1 1 1 I I 1 1 I I I 1 1 I I
803
0s0 lal 150 202 2X) 31XI 3$0 4ml 45U SC@ 550 o lal 2rK3 m soil 6CQ

Time (reins) Time3~mins)

(b)

121m ,
I I I I I i

~, I I I I 1 I I 1 1
o I n
050100 I 50 202 230 300 3$3 W 450 m 5X3 0 Ictl 2CN 3(I3 400 5C0 KQ

Time(reins) Time (reins)

Fig.16. Comparison between actual field data and model Fig.17. Comparison between ‘Perfect’ or Idealised Driller’s
simulation for a kick in a water-base mud. (a) Standpipe Method kill of field kick shown in Fig. 16 and a simulation of
pressure, (b) Choke Pressure. the actual kill. (a) Casing shoe pressure; (b) Mass of gas.

156