Professional Documents
Culture Documents
Fallah Dissertation 2021
Fallah Dissertation 2021
Fallah Dissertation 2021
by
AmirHossein Fallah
2021
The Dissertation Committee for AmirHossein Fallah Certifies that this is the
approved version of the following Dissertation:
Committee:
Vaibhav Bahadur
Yaguo Wang
by
AmirHossein Fallah
Dissertation
Presented to the Faculty of the Graduate School of
in Partial Fulfillment
of the Requirements
Doctor of Philosophy
First and foremost, I would like to thank my supervisor, Dr. Dongmei Chen for
giving me the opportunity to join her research group and for guiding me throughout my
PhD endeavor. Without her feedback, support, guidance, and encouragement, this
I would like to thank RAPID consortium within the University of Texas at Austin
for the financial support of this project. I would like to specifically thank Dr. Ali Karimi
Vajargah, Dr. Eric van Oort, and Dr. Pradeepkumar Ashok for their inputs and help
Last but not least, I wish to thank my family and friends for believing in me any
time I felt hopeless, encouraging me any time I was lost, and assisting me any time I needed
iv
Abstract
and is more reliable than its clean energy competitors. To generate geothermal energy at a
necessary. Current drilling technology expertise in the field of oil & gas such as managed
pressure drilling (MPD) is currently used to drill challenging extended reach drilling (ERD)
and high-pressure high temperature (HPHT) wells under extreme conditions. Applying
and globally scalable energy generation. However due to the complexities associated with
HPHT conditions and extended lateral sections, efficient drilling and reliable well control
during the drilling and operation of geothermal wells is challenging and requires a
flow.
transient governing equations, allowing the use of small time-steps to simulate automated
drilling and well control operations in real-time. To apply the model to the wide range of
scenarios observed in drilling and operation of DCLG wells, various sub-models are
developed and integrated that simulate complex phenomena such as gas solubility and
break-out, cuttings settling and bed blockage, rock formation cooling, temperature-
dependent fluid properties, pressure waves and automated MPD choke control, etc.
The developed modeling approach is applied to various scenarios during the drilling
and operation of deep CLGS (DCLGS) wells. The following applications are explored in
this work: (1) Reservoir influx detection and automated gas kick control under HPHT
conditions, which is governed by liquid-gas multiphase flow; (2) Hole cleaning and
multiphase flow; and (3) Operation and performance analysis of DCLGS wells with
vi
Table of Contents
List of Tables .......................................................................................................................x
1.1 Motivation.............................................................................................................1
1.2.1 Wellbore Hydraulic Modeling for Kick Detection (Drilling Stage) ......5
3.4 Conclusions.........................................................................................................82
viii
4.4 Simulation Scenarios ..........................................................................................95
4.5 Conclusions.......................................................................................................107
5.3 Conclusions.......................................................................................................129
Glossary ...........................................................................................................................139
References ........................................................................................................................141
ix
List of Tables
Table 2-2. Pipe, casing, and cement sizes used in the dynamic kick control
scenario. ........................................................................................................41
Table 4-1. Relevant geometrical, formation, fluid and geothermal parameters used
x
List of Figures
Figure 1-1. Multi-phase simulation models used in present day drilling applications
Figure 1-2. Key variables controlling cuttings transport in deviated wells (after
Figure 1-4. HPHT tiers and operating range for deep CLGS wells (well information
Figure 2-1. 5 Different cases of the heat transfer network for an offshore well. For
Figure 2-2. Nusselt number for laminar annulus flow (from Nellis and Klein [77]). .....26
Figure 2-3. Solution procedure of the semi-implicit numerical scheme at each time-
step. ...............................................................................................................29
Figure 2-4. Discretization of rock formation for heat transfer calculations (with bi-
Figure 2-5. Steady-state temperature profile compared to the results of Hasan and
Kabir model and commercial software. ........................................................33
Figure 2-6. Drillstring, wellbore and formation nodes' temperature profile at steady-
state. ..............................................................................................................34
Figure 2-8. Bottomhole temperature versus time compared against the results of
Figure 2-9. Wellbore schematic and configuration used in controller tuning. ................37
xi
Figure 2-10. Choke opening versus time. ..........................................................................38
Figure 2-15. Deviated well used in the dynamic kick control scenario.............................41
Figure 2-16. Initial temperature profile for the dynamic kick control scenario. ...............42
Figure 2-18. Comparison of the dissolved gas density (DG) and the solubility
Figure 2-19. Comparison of the gas volume fractions at different times. .........................46
Figure 2-20. (a) BHP, (b) surface backpressure and (c) SPP versus time. ........................47
Figure 2-21. (a) flow out, (b) pit gain and (c) choke opening plots. .................................48
Figure 2-23. Comparison of the dissolved gas density (DG) and the solubility
Figure 2-24. Comparison of the gas volume fractions at different times. .........................51
Figure 2-25. (a) BHP, (b) surface backpressure and (c) SPP versus time. ........................51
Figure 2-26. (a) flow out, (b) pit gain and (c) choke opening plots. .................................52
Figure 3-1. Solution procedure of the semi-implicit numerical scheme at each time-
step. ...............................................................................................................63
Figure 3-2. Well inclination and simulation parameters used in case study 1. ...............66
xii
Figure 3-3. Steady-state bed blockage, cuttings concentration, annular pressure, and
Figure 3-4. Effect of pipe rotation on bed blockage (case 1A). ......................................68
Figure 3-5. Cuttings concentration in suspension during the clean-up cycle (case
1A). ...............................................................................................................70
Figure 3-6. Cuttings mass out and BHP during the clean-up cycle (case 1A). ...............70
Figure 3-7. Bed blockage, cuttings concentration, annular pressure, and velocity
Figure 3-8. Cuttings mass out and BHP during the clean-up cycle (case 1B).................72
Figure 3-9. Well inclination and simulation parameters used in case study 2. ...............73
Figure 3-10. Mud and cuttings velocity profiles for different pump rates (case 2). ..........74
Figure 3-11. Steady-state bed blockage, cuttings concentration, and annular pressure
Figure 3-12. Well inclination and simulation parameters used in case studies 3 and 4. ...76
Figure 3-13. Steady-state bed blockage, cuttings concentration, annular pressure, and
Figure 3-14. Cuttings mass out and BHP during the clean-up cycle (case 3). ..................78
Figure 3-15. Steady-state bed blockage, cuttings concentration, annular pressure, and
Figure 3-16. Cuttings concentration profile at different times during the MPD-
Figure 3-17. Cuttings concentration profile at different times during the MPD-
xiii
Figure 4-1. Schematic of the MPO-controlled deep geothermal well concept,
including (1) inlet section (cased), (2) lateral section (open-hole), (3)
Figure 4-2. Geometric discretization of rock formation for heat transfer calculations
Figure 4-3. Comparison of the radial rock temperature profiles at 1 day and 5 days
Figure 4-4. Temperature profile at different times for the “7/7” geometry (with
Figure 4-5. Temperature profile at different times for the “10/5” geometry (with
Figure 4-6. Pressure profile at different times for geometries (with open-hole
lateral). ........................................................................................................100
Figure 4-7. Outlet temperature, thermal power and flow out versus time for both
Figure 4-8. SBP and choke opening versus time for both geometries (with open-hole
lateral). ........................................................................................................103
Figure 4-9. Temperature profile at different times for the “7/7” geometry (with
Figure 4-10. Temperature profile at different times for the “10/5” geometry (with
Figure 4-11. Outlet temperature and thermal power versus time for both geometries
Figure 4-12. Comparison of the temperature profiles for different pump rates. .............106
xiv
Figure 4-13. Comparison of the outlet temperature and the generated power for
Figure 5-1. J-shaped closed-loop design with the integrated MPO controller. .............110
Figure 5-2. Temperature profile at different times for the U-shaped design. ................113
Figure 5-3. Temperature profile at different times for the J-shaped design. .................114
Figure 5-4. Outlet temperature, thermal power, choke opening, and surface back
Figure 5-5. Temperature (left) and thermal power (right) at 10 h versus total well
Figure 5-6. Outlet temperature (left) and thermal power (right) versus TVD for
Figure 5-7. Temperature (left) and thermal power (right) versus pump rate for
wellbores. ....................................................................................................122
Figure 5-9. Comparison of the outlet temperature and the generated power for
Figure 5-10. Fluid temperature profile at different times for 21.59, 25.08, and 31.12
cm wellbores. ..............................................................................................124
Figure 5-11. Outlet temperature, thermal power, and flow out rate for 21.59, 25.08,
Figure 5-13. Fluid temperature profile at different times for 200, 250, and 300 ℃
xvi
Chapter 1: Introduction & Literature Review
1.1 MOTIVATION
More complex wells are being drilled every day, and with the improvements in
drilling technology, reservoirs that once were assumed to be “undrillable” are becoming
accessible, reducing the cost and issues associated with drilling. Recent improvements in
managed pressure drilling (MPD) techniques enables accurate pressure control in tight
drilling margins and extreme conditions. While state-of-the-art drilling techniques are
being used daily to drill oil & gas wells, little attention has been paid to the geothermal
industry, where the same drilling expertise could be leveraged to drill deep geothermal
While geothermal energy is considered green energy, there are issues associated
with the current geothermal technology, i.e., enhanced geothermal systems (EGS). In EGS
wells, fluid is pumped into the ground through an injection well and is returned to surface
through a production well. The wells are connected through a matrix of natural or induced
fractures, where the flow through this fracture network provides sufficient time and area
for heat transfer. Due to the dependence on the fracture network, however, EGS systems
have technical and environmental risks such as surface gas emissions, short circuiting, and
induced seismicity [2–4]. Earthquakes have also been observed in Switzerland and South
Korea due to the failure of EGS power plants [5,6]. Closed-loop geothermal systems
(CLGS) wells have been proposed as an alternative, where the heat transfer occurs across
and extended lateral section in a high temperature formation, allowing for sufficient heat
transfer to the circulating fluid [7]. Companies such as Eavor Technologies and GreenFire
1
Various geothermal power plants are available that are designed to maximize the
electricity generation based on the outlet fluid temperature. Power plants that use water as
the working fluid are divided into dry steam, flash steam, and binary plants [10,11]. Dry
steam and flash steam power plants directly use steam as the working fluid of the turbine,
and therefore require water temperatures of above 180 ℃. Since typical operational
temperature of CLGS wells is below 180 ℃, binary plants are used for electricity
generation where the hot water/steam from the well is entered into a heat exchanger to heat
another fluid with a low evaporation point. This fluid is then circulated into a turbine for
electricity generation. Cool water from the heat exchanger is also pumped back into the
well for continuous heat transfer from the high-temperature rock formations downhole.
source of clean energy, more work needs to be done to maximize the performance and
power generation of these systems. Potential power generation of CLGS designs need to
maximize the energy production. Binary power plants normally have a thermal efficiency
where 𝑊̇ is the electrical power, and 𝑄̇𝑖𝑛 is the heat received by the working fluid (either
the outlet water from the CLGS well or the inlet working fluid into the turbine). Here, the
net thermal power is used to study the performance of the wellbore. Thermal power is
2
defined as the heat difference between the outlet and inlet of the geothermal well, which is
Moreover, due to the high cost of drilling and maintaining deep high-pressure high-
need to be addressed to avoid operational drilling issues, and to minimize the cost of
drilling. While drilling technology in the field of oil & gas certainly benefits exploration
temperatures and ensure well control during drilling and operation of CLGS wells. These
models must consider various phenomena observed in drilling and well control, such as
gas influx, solubility and break-out, cuttings bed formation and hole cleaning challenges,
geometry and area discontinuities, thermal cooling of the surrounding rock formation,
dynamic well control and pressure wave dynamics in the well, temperature-dependent fluid
properties, rapid and automated choke adjustments, etc. Current models used in the oil &
gas and geothermal industries are overly simplified and cannot simulate novel drilling and
production techniques accurately. More sophisticated models are needed to reduce the
drilling time and cost, minimize drilling hazards, ensure the safety of the drilling crew,
3
equipment, and the environment, and provide real-time information for well planning,
simulate various drilling and operation applications of deep CLGS wells. The model is
adjusted for each application to provide maximum capability and performance in real-time.
There are three main topics explored in this dissertation (the first two topics could be
applied to the drilling of any well, and the last topic is specific to the production stage of
novel MPD controllers, and calculate important parameters such as the kick
tolerance.
circulating temperature and generated power, control the phase behavior of the
These tasks are thoroughly studied in the following chapters, and simulation results
In this section, the literature review of the current work is presented. The literature
review is divided into 3 different sections, covering the drilling and operation of deep
geothermal wells. These sections are: (1) wellbore hydraulic modeling for kick detection,
4
(2) cuttings transport modeling, and (3) hydraulic simulation of closed-loop geothermal
systems.
MPD techniques are currently widely used on various onshore and offshore wells
to precisely control the bottom-hole equivalent circulating density (ECD) and the annular
pressure profile along the wellbore during circulating conditions [13]. Since it is not
practically feasible to have downhole sensors along the wellbore (unless more sophisticated
and costly technologies such as wired drillpipe are used [14]), the knowledge of downhole
parameters is limited. Due to this limitation, there is a growing need for comprehensive
multi-phase hydraulic models that can accurately take into account complex well
geometries, HPHT conditions, and complex well control situations and the fast transient
events and phenomena associated with them, e.g., riser gas unloading, automated choke
During kick control, gas influx travels uphole at a higher velocity than the
surrounding drilling mud because of the lower density. Moreover, pressure dynamics in
the wellbore affect various parameters during drilling operations, such as mud and gas
densities, reservoir influx, fluid loss at the fractures, and gas solubility and break-out.
Temperature also affects fluid density and gas solubility in HPHT wells, and valid
estimations of the dynamic temperature profiles within the wellbore and the surrounding
rock formation are required for guiding appropriate well control actions. Hence,
To date, several models have been developed with a wide range of capabilities and
sophistication to simulate fluid flow through the wellbore. Aarsnes et al. [15] provided a
5
review of these models and categorized them according to their mathematical structure and
level of sophistication. The developed models have often neglected temperature effects and
simplification aims to reduce the computational expense for real-time control design.
However, since temperature affects several other crucial system parameters such as
pressure, this simplification may result in an inaccurate estimation of the ECD and non-
thermal stress variations due to mud circulation and associated cooling or heating [18,19]
and the break-out point of dissolved gas in non-aqueous drilling fluids during a well control
that incorporate suitable thermal models are routinely used. An example is the model
developed by Hasan and Kabir [20], which provides an analytical solution for the steady-
state temperature profile in the wellbore. Such models simplify the momentum and mass
conservation equations, and thus are not suitable for dynamic temperature prediction
during a gas kick scenario due to varying velocity and pressure profiles along the wellbore
and their effects on the temperature profile. On the other hand, Two-Fluid Models (TFMs)
Bendiksen et al. [21] is arguably one of the most sophisticated models in this respect. Other
examples include the models developed by Goldszal et al. [22] and Yin et al. [23]. The
implicit numerical scheme, separate momentum equation for each phase (characteristic of
multi-fluid models) and sophisticated closure relations used in these models enables them
to simulate transient multi-phase flow in pipes and annuli with minimal error. However,
the level of complexity involved with the two (or multi)-fluid modeling assumption comes
with a high computational cost that makes such models impractical for real-time control
6
applications. Moreover, such models may not be numerically stable for all cases (e.g. in
case of high gas fractions) [24], and currently lack the functionality to simulate MPD
operations due to their slow numerical schemes. Also, the TFMs are more suitable for
segregated flow regimes such as “stratified flow” which rarely occur in drilling operations.
Since early gas kick detection is an inherent capability of MPD systems that track inflow
into - and outflow from - the well, most kicks are relatively small in size, such that the
anticipated flow regime is “bubble flow” or “dispersed bubble flow”. This makes TFMs
Drift-Flux Models (DFMs) have been developed to address the need for
The use of a mixture momentum equation with an integrated slip law makes DFMs
computationally cheaper and more stable compared to the two-fluid modeling approach.
However, DFMs that incorporate the energy equation are rare in the published literature.
One exception is the model developed by Petersen et al. [25], which uses a de-coupled
energy equation to solve for the 2-D temperature profile (in axial and radial directions) in
a well. However, their focus was on the pressure estimation of the problem, whereas
validation and results of their thermal model were not provided. This makes it impossible
to verify the temperature estimations independently and evaluate the validity of the various
simplifying assumptions that were made, such as using the steady-state solution for the
radial temperature profile. The models developed by Sun et al. [26] and Xu et al. [27] are
other examples of DFMs that take into account heat transfer effects. However, using
constant density and simplified viscosity models for the mud, their models are less accurate
for estimating the downhole parameters due to the dependence of density and viscosity on
pressure, temperature, and flow conditions. Moreover, the incorporated fully implicit
numerical schemes in these models come with high computational costs and larger time-
7
steps that neglect pressure wave dynamics. As a result, while these models are capable of
simulating conventional kick control methods, they cannot model automated choke control
and capture the fast transients during dynamic kick control practices and MPD operations.
Figure 1-1 provides a comprehensive comparison of the published DFMs that account for
Figure 1-1. Multi-phase simulation models used in present day drilling applications and
comparison of available DFM-WE approaches [28,29].
8
The tree graph in Figure 1-1 shows different multi-phase flow modeling approaches
used in drilling applications. The top row shows the models With Energy Equation (WE),
and the bottom row shows the same models Without Energy Equation (WOE). The
complexity and numerical cost decrease from left to right, through incorporating a mixture
momentum equation in DFMs, and relaxing the pressure dynamics in Reduced DFMs
(RDFMs).
Temperature variations and their effect on kick control during MPD operations are
not addressed in the literature and current DFM-WEs use implicit numerical schemes,
making them incapable of simulating automated choke control. A hydraulic model was
non-aqueous drilling fluids [32], the use of non-Newtonian fluids, arbitrary 3-D well paths,
and area discontinuities in the drillstring and wellbore. However, the developed model did
not take the temperature variations into account, making the estimations of density,
conditions.
To avoid hole cleaning problems such as slow rate of penetration, tripping and
cementing issues, stuck pipe, excessive pressures, etc., cuttings need to be effectively
removed from the borehole. Cuttings have a higher density than the drilling fluid and are
therefore transported at lower relative velocities in relation to the mud. Since it is generally
challenging and costly to place sensors along the wellbore to measure the concentration
and location of solid particles in the well, proper hole cleaning of deep wells presents a
significant challenge. This issue is more pronounced for deviated and horizontal wells with
9
longer lateral sections, where the cuttings particles tend to drop to the low side of the
borehole under the influence of gravity. If the mud velocity is not sufficient to keep the
cuttings in suspension and move them uphole, the cuttings will settle down and form a
cuttings bed [33]. The cuttings bed does not move with the mud flow and partially blocks
the annular cross-section, leading to a higher mud velocity through the unblocked region.
Once a cuttings bed begins to build up, it will keep increasing until the fluid velocity
reaches a level that is high enough to keep the solid particles in suspension and avoid more
settling. This velocity is known as the critical transport fluid velocity (CTFV) [34].
Several factors affect cuttings transport along the well, including but not limited to
hole and pipe sizes, pipe rotation, pipe eccentricity, mud rheology, cuttings size and
density, flow rate, etc. In vertical wells, cuttings are transported to the surface by
overcoming the slip velocity between the solid particles and the mud. In such wells, the
flow rate (which determines the annular velocity) and mud rheology are the leading factors
in cuttings transport. Due to the bed formation in deviated and horizontal wells, the flow
rate, pipe rotation, and pipe eccentricity play critical roles in hole cleaning. Figure 1-2
shows the effect of various parameters on hole cleaning of the deviated wells and how
much control the drilling crew has over each parameter [35]. While some variables are not
controllable, the flow rate, pipe rotation and rate of penetration (ROP) can be controlled in
10
Figure 1-2. Key variables controlling cuttings transport in deviated wells (after Adari et
al. [35]).
parameters on hole cleaning process, most importantly the bed height and the slip velocity.
For example, Shah et al. [36] and Kelessidis and Mpandelis [37] present experimentally
developed models for settling velocity of solid particles in non-Newtonian fluids. Many
experimental models have also been proposed to estimate the CTFV and the bed height in
deviated wells [34,38–40]. Xiaofeng et al. [41] provide a review and comparison of such
models. While these experimental studies provide basic understanding of the steady-state
conditions, they fail to provide real-time information on transient bed formation and
cuttings flow, which are crucial for effective hole cleaning in actual practice. It is therefore
desirable to develop multi-phase flow models that can describe the cuttings transport
dynamics and estimate the cuttings bed height and cuttings concentration reliably.
11
Developing a multi-phase flow model to describe the cuttings and mud is complex
due to the intricate forces between the phases, particle settling, collision between the
particles, pipe eccentricity and hole size effects, etc. Computational fluid dynamics (CFD)
models have been proposed to simulate the cuttings transport considering the collision
dynamics [42]. Another example is the approach by Erge and van Oort [43] where the
annular velocity profile is used to determine the cuttings bed height. These models provide
2D/3D estimations of the cuttings bed and are practical for analyzing specific sections of
the well, as well as determining the optimal parameters for efficient hole cleaning.
However, the high computational cost associated with these models generally prevents
them from being implemented in real-time. Therefore, they are not suitable for estimating
the dynamic solid concentration and pressure profiles along the entire wellbore during
proposed that rely on experimentally developed sub-models to estimate the bed height and
slip velocity. A review of such models is provided by Nazari et al. [44]. Due to the one-
dimensional discretization of the wellbore and the use of experimental formulas, these
models can simulate the entire wellbore with a much-reduced computational cost.
However, automated choke control is neglected in these models, making them incapable
of simulating the hole cleaning under challenging MPD operations. Moreover, many of
these models make simplifying assumptions, such as incompressible mud, limited range of
fluid rheology, neglecting the effect of pipe rotation, etc. For instance, the proposed model
of Naganawa et al. [45] is developed based on a semi-implicit numerical scheme for robust
simulations. However, the assumption of incompressible drilling fluid and neglecting the
effect of mud rheology on slip velocity affect the accuracy of estimating the pressure profile
and cuttings transport. Another example is the model developed by Cayeux et al. [46]. This
12
model includes the effect of mud compressibility and expansion. However, the numerical
solution procedure is not provided in the published work, making it difficult to verify the
fully transient behavior of the model and the capability to be integrated with automatic
controllers. Furthermore, the inclusion of the energy equation results in high computational
costs that limits the real-time simulation performance of the model. Table 1-1 compares
the modeling approach and capabilities of the cuttings transport models mentioned above
Erge and van Akhshik et al. Naganawa et Cayeux et al. This model
Oort (2020) (2015) al. (2017) (2014) (Chapter 3)
Space 3D 3D 1D 1D 1D
Discretization discretization discretization discretization discretization discretization
Numerical Experimental Experimental
modeling of bed Numerical Numerical sub-models sub-models
Cuttings Bed
blockage using modeling of modeling of used for bed used for bed
Modeling
experimental bed blockage bed blockage blockage blockage
correlations estimation estimation
Geothermal energy is considered clean and renewable, and geothermal wells have
been drilled and used for heat and electricity generation for many years [2,47,48].
13
However, the economical extraction of geothermal energy has been mostly limited to
countries and regions with high subsurface temperature gradients (e.g. areas with active
volcanism and tectonic activity) and permeable aquifers [3,49]. Conventional enhanced
geothermal systems (EGS) require high rock porosity and permeability, sufficient fluid in
place, and adequate fluid recharge, which are not always available [48,50]. Moreover, the
direct contact between fluid and rock through fractures might cause problems such as fluid
contamination, surface gas emissions, and induced seismicity [51–53]. To avoid these
geothermal systems (CLGS) were introduced [7]. CLGS wells essentially work as an
indirect heat exchanger, where the circulating fluid absorbs energy from the formation rock
as it flows through the well, eliminating the reliance on fractures and the associated issues
Different CLGS designs have been proposed in the published literature [54]. These
include pipe-and-annulus designs [9] and the U-shaped loops [8]. Abandoned oil and gas
wells are also considered for geothermal energy production [55–57]. However, the
proposed concepts are mostly based on shallow wells and the estimated electricity
generation in such wells is on the order of a few megawatts only, which may not be
sufficient to generate electricity on a large utility scale (see e.g. Oldenburg et al. [58]). To
generate commercially viable power at any location, deep CLGS (DCLGS) wells are
study by the Geothermal Laboratory at SMU [59] shows that geothermal reservoirs with
temperatures above 200 ℃ at the depth of 7.5 km are available in most regions within the
US, most prominently in the western states and in south-east Texas. Similar geothermal
temperatures are observed in Australia and Europe as well [60,61]. Thus, to access such
reservoirs, it is necessary to drill wells that are deeper than existing wells.
14
Figure 1-3. Geothermal temperatures at 7.5 km within the US (courtesy of SMU
Geothermal Laboratory [59]).
Drilling deeper CLGS wells with longer lateral sections is technically difficult due
to the extreme HPHT conditions, the challenges associated with accurate directional
drilling at depth, well integrity issues, etc. However, technologies already developed in the
oil and gas sector have the potential to meet these challenges and can be extended to deep
geothermal drilling [1,62–64]. Note that deep vertical wells have already been drilled up
to 12 km true vertical depth (TVD). Two examples are the deep vertical wells drilled in
Russia at the Kola Peninsula (12+ km) [65] and the KTB site in Germany (9+ km) [66].
Directional drilling techniques have also enabled drilling of very long horizontal sections
[67,68]. Wells with total Measured Depths (MD) of up to 15 km have been drilled using
be needed to drill deep geothermal ERD wells. However, given historical HPHT drilling
challenge. As of 2012, multiple oil and gas exploration wells have been drilled in the
15
operating range of the proposed deep CLGS, with in-situ temperature up to 290 ℃ (Figure
1-4). Geothermal wells with temperatures up to 450 ℃ have also been drilled, although
these wells are typically less than 5 km deep [70]. It is assumed that deep directional
drilling of CLGS wells is technically possible in the not-too-distant future (i.e. within a
Figure 1-4. HPHT tiers and operating range for deep CLGS wells (well information
courtesy of Total, 2012 [62]).
multi-phase flow that allows for automated MPD control. Validation of the
16
• Chapter 3 covers hole cleaning modeling during horizontal well drilling
estimates the status of the bed blockage and cuttings concentration in real-
time. Multiple case studies are analyzed that cover a wide range of the
insulation of the return flow. A coupled wellbore hydraulics and rock heat
geothermal system.
17
Chapter 2: Modeling of Wellbore Hydraulics1
2.1 INTRODUCTION
hydraulic modeling framework is presented. This integrated model can estimate the mud
temperature in the drillstring and the annulus, as well as the temperature variations of the
formation during complex well control situations. The novelty here is that the integrated
thermal and hydraulic model considers the effects of the dynamic temperature profiles
when estimating different parameters during conventional and MPD drilling operations,
such as liquid and gas densities, ECD and pressure profiles, gas solubility and break-out,
etc. Moreover, the thermal model takes into account the temperature variations within the
surrounding rock formation, and the effect of rock cooling/heating on the annulus and
used to simulate automated choke control and other MPD functionality. The semi-implicit
discretization allows for fast but accurate simulations that capture the effects of pressure
wave dynamics and the fast transients associated with dynamic well control, enabling
automated choke control during MPD simulations. This is currently missing from the
DFM-WEs in the published literature. The integrated model can also be used for
conventional kick control, in case a large kick size requires shutting in the well on blow-
out preventers [72]. Moreover, for a given well geometry and fluid rheology, different kick
sizes can be simulated to estimate the maximum pressure at the casing shoe and thereby
1 The work in this chapter is previously published [89]. Q. Gu, Z. Ma, A. Karimi Vajargah, D. Chen, P.
Ashok, and E. van Oort contributed to the model development and manuscript revision of the published
article.
18
Details of the mathematical modeling are presented in this chapter, including the
formulation of the conservation equations; the thermal resistance network between the
drillstring and annulus flows, and between the annulus flow and the rock formation; the
heat transfer network within the rock formation; the details of the solution procedure used
to solve the flow parameters; and the heat transfer network within the rock formation. The
thermal model is validated against analytical solvers and commercial software. The
performance and results of the developed automatic controllers are provided to show
efficient MPD well control. Finally, MPD well control scenarios are simulated to show the
effect of temperature variations in gas solubility and break-out during kick control.
dimensional representation of the DFM along the wellbore, are provided with a focus on
the incorporation of the energy equation. The details of the mass and momentum equations
are provided elsewhere [30]. The following assumptions are made in developing the
thermal model:
variations during a single time-step (on the order of 10-2 s) are negligible.
• Heat storage within the drillpipe, casing and cement layers is neglected due
to the small thickness of these layers. These layers are modeled as surfaces
through which heat is conducted between the annulus and drillstring flows,
19
• While the flow temperature is modeled in 1D, the surrounding formation
gradients.
• Gas and liquid phases are assumed to be at thermal equilibrium at each point
along the well, i.e. there is one temperature to be solved from the mixture
energy equation.
equation for each element (liquid or gas), a mixture momentum conservation equation, and
a de-coupled mixture energy equation, which is provided in the set of Equations 2-1 - 2-3
below. In the energy equation, all the phases are assumed to be at the same temperature,
i.e. thermal equilibrium is assumed between the phases at the same depth; hence, the only
𝜕 ∑𝑛
𝑖=1 𝛼𝑖 𝜌𝑖 𝑣𝑖 𝜕 ∑𝑛 2
𝑖=1 𝛼𝑖 𝜌𝑖 𝑣𝑖 𝜕𝑝
+ = − 𝜕𝑥 + 𝑓𝑔 + 𝑓𝑤 + 𝑀̇𝑠𝑜𝑢𝑟𝑐𝑒 ································ (2-2)
𝜕𝑡 𝜕𝑥
1 1
𝜕 ∑𝑛 2
𝑖=1 𝛼𝑖 𝜌𝑖 (𝑒𝑖 + 𝑣𝑖 +𝑔𝑧) 𝜕 ∑𝑛 2
𝑖=1 𝛼𝑖 𝜌𝑖 𝑣𝑖 (ℎ𝑖 + 𝑣𝑖 +𝑔𝑧)
2 2
+ =
𝜕𝑡 𝜕𝑥
𝜕 𝜕𝑇
(𝐾𝑚𝑖𝑥 𝜕𝑥 ) + 𝑞̇ 𝑤𝑎𝑙𝑙 + 𝐻̇𝑠𝑜𝑢𝑟𝑐𝑒 + 𝑞̇ 𝑔𝑒𝑛 ·············· (2-3)
𝜕𝑥
20
In these equations, 𝑡 is time, 𝑥 is length in the direction of the well (measured depth
(MD)), 𝛼 is volume fraction, 𝜌 is density, 𝑣 is velocity, 𝛤̇ is gas dissolution rate per unit
volume, 𝑚̇𝑠𝑜𝑢𝑟𝑐𝑒 is the mass influx from sources (per unit volume), p is pressure, 𝑓𝑔 is
gravitational force, 𝑓𝑤 is wall friction, 𝑀̇𝑠𝑜𝑢𝑟𝑐𝑒 is the momentum influx from sources (per
unit volume), 𝑒 is internal energy, 𝑔 is the gravitational acceleration, 𝑧 is the true vertical
rate of external heat transfer through the walls per unit volume, 𝐻̇𝑠𝑜𝑢𝑟𝑐𝑒 is the energy influx
from sources (per unit volume), 𝑞̇ 𝑔𝑒𝑛 is the heat generation rate at the bit per unit volume,
Equations (2-1) and (2-2) are discussed in detail by Ma et al. [30]. In the energy
equation (Equation (2-3)), the two terms on the left-hand side refer to the conservation and
convective terms, respectively. On the right-hand side, the first term represents the axial
conduction in the mud; the second term is the heat transfer to the outer and inner (if any)
walls; the third term refers to the inlet energy source (in the form of enthalpy) from kicks,
pumps, injections, or the bit; and the last term is the rate of heat generation at the bit.
The DFM assumption enables this model to include the gas rising velocity and
differences in gas and liquid flow (which is necessary for simulating multi-phase flow),
while avoiding the complexities of the two-fluid models and their instabilities associated
with flow regime changes. Various sub-models are used to calculate the temperature- and
21
2.2.2 Closure Relations.
In order to be able to solve the equations, relevant closure equations are used for
the unknown terms in the energy equation. Internal energy and enthalpy of each element
𝑒 = 𝑐𝑣 𝑇 ························································································ (2-4)
where 𝑐𝑣 is the specific heat at constant volume, 𝑔 is the gravitational acceleration, 𝑧 is the
where mass flux from the source (𝑚̇𝑠𝑜𝑢𝑟𝑐𝑒 ) is calculated from the models of various sources
The external heat transfer for the drillstring and annulus fluids is modeled using a
radial heat resistance network, where heat is transferred between the drillstring and annulus
fluids, between the annulus fluid and the formation, and between the fluid in the riser or
choke line and the seawater. Figure 2-1 shows a schematic of an offshore well with 5
general cases of external heat transfer depending on the cell location. The wall heat transfer
rate for the pipe and annulus flows in Equation (2-3) is calculated according to the
• Heat transfer between the drillstring and annulus fluids through convection
in the drillstring fluid (on the inner side of the drillpipe), conduction through
22
the drillpipe, and convection in the annulus fluid (on the outer side of the
drillpipe).
• Heat transfer between the annulus fluid and the rock formation in openhole
• Heat transfer between the annulus fluid and the rock formation above the
casing shoe through convection in the annulus fluid, conduction through the
• Heat transfer between the fluid in the choke line and the seawater through
convection in the drilling fluid (on the inner side of the choke line),
conduction through the choke pipe, and convection in the seawater (on the
• Heat transfer between the fluid in the riser and the seawater through
convection in the drilling fluid (on the inner side of the riser), conduction
through the riser, and convection in the seawater (on the outer side of the
riser).
23
Figure 2-1. 5 Different cases of the heat transfer network for an offshore well. For
onshore wells, the network reduces to the first 3 cases.
Based on the modeled resistance network shown in Figure 2-, wall heat transfer on
24
where 𝑅𝑜𝑢𝑡𝑒𝑟 and 𝑅𝑖𝑛𝑛𝑒𝑟 resistances are the sum of individual conduction and convection
resistances on the outer and inner sides. These individual resistances are given by (note
that resistances are multiplied by the cell volume in order to get the final heat transfer per
unit volume):
𝑟
𝐴𝑙𝑛( 𝑜 )
𝑟𝑖
𝑅𝑐𝑜𝑛𝑑𝑢𝑐𝑡𝑖𝑜𝑛 = ········································································· (2-9)
2𝜋𝐾
𝐴
𝑅𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑜𝑛 = ℎ𝜋𝐷 ········································································· (2-10)
𝐻𝑇
where 𝐴 is the cross-sectional area (pipe or annuli), 𝑟𝑜 and 𝑟𝑖 are the outer and inner radius
of the solid across which heat is being conducted, 𝐾 is the conductivity of the solid, 𝐷𝐻𝑇 is
the diameter of the wall at which heat is being convected, and ℎ is the convection
where 𝐾𝑚𝑖𝑥 is the volume-averaged mixture conductivity, 𝐷ℎ is the hydraulic diameter and
𝑁𝑢𝑛𝑜𝑛−𝑁𝑒𝑤𝑡𝑜𝑛𝑖𝑎𝑛 is the Nusselt number for the non-Newtonian fluid, corrected by the fluid
behavior index, 𝑛 [73]:
where Δ=(3n+1)/4n.
Newtonian Nusselt number for laminar and turbulent flow is calculated from
Equations (2-13) and (2-14) and Figure 2 [74]. For laminar flow in pipes:
25
For laminar flow in the annulus, the Nusselt number is as given in Figure 22, where
the radius ratio is defined as the ratio between the inner radius and the outer radius of the
where 𝑅𝑒 and 𝑃𝑟 are the Reynolds number and Prandtl number respectively, and 𝑓 is the
friction factor, which can be obtained in real-time [75] or estimated from the fluid
rheological properties [76]. Other parameters in the energy equation or closure relations
are either given by the user or calculated from the mass and momentum conservation
equations.
Figure 2-2. Nusselt number for laminar annulus flow (from Nellis and Klein [77]).
26
• For solids: The specific heat capacity is assumed to be constant.
• For liquids: A polynomial model is used for the specific heat capacity at
𝑐𝑝 = 𝑎 + 𝑏𝑇 + 𝑐𝑇 2 + 𝑑𝑇 3 ································································ (2-15)
𝑐𝑣 = 𝑐𝑝 ························································································ (2-16)
• For gases: For monatomic gases, the specific heat capacities are constant.
The specific heat ratio (γ) and molar mass of the gas (M) are inputs to the
𝑅𝑢⁄
𝑐𝑝 = 𝑐𝑣 + 𝑀 ············································································· (2-18)
For other ideal gases, Equation (2-18) is used with the following polynomial
relation [79]:
𝑐𝑝
= 𝛼 + 𝛽𝑇 + 𝛾𝑇 2 + 𝛿𝑇 3 + 𝜀𝑇 4 ··················································· (2-19)
(𝑅𝑢 /𝑀)
27
2.2.4 Numerical Scheme.
In order to solve the system of conservation equations, the drillstring and annulus
are discretized along the well direction, and a second-order semi-implicit numerical
scheme is used to obtain a high level of accuracy. Details of the numerical scheme and the
second-order discretization are presented in the work of Evje and Fjelde [71]. The
numerical validation of the deployed algorithm is provided by work of Ma et al. [31]. The
solution procedure of the integrated hydraulic and thermal model is presented in the
flowchart shown in Figure 2-3. The subscripts and superscripts in the figure refer to the
cell number and time-step, respectively. The pressure, densities and volume fractions are
calculated from the mass conservation equations and the density models (which relate
densities to pressure). Subsequently, velocities are updated at the new time-step using the
mixture momentum equation and the slip law. Finally, the decoupled energy equation is
solved to find the temperature at the new time-step. This is done for all the cells while
appropriate boundary conditions are used. This process is repeated for each new time-step,
28
Figure 2-3. Solution procedure of the semi-implicit numerical scheme at each time-step.
stability, the pressure is updated using the mass conservation equations first. The updated
pressure is then used to calculate the pressure gradient term in the momentum equation
(Equation (2-2)) implicitly. All other terms in the conservation equations (i.e. the
convective terms on the left-hand side and the source terms on the right-hand side) are
while maintaining numerical stability through the implicit calculation of the pressure
gradient term [71]. Using this technique, fast, stable, and accurate simulations are achieved
29
that are on average 3 times faster than the real kick control operation, which is necessary
for real-time simulations and operating automatic controllers. Moreover, the robust
numerical scheme allows for simulations with small time-steps on the order of 10-2 s,
thereby enabling the simulator to capture the pressure waves during dynamic choke control.
The formation temperature in the well’s direct vicinity changes due to heat
exchange between the formation and wellbore fluid. Assuming a constant formation
temperature in external heat transfer calculations (Equation (2-8)) will lead to large errors
at high pump rates in HPHT wells or when the circulation is continued for prolonged
periods of time, when the formation temperature can become significantly different from
its initial value (i.e. the far-field temperature). To address this issue, the formation is
discretized radially at each cell to get the radial distribution of temperature and heat
conduction in the formation adjacent to the well (Figure 2-4). For simplicity, only
conduction in the radial direction is considered, and the axial heat conduction in the
formation is neglected. This can be justified due to the relatively small temperature
gradients and 2nd-order derivatives in the axial direction, resulting in small heat conduction
in this direction, which is negligible compared to the temperature gradient and heat
conduction in the radial direction. The number of discretized cells in the formation is
changed with time, such that the last cell (cell furthest from the well on the leftmost side
of Figure 2-4) has a near-zero temperature change from the initial in-situ geothermal
temperature.
30
Figure 2-4. Discretization of rock formation for heat transfer calculations (with bi-
lateral symmetry).
heat transfer from the adjacent cells. For each cell within the rock formation, the
temperature change is calculated based on the conduction heat transfer in radial direction:
𝜕𝑇𝑖,𝑗
𝜌𝑓𝑜𝑟𝑚 𝑐𝑣,𝑓𝑜𝑟𝑚 𝑉𝑖,𝑗 = 𝑄̇(𝑖+1,𝑗) 𝑡𝑜 (𝑖,𝑗) − 𝑄̇(𝑖,𝑗) 𝑡𝑜 (𝑖−1,𝑗) ······························· (2-20)
𝜕𝑡
where 𝜌𝑓𝑜𝑟𝑚 and 𝑐𝑣,𝑓𝑜𝑟𝑚 are the density and specific heat capacity of the rock formation
respectively, 𝑉 is the cell volume, and 𝑄̇ is the conduction heat transfer from the adjacent
cells, which is calculated as:
𝑇 −𝑇𝑖,𝑗
𝑄̇(𝑖+1,𝑗) 𝑡𝑜 (𝑖,𝑗) = 𝐾𝑓𝑜𝑟𝑚 𝐴𝐻𝑇 𝑖+1,𝑗 ····················································· (2-21)
𝑟 −𝑟 𝑖+1 𝑖
where 𝐾𝑓𝑜𝑟𝑚 is the conductivity of the rock formation, 𝐴𝐻𝑇 is the heat transfer area between
the adjacent cells, and 𝑟 is the radial location of the cell center.
31
2.3 MODEL VALIDATION
The multi-phase flow model has been previously validated [30] against
experimental data from a test well [80]. For validating the thermal model, there is a distinct
lack of reliable temperature data from test wells and actual wells out in the field. Until the
time where such data is actively gathered and more openly shared, we have to take a
pragmatic approach by comparing our model, defined as DFM with energy equation
conducted to verify the proposed thermal model against the single-phase steady-state
Hasan and Kabir model [20] as well as commercial software (Drillbench, which is
developed based on the hydraulic model of Petersen et al. [81]). In the example shown
here, the mud is circulated at 26.12 m3/h and 26.7 ℃ in a 4200 m deep well through a
drillstring with the inner and outer diameters (ID and OD) of 140 mm and 165 mm
respectively [18]. Annulus clearance is set to be 25.4 mm and circulation continues until
the mud temperature in the drillstring and annulus do not change any further (i.e. the system
reaches a steady-state). The mud initial temperature profile is assumed to follow a linear
formation gradient of 0.018 ℃/m and the surface temperature is assumed to be 26.7 ℃.
Table 2- shows the density, specific heat capacity and thermal conductivity of the mud,
32
As shown in Figure 2-5, the steady-state results from the proposed model are
compared with the results from both the Hasan and Kabir model and commercial software.
The maximum discrepancy between our simulation results and these existing model
predictions occurs at the bottom of the wellbore, where the temperature differences of 1 ℃
and 2 ℃ are observed in comparison with the Hasan and Kabir model and commercial
software, respectively. This difference can be attributed to the fact that the Hasan and Kabir
model uses a mud of constant density, whereas the DFM-WE uses a density model that is
dependent on the pressure and temperature dynamics. Moreover, in the current study, the
heat transfer within the rock formation is modeled using a conduction heat transfer
network, whereas the Hasan and Kabir model uses a dimensionless temperature coefficient
to include the effect of formation temperature variations [20]. However, aside from the
small deviation on bottom, the comparisons show a good match and the temperature
profiles in both the drillstring and the annulus have very similar trends.
Figure 2-5. Steady-state temperature profile compared to the results of Hasan and Kabir
model and commercial software.
33
Figure 2-6 shows the results of the temperature profile under steady-state condition
for the drillstring, annulus and radial formation cells. In this simulation, five radial nodes
(20 mm, 163 mm, 269 mm, 378 mm, and 485 mm into the rock formation respectively)
were used for the formation discretization. The formation node closest to the well (node 1)
shows the largest temperature change from the initial temperature. As the distance from
the well increases, temperature values get closer to the initial far-field value. It is observed
that at 1720 m TVD, the annulus mud temperature is equal to the initial formation
temperature and the formation temperature remains constant at this depth in this particular
case.
Figure 2-6. Drillstring, wellbore and formation nodes' temperature profile at steady-
state.
Results were also obtained for the transient temperature profiles at 1, 2 and 10 hours
after the mud circulation was initiated. Figure 2-7 shows the comparison against the
commercial software results (note that the Hasan and Kabir approach is a steady-state
solution that cannot capture the time-dependent behavior). Our simulation results are
34
generally in a good agreement with that from the commercial software except a temperature
difference of 3.5 ℃ at around 3750 m TVD (where peak temperature is reached) at 1 hour.
Figure 2-8 shows the bottomhole temperature of the DFM-WE versus the
commercial software result as a function of time for the first 1000 minutes of the
simulation. Our simulation results agree with those from the commercial software with a
35
Figure 2-8. Bottomhole temperature versus time compared against the results of
commercial software.
The developed hydraulic model provides an enabling tool for controller design that
previously reported models lack [82]. To demonstrate this capability of the model, an
example of controller tuning is presented in this section.
bottomhole pressure (BHP) constant during a dynamic kick control scenario in a 4267 m
vertical well. The controller is then tuned using the hydraulic model. The results are
compared against those of the untuned controller and manual choke control for a 0.8 m3
gas kick in water-based mud (WBM). In all cases, the BHP is set to 0.69 MPa above the
reservoir pressure while circulating out the kick. The WBM has a density of 1414 Kg/m3,
36
fluid behavior index of 0.5383 and consistency index of 0.4079 Pa. s0.5383 , and negligible
yield stress [83]. Wellbore schematic and configuration are presented in Figure 2-9. In all
cases, the pump rate is set to 159 m3/h at 1 minute into the simulation. At 3 minutes, a kick
is introduced by increasing the BHP to 64.1 MPa. The kick is detected at 8 minutes into
the simulation (i.e. when the pit level is increased by 0.8 m3), and the choke is adjusted
(automatically or manually) to stop the kick and maintain the BHP at 64.8 MPa.
Figure 2-10 - 2-14 show the choke opening, BHP, liquid flow out, surface
backpressure, and pit gain results during the kick control process, respectively. It can be
seen that once the kick is detected at 8 minutes, the tuned controller rapidly adjusts the
choke opening and applies backpressure to keep the BHP at 64.8 MPa, with some minor
fluctuations when the gas is rapidly expanding near surface and at the choke. The untuned
controller lags and is not able to apply the required backpressure quickly enough, resulting
in fluctuations in BHP. These fluctuations lead to high pressures at around 35 minutes that
could fracture the formation, as well as a secondary kick at around 45 minutes. The manual
choke control method cannot keep the BHP constant, with sudden jumps in pressures
during the kick control process. Moreover, the slow reaction time after the kick is detected
37
results in a larger kick size as the kick is stopped later than the other cases. A secondary
38
Figure 2-13. Surface backpressure versus time.
The pit gain plot (Figure 2-14) also shows how the tuned controller outperforms
the untuned controller and manual control. Taking the gas expansion into account, the
tuned controller increases the choke opening to keep the BHP constant. The untuned
controller fails to open the choke efficiently and results in applying excessive
backpressures lower observed pit gain values. The manual control has the worst
performance. It not only fails to open the chock efficiently to prevent excessive
backpressures, but also introduces a much larger kick size, resulting in higher pit gain
values throughout the kick circulation. Through this example, one can see that the
design.
39
2.5 SIMULATION SCENARIOS
temperature profile during conventional kick control operations were studied earlier
[72,84] where a Weight & Wait (W&W) kick control situation in an onshore well with
aqueous drilling fluids was simulated. The results show that neglecting the temperature
dynamics underestimates the kick size, which can be hazardous and lead to additional kicks
while the gas kick is being circulated out of the well. Here, a dynamic kick control scenario
is simulated to show the robustness of the model and the importance of considering
thermodynamics in the multi-phase flow simulations of dynamic kick control scenarios and
MPD operations. The heat transfer network in the riser, choke line and the wellbore, gas
solubility in the non-aqueous drilling fluid and the fast transients associated with automated
choke control are considered when comparing the results of the DFM-WE and the DFM
Without Energy Equation (DFM-WOE). The dynamic kick control scenario is modeled for
two cases: a high-pressure riser case, where the gas is allowed to enter the riser, and a
conventional case, where the gas is circulated out through the choke line.
A plot of the well geometry is given in Figure 2-15 and the wellbore configuration
is presented in Table . The initial seawater and formation temperature profiles are also
given in Figure 2-16. In the simulations, the seawater temperature is assumed to remain
constant as the heat will dissipate quickly into the water. Since the deepwater currents are
less than 1 m/s [85], the free convection on the outer wall of the riser and choke line is
modeled with a constant Nusselt number of 100 [74]. The thermal conductivity of the
seawater is set to be 0.63 W/m.K for heat transfer calculations. The water depth is 610 m
40
Figure 2-15. Deviated well used in the dynamic kick control scenario.
Table 2-2. Pipe, casing, and cement sizes used in the dynamic kick control scenario.
Drillstring Drillstring Wellbore/Casing
MD
ID OD ID
(m)
(mm) (mm) (mm)
0 131 149 226
610 131 149 226
1890 131 149 226
4439 131 149 216
4629 101 149 216
4803 78 167 216
41
Figure 2-16. Initial temperature profile for the dynamic kick control scenario.
• 1-4 mins: SBM with a density of 1510 Kg/m3 is pumped at 136 m3/h and
25 ℃. The mud has an oil fraction of 0.7, a yield stress of 10.06 Pa, fluid
behavior index of 0.7323 and consistency index of 0.3285 Pa. s0.7323 [86].
Specific heat capacity and thermal conductivity of the mud are assumed to
be 1298 J/Kg.K and 0.8 W/m.K respectively (1438 Kg/m3 (12 ppg) SBM
• 4-15 mins (4-18 mins for the DFM-WOE): Reservoir pressure is set to be
73.1 MPa to introduce a kick. The kick is detected when the pit gain
increases by 0.8 m3 from the value after the start of circulation (which is
42
• 15-80 mins (18-80 mins for the DFM-WOE): After the kick is detected,
reservoir pressure is set to be 74.8 MPa using a PI controller and the same
pump rate is kept until the kick is circulated out of the well. Circulation is
continued until the gas has entirely circulated out of the well.
Figure 2-17 shows the temperature profiles in the drillstring, annulus and the first
formation node at different times in the simulation for the DFM-WE. The seawater
temperature stays constant as any water heated by the riser will not remain close to it,
whereas the formation temperature below the seafloor can change due to the heat transfer
between the formation and the annulus. The discontinuities in formation temperature result
from the different thermal resistances due to the difference in casing and cement
thicknesses and annulus area around the drill collars, which in turn result in different
conduction and convection resistances respectively, as well as the change from seawater to
43
The density of the dissolved gas profile and the solubility threshold for the two
models at different times are shown in Figure 2-18. The solubility threshold in SBM is a
function of pressure and temperature, and becomes infinity at high pressures [88]. As a
result, when the kick starts at the bottom of the hole at high pressure, it goes into solution
in its entirety. The mixing of the mud and the gas does not follow ideal mixing rules,
resulting in a much smaller surface pit gain than that of a kick in a WBM. Hence, the kick
is detected later (after the pit gain reaches a total of 0.8 m3, the detection threshold) and
kick control action, i.e. setting the BHP to a value higher than the reservoir pressure, is
delayed.
Figure 2-18. Comparison of the dissolved gas density (DG) and the solubility threshold
(ST) at different times.
Figure 2-18 shows that the kick influx is larger in the DFM-WE, resulting in an
earlier detection of the kick (by 3 minutes). Although both cases have similar kick size
(~0.8 m3) when detected, it takes longer for the kick to break out of the solution using the
DFM-WOE. The solubility threshold values of the two models are also slightly different
44
due to the difference in temperatures, which is not significant in this case, but can be of
more importance if the temperature difference near the surface is significant. As the gas
solubility decreases with pressure decrease, the dissolved gas eventually breaks out of the
solution while being circulated out of the wellbore. In some cases, the break-out depth
could be very close to the surface. This gives the drilling crew a short time to react, which
is a significant safety concern. The break-out occurs at 46 minutes in the DFM-WE and at
49 minutes in the DFM-WOE. The difference is due to the higher gas concentration of the
kick in the DFM-WE, resulting in a deeper break-out point as the dissolved gas is being
circulated out.
The volume fraction profile of the kick is plotted at different times in Figure 2-19
for both models. After the break-out point, the gas expands quickly as the pressure is
decreasing at shallower depth in the high-pressure riser. Although both models have similar
initial kick sizes, the higher gas concentration in the DFM-WE causes an earlier break-out
point and higher free gas volume fractions. Circulation is continued until the gas is entirely
circulated out of the well, at 69 minutes and 73 minutes into the simulation, for the DFM-
45
Figure 2-19. Comparison of the gas volume fractions at different times.
Figure 2-20 shows the BHP, surface backpressure and SPP as a function of time for
both models. BHP values are very close for both cases, except for the aforementioned
three-minute time shift. The reason for this can be seen in the BHP plot itself. When the
kick occurs, the BHP predicted by the DFM-WE is slightly lower than that predicted by
the DFM-WOE due to the temperature effects. The temperature change can be observed in
the temperature plot at 4 minutes in Figure 2-17. Lower BHP causes a more negative
pressure difference between the mud and the reservoir, leading to a higher gas influx. This
is the reason behind the higher gas concentration and the earlier detection of the kick
mentioned earlier. After the kick is detected, the PI controller keeps the BHP constant at
46
Figure 2-20. (a) BHP, (b) surface backpressure and (c) SPP versus time.
The surface backpressure behavior is also affected by the kick detection time shift.
The backpressure peak happens at around 53 minutes when the gas reaches the surface and
is at its highest volume. As the gas concentration is higher in the DFM-WE while the
volume fraction of the gas at surface is also higher, a higher surface backpressure is
required to keep the BHP above the reservoir pressure. SPP results are similar with small
Figure 2-21 shows the flow out, pit gain, and choke opening behavior versus time.
As indicated, the pit volume change is reduced to -0.64 m3 when the pumping starts (4
minutes) due to the mud compressibility for both cases. The kick is detected and stopped
after a 0.8 m3 increase in pit gain. Subsequently, the pit gain in the DFM-WOE remains
constant as long as the gas remains entirely dissolved in solution at high pressures and
temperatures; the pit gain for the DFM-WE, however, slightly decreases with time due to
the reduction of the average mud temperature. Once the gas breaks out of the solution, the
pit gain suddenly increases and reaches its peak of 2.0 m3 and 1.4 m3 in the DFM-WE and
47
DFM-WOE respectively when the gas is at surface. This jump in the pit gain is higher in
the DFM-WE as the gas has a higher density and reaches a larger volume fraction. Once
the gas is circulated out of the well, the pit volume change in the DFM-WOE goes back to
-0.94 m3, resulting from the compression of the mud from the circulation and the added
surface backpressure; the pit volume change of the DFM-WE after the gas is circulated out
is -1.3 m3 due to the compression of the mud by circulation, surface backpressure and
Figure 2-21. (a) flow out, (b) pit gain and (c) choke opening plots.
In this case, a similar kick is controlled by closing the blowout preventer (BOP)
and circulating the kick through the choke line. Since the model simulates the flow in a
fixed preset geometry, simulation is started while pumping through the choke line and
reservoir pressure is selected such that the resulting influx is similar to that of the previous
case. The results are studied after the kick is detected and circulating through the choke
48
line is started. For this case, the kick is detected at 15 minutes for the DFM-WE and 17
minutes for the DFM-WOE. Once the kick is detected, the subsea BOP is closed and the
kick is circulated at 22.7 m3/h through the 114 mm choke line. Simulation is continued
until the kick is circulated out of the choke line, which takes more time due to the lower
pump rate compared to 136 m3/h used in the high-pressure riser case (300 minutes into the
simulation).
Similar to the previous case, plots of relevant parameters are shown in Figure 2-22
- 2-26, including temperature profiles, dissolved gas density, gas volume fraction, BHP,
surface backpressure and SPP versus time, and flow out, pit gain and choke opening. In
this case, a higher backpressure is needed to keep the BHP constant due to the lower pump
rate. This causes higher pressures in the choke line compared to the high-pressure riser
used for gas circulation in the previous case. These higher pressures will increase the
solubility threshold, and hence a lower amount of the gas will break out. This results in
lower pit gains when circulating out the kick at a lower pump rate through the choke line.
Temperature profiles in this case differ from those of the previous case due to the smaller
area and heat transfer network change for the choke line. Comparison between the DFM-
WE and DFM-WOE shows results similar to that seen in the high-pressure riser case.
49
Figure 2-22. Drillstring, annulus and near-wellbore formation temperature profiles at
different times.
Figure 2-23. Comparison of the dissolved gas density (DG) and the solubility threshold
(ST) at different times.
50
Figure 2-24. Comparison of the gas volume fractions at different times.
Figure 2-25. (a) BHP, (b) surface backpressure and (c) SPP versus time.
51
Figure 2-26. (a) flow out, (b) pit gain and (c) choke opening plots.
simulator. This new methodology better explains wellbore thermodynamics and enhances
the modeling accuracy and capabilities of the DFM when simulating well control events
and MPD operations, especially under HPHT conditions. Previous DFM-WE model
models, neglect pressure waves, and/or make simplifying assumptions in the energy
equation that lead to a loss of modeling accuracy. Moreover, temperature effects during
MPD operations and automated choke control scenarios are not considered in the published
literature. The work presented here is a modeling approach that integrates a DFM, an
extensive thermal model and suitable closure relations to simulate the multi-phase flow in
52
To model dynamic well control scenarios, a comprehensive second-order semi-
accurate simulations in real-time that are numerically less expensive than similar implicit
models. As a result, small time-steps can be used to capture the pressure waves and
simulate the fast-transient events that are associated with the use of MPD systems. Using
the robust numerical schemes, a PI controller is designed and tuned to keep the BHP
constant during MPD kick control simulations. Results show that rapid adjustments of the
choke opening via the automatic controller improves the BHP control greatly compared to
Using the developed thermal model, temperature can be estimated in the drillstring,
annulus, and formation adjacent to the well during drilling and well control scenarios. The
model considers axial heat conduction in the mud, heat generation at the bit, non-
Newtonian convection, and heat storage in the formation to precisely estimate dynamic
temperatures during complex well control situations. Comparisons with existing thermal
models show a very good match for both steady-state and transient temperature behavior,
but verification is unfortunately limited by the lack of experimental data. The integrated
thermal and hydraulic model can simulate HPHT situations, non-Newtonian liquids,
comprehensive density and viscosity effects, arbitrary 3D well path and geometry, multiple
gases and liquids in the well, MPD operations and gas solubility in non-aqueous drilling
fluids.
an offshore dynamic kick control scenario with SBM was simulated to study the effect of
the thermal model on the gas solubility and break-out depth. Although the effect of
temperature on the gas solubility was not significant for the presented case, the DFM-WE
predicts a higher dissolved gas density for the same influx size, and hence leads to a break-
53
out point that occurs earlier in time and deeper in the well. The amount of the gas that
breaks out of solution is also higher for the DFM-WE. The same simulation scenario was
modeled for a case where the kick is allowed to enter the high-pressure riser and controlled
by back-pressure, and for a case where the kick is circulated out conventionally through
the choke line. Due to the slower pump rate through the choke line and higher solubility
thresholds as a result of increased backpressure, the amount of gas that breaks out of
solution is lower in the latter case, while the time it takes to circulate out the kick is longer.
In both cases, however, omitting the temperature dynamics in the DFM-WOE leads to an
the kick. By considering the temperature effects on well control, the developed DFM-WE
can provide accurate estimations of the gas kick behavior under HPHT conditions and
assist the drilling crew to safely stop and circulate out the kick.
54
Chapter 3: Cuttings Transport Modeling2
3.1 INTRODUCTION
Current hole cleaning practices are mostly based on experience, rules of thumb and
simplistic calculations. Hence, they are not reliable and do not work as expected in all
scenarios. There is the need for a robust, fast, and accurate approach to simulate cuttings
transport, and provide reliable and useful estimations of the hole conditions in real-time.
processes and estimate the cuttings concentration and the dynamic bed height during
drilling and hole cleaning operations. The model uses a DFM approach and a robust semi-
implicit numerical scheme for fast real-time simulations (developed based on the model
used in 0). The semi-implicit algorithm enables using small time-steps (on the order of 10-
2
s) that capture the fast transients and the pressure waves in the well, which is missing
from the other hole cleaning models in the published literature. This is necessary for
simulating MPD operations, where the automatic choke controller adjusts the BHP rapidly
to maintain it within the drilling margin. The cuttings concentration and blockage effects
are considered in the model, allowing the automated controllers to maintain the BHP during
dynamic hole cleaning operations. The developed model considers the pressure-dependent
cuttings slip velocity, pipe rotation and eccentricity effects, complex 3D well path and
geometry, etc.
2The work in this chapter is previously published [114]. Q. Gu, G. Saini, A., D. Chen, P. Ashok, E. van
Oort, and A. Karimi Vajargah contributed to the model development and manuscript revision of the
published article.
55
Case studies are performed based on field experiments to analyze the effectiveness
of the developed model on avoiding operational problems such as pack-off and stuck pipe.
It is shown that by monitoring the real-time cuttings concentration and bed height along
the wellbore, the developed model can detect improper hole cleaning conditions and
provide optimum drilling parameters to resolve problems, thereby minimizing NPT. The
robust numerical scheme allows for simulations that are several times faster than the real-
time operation on a standard desktop PC, providing the crew with enough time to take
preventive actions. Clean-up cycles can also be simulated by the model to calculate and
optimize the required parameters for optimum hole cleaning results. Required clean-up
times are calculated for the field cases to ensure that cuttings are effectively removed from
the wellbore before pulling out of hole and running casing. Moreover, MPD operations can
be simulated using the model that consider the effects of cuttings concentration and bed
conditions to the drilling crew during the drilling process and give adequate time to the
crew to take timely corrective action if necessary. Moreover, it is shown how the model
can simulate the planned drilling process and calculate optimum drilling parameters to
In this section, the mathematical model which consists of the general conservation
equations, closure sub-models, and the numerical scheme are discussed in detail. The
model at its core is based on the gas-liquid multi-phase model introduced in 0. Necessary
modifications are made to optimize the model for solid-liquid flow and cuttings bed
56
generation. To model the multi-phase flow of solid particles and liquid in the wellbore, the
as velocity are averaged over the cross-sectional area. All parameters are
calculated at cell centers and are assumed to be unchanged within the cell.
• In near-vertical sections (inclination < 35º), the cuttings bed will not form,
and all the cuttings particles remain in suspension. In the deviated and
cuttings bed depending on the maximum cuttings bed area. This maximum
bed area depends on the drilling conditions (e.g. flow rate, pipe rotation,
is lower due to the random settling and various cuttings shapes. Here, we
cuttings transport models [45]. The cuttings bed is also assumed to be stable
temperature profile.
thermal models (e.g., Fallah et al. [89]). In this way, the complexity of
low-temperature wells.
The model consists of mass conservation equations for the liquid and solid phases
in suspension, a mass conservation equation for the cuttings bed (for sections of the well
where cuttings bed can form), and a mixture momentum conservation equation for the
𝜕𝛼𝑠𝑢𝑠𝑝 𝑐𝑠 𝜕𝛼𝑠𝑢𝑠𝑝 𝑐𝑠 𝑣𝑠
𝜌𝑠 + 𝜌𝑠 = −𝑆𝑠̇ + 𝑚̇𝑠𝑜𝑢𝑟𝑐𝑒,𝑠 ············································ (3-2)
𝜕𝑡 𝜕𝑥
where 𝑡 is time, 𝑥 is the length along the wellbore (measured depth (MD)), 𝛼 is volume
fraction, 𝑐 is concentration, 𝜌 is density, 𝑣 is velocity, 𝑆̇ is the settling rate in bed per unit
volume, 𝑚̇𝑠𝑜𝑢𝑟𝑐𝑒 is the rate of mass generation from sources per unit volume (at the pump
or bit), 𝑝 is pressure, 𝑓𝑔 is gravitational force, 𝑓𝑤 is wall friction, 𝑀̇𝑠𝑜𝑢𝑟𝑐𝑒 is the rate of
momentum generation from sources per unit volume, and subscripts 𝑙 and 𝑠 represent liquid
58
The time-derivative terms in Equations (3-1 – 3-4) are the conservation terms, the
x-derivative terms are the convective terms (note that the stationary bed does not have a
convective term since the bed velocity is assumed to be zero), and the terms on the right-
hand side are the mass and momentum source terms. The source terms in the mass
equations are either external (i.e. the mud pump and cuttings generation at the bit), or
internal (i.e. the liquid and solid settling in the bed and vice versa). In the right-hand side
of the momentum equation, the first term is the momentum loss due to liquid and solid
settling in the bed; the second term is the gravitational force; the third term is the
momentum loss due to the wall friction; and the last term is the momentum source from
external sources (such as the mud pump and cuttings generation at the bit).
Note that the density of the cuttings particles is known since the solid is assumed
to be incompressible. The solid and liquid concentrations in the bed are also known from
the packing efficiency. System of Equations (3-1 – 3-4) consists of four conservation
equations and eight unknowns (undetermined system), namely, suspension and bed volume
fractions, liquid and solid concentration in suspension, liquid and solid velocity in
suspension, liquid density, and pressure. Therefore, four additional closure relations are
𝑐𝑠 + 𝑐𝑙 = 1 ···················································································· (3-6)
59
• Liquid density is a function of pressure and temperature. A linear density
model for the mud is used (it is assumed that the temperature profile is
known):
𝑝−𝑝𝑙0
𝜌𝑙 = 𝜌𝑙0 + − 𝐶(𝑇 − 𝑇𝑙0 ) ····························································· (3-7)
𝑎𝑙2
temperature and 𝜌𝑙0 , 𝑝𝑙0 and 𝑇𝑙0 are reference density, pressure and
temperature, respectively.
• Liquid and solid in the suspension do not travel at the same velocity. In
other words, slippage exists between the two phases, which is defined as
follows:
The slip velocity is calculated using force equilibrium for a single particle in
suspension where the effect of other particles is neglected [90]. From the force equilibrium:
𝜋
𝐷= 𝑔 (𝜌𝑠 − 𝜌𝑙 ) 𝑑𝑠3 cos (𝜃)······························································ (3-9)
6
where 𝐷 is the drag force, 𝑔 is gravitational acceleration, 𝑑𝑠 is the particle diameter, and 𝜃
𝜋 2
𝐷= 𝑓 𝜌𝑙 𝑑𝑠2 𝑣𝑠𝑙𝑖𝑝 for 𝑅𝑒𝑝 > 0.1 ····················································· (3-10b)
8
60
where 𝜇 is viscosity, 𝑣𝑠𝑙𝑖𝑝 is the slip velocity between the solid and liquid, 𝑅𝑒𝑝 is particle
Reynolds number, and 𝑓 is the friction factor calculated from Bourgoyne Jr et al. [90]. This
slip law includes the effect of the cutting size and shape, flow regime (laminar or turbulent),
The frictional pressure drop for the drilling fluid is obtained using the approach
proposed by Ahmed and Miska [91]. In friction factor calculations, the mixture viscosity
1/3
𝜇𝑚𝑖𝑥𝑡𝑢𝑟𝑒 = 𝜇𝑙 /(1 − 𝑐𝑠 ) for 𝑐𝑠 > 0.74 ············································· (3-11b)
where the non-Newtonian liquid viscosity is calculated from the yield power law (YPL)
properties. The closure relations and sub-models make a determined system of equations
which can be solved by using a numerical approach discussed further in this chapter.
In order to describe the hole cleaning dynamics, the height of the cuttings bed under
transient conditions needs to be calculated. The precise simulation of the cuttings bed
height depends on several parameters and is computationally expensive. Here, we use the
experimental model initially proposed by Larsen et al. [34], and then modified by Jalukar
[92] and Bassal [93] to calculate the terminal velocity at every cell at each time-step. The
steady-state bed area is derived from the terminal velocity obtained from the Larsen model
(along with Jalukar and Bassal modifications), which is then included in the transient
cuttings transport model as a threshold for cuttings settling. Additional cuttings can settle
in the bed in each cell if the bed area is lower than the threshold. By contrast, if the bed
61
area is larger than the threshold (e.g. when the pump rate or pipe rotation is increased),
cuttings settled in the bed will go back into suspension and will be carried upward by the
drilling fluid. Use of the Larsen model avoids the complexities and limitations associated
and still takes into account multiple parameters that affect the bed height, such as flow rate,
pipe rotation, inclination, pipe eccentricity, particle size, mud rheology, hole sizes, etc.
based on the discretization method proposed by Evje and Fjelde [71] and used in 0. Using
real-time. Figure 3-1 shows a flowchart of the solution procedure used in this study.
62
Figure 3-1. Solution procedure of the semi-implicit numerical scheme at each time-step.
The subscripts and superscripts in the figure refer to the cell number and time-step,
respectively. First, the mass conservation equations are solved to obtain the volume
fractions of the bed and suspension, as well as the cuttings and liquid concentration in
suspension (note that the cuttings density and the solid and liquid concentrations in the bed
are known). Using Equations (3-1) and (3-7), liquid density and pressure are updated. The
pressure gradient term in Equation (3-4) is then calculated using the updated pressures (i.e.,
numerical stability of the algorithm [71]. Subsequently, the momentum conservation term
is updated (Equation (3-4)), and the liquid and solid velocities are calculated using the slip
velocity (Equation (3-8)). Finally, the remaining parameters such as viscosity are updated,
and the terms on the right-hand side of Equations (3-1 – 3-4) are calculated for the next
time-step.
Simulations on a normal PC are around eight times faster than the real operation, providing
the drilling crew with enough time to take necessary actions and to plan ahead. Relying on
the fast algorithm, small time-steps are used in solving the equations (which are on the
order of 10-2 s). Using small time-steps captures the pressure waves, and considers fast
transients, which are necessary for automatic choke control. This allows the simulation of
MPD operations while considering the cuttings transport effects, which is currently missing
63
3.3 SIMULATION CAPABILITIES AND CASE STUDIES
The developed model is capable of simulating transient drilling and hole cleaning
operations to provide important information to the drilling crew in real-time. The model
• Drilling operation: The model can estimate the bed blockage, cuttings
results, ROP, pump rate, pipe RPM and fluid rheology can be tuned for
effective cuttings removal from the borehole. During the drilling process,
the model can monitor the cuttings flow in real-time and inform the drilling
crew of the results of changing the parameters (e.g. increasing the ROP for
clean-up cycles after drilling to the target MD and before pulling out of hole
(POOH). It is desirable to use the maximum allowable pump rate and pipe
RPM for best hole cleaning results (note that it is not always practically
viable to remove all the cuttings from the wellbore, especially in large
boreholes). The model can simulate the transient clean-up cycles and
provide the crew with useful information such as: maximum observed BHP,
wellbore, the expected volume of cuttings that are removed from the well,
and the bed blockage and remaining cuttings in the well after the clean-up
64
cycle. This information will help minimize the risk of “knock-on” problems
In case 1, two similar horizontal wells which experienced cementing problems are
analyzed. The well geometries are described in Figure 3-2. Well 1.1 was drilled to the target
MD of 2,500 ft. While POOH, increased torque was observed followed by erratic torque
readings during back reaming. Moreover, the well unloaded at approximately 950 ft. While
running casing after POOH, no issues were observed. However, during the cementing
stage, returns were lost, and the casing string became stuck after pumping the cement
slurry. Similarly, Well 1.2 was drilled to the target MD of 2,434 ft, increased torque
readings were observed during POOH and the well unloaded, followed by intermittent
losses of returns while cementing. It was suspected that pack-off behind the casing string
occurred in both cases, which led to the induced losses. The proposed model was used to
analyze the cuttings transport in these two wells. The modeling parameters are provided in
Figure 3-2.
65
Well Geometry Fluid Properties
MD 762 / Density
1090
(m) 742 (Kg/m3)
Openhole Yield stress 7.7 /
311
diameter (mm) (Pa) 8.7
Drillstring
127 Drilling Parameters
diameter (mm)
Average
6.4 Pipe RPM 60
diameter (mm)
Figure 3-2. Well inclination and simulation parameters used in case study 1.
Figure 3-3 shows the bed blockage, cuttings concentration in suspension, pressure,
and annular solid velocity profiles along the well at steady-state under the given drilling
conditions for both wells. Due to the large hole size, the annular velocity is not sufficient
to avoid bed formation, and about 35% and 28% of the annular area is blocked by the
cuttings bed for Wells 1.1 and 1.2, respectively. The cuttings concentration is roughly 5%
for both cases, with a slight increase in the vertical section. This increase is due to the
higher slip velocity in the vertical section, caused by the weight of the cuttings. The
pressure profile is affected by the cuttings concentration and the frictional pressure loss,
leading to a BHP of 4916 KPa and 5061 KPa for Wells 1.1 and 1.2, respectively. The mud
and cuttings velocity profiles are affected by the bed blockage, where the annular velocities
are almost 1.5 times higher as compared to the annular velocities through the unblocked
66
annulus (1.0 m/s and 0.8 m/s). This increased velocity keeps the solid particles in
suspension and prevents additional blockage. However, the frictional pressure drop
Figure 3-3. Steady-state bed blockage, cuttings concentration, annular pressure, and
velocity profiles while drilling (case 1).
Note that when drilling stops (at the steady-state condition of Figure 3-), cuttings
particles that were in suspension settle down in the bed, increasing the bed blockage by
5%. During the tripping out operations, pulling the bottomhole assembly (BHA) through
regions of high cuttings bed can cause excessive torque and drag. Moreover, the remaining
cuttings in the well could have been trapped outside of the casing string and led to lost
circulation while cementing. To avoid such problems, two hole cleaning case studies were
explored using the model. Recommendations based on the simulation results are reported
as below.
In this case, the hole is cleaning using higher pipe RPM values, assuming the pump
rate is limited by other factors, e.g. fracture pressure and/or mud pump discharge pressure
67
limitations. Simulations were conducted to determine the amount of cuttings that can be
cleaned out of the well and the required time for effective cuttings removal. It is known
that within the operation limits, increasing the pump rate during the clean-up cycle can
further improve the cuttings removal from the well. Figure 3-4 quantitatively shows the
effect of pipe rotation speed on the cuttings bed height after the clean-up cycle. Higher pipe
rotation speeds lead to smaller bed blockage and less severe hole cleaning problems. For
example, a complete clean-up cycle with a 120 RPM pipe rotation removes 4.45 m3 of
cuttings from the well, which is approximately 60% of the total volume of cuttings that
remained in the borehole during drilling. The model can help the drilling crew find the root
The model can also describe the transient behavior during the hole cleaning process,
which is important for real-time operation and monitoring of the cuttings mass coming out
68
of the well. Figure 3-5 shows the cuttings concentration profile at different times during a
clean-up cycle, where the pump rate is maintained at 209 m3/h and the pipe rotation is
increased from 60 to 120 RPM for improved hole cleaning. The clean-up cycle follows
after the steady-state drilling conditions of Figure 3-3. Due to the increased pipe rotation,
about 50% of the cuttings that had settled in the bed move back into suspension and are
suspension compared to that of the drilling conditions (at 0 mins). When these highly-
concentrated cuttings reach the vertical section, the higher density of the mud-and-cuttings
mixture increases the BHP, which could exceed the fracture pressure. Figure 3-6 shows the
cuttings mass out and BHP during the clean-up cycle. During the first 4 minutes of hole
cleaning, high-concentration cuttings are entering the vertical section, increasing the BHP
to a maximum of 5288 KPa. The elevated pressure profile needs to be monitored to avoid
exceeding the fracture pressure. At 4 minutes, these cuttings reach the surface and exit the
well at a much higher rate compared to that of the normal drilling operation due to the
increased concentration. The maximum cuttings mass out rate is around 1263 Kg/min,
which is about three times the mass exit rate during normal drilling. As the cuttings exit
the well, BHP and cuttings flow out rate decrease. At 24 minutes, the suspended cuttings
are removed from the well (with some cuttings remaining in the cuttings bed) and the BHP
reduces to 4447 KPa. The required number of bottoms-up cycles for the clean-up cycle is
1.7. Similar results are observed for both Well 1.1 and Well 1.2.
69
Figure 3-5. Cuttings concentration in suspension during the clean-up cycle (case 1A).
Figure 3-6. Cuttings mass out and BHP during the clean-up cycle (case 1A).
70
3.3.1.2 Case 1B: Changing the Fluid Rheological Properties
Increasing the flow rate improves hole cleaning, however, the fluid rheology and
frictional pressure drop limit the flow rate. In this case, the mud rheology is changed to
allow drilling at a higher pump rate, where a less-viscous drilling fluid is proposed to enable
higher pump rates without increasing the BHP. The proposed mud has a density of 1054
Kg/m3, a yield stress of 5.6 Pa, a consistency index of 1.0 Pa.s0.42, and a fluid behavior
index of 0.42. Limiting the standpipe pressure (SPP) to 27579 KPa, the pump rate is set to
be 363 m3/h with a pipe rotational speed of 120 RPM during both drilling and hole cleaning.
Figure 3-7 shows the bed blockage, cuttings concentration, pressure, and solid velocity
profiles before and after the clean-up cycle. The bed blockage while drilling is 7%, with a
solid concentration of 3%. Compared to the result showed in Figure 3-3, the bed blockage
is reduced by an additional 75% due to the increased flow rate and pipe rotation speed. The
decreased viscosity of the mud compensates for the increased flow rate, leading to a minor
reduction of the BHP from 4640 KPa before to 4357 KPa after hole cleaning.
Figure 3-7. Bed blockage, cuttings concentration, annular pressure, and velocity profiles
before and after hole cleaning (case 1B).
71
Figure 3-8 shows the cuttings mass out and the BHP during the clean-up cycle.
Similar to Figure 3-6, the BHP and cuttings mass out are increased when the initially settled
cuttings reach the vertical sections and exit the wellbore. However, due to the reduced bed
blockage, the cuttings mass out reaches a maximum of 899 Kg/min at 5 minutes. The
maximum observed BHP is 4757 KPa. The required clean-up time in this case is 14
minutes, which is 10 minutes less than case 1A due to the increased pump rate and the
reduced bed blockage. The steady-state BHP after hole cleaning is 4357 KPa, which is
close to that of case 1A where BHP was 4447 KPa. Due to the reduced cuttings loading in
the annulus, the BHP variations during the clean-up cycle are about half compared to those
of case 1A. Similar results are observed for both Well 1.1 and Well 1.2.
Figure 3-8. Cuttings mass out and BHP during the clean-up cycle (case 1B).
In this case, the model is used to determine the minimum required pump rate for
efficient cuttings transport. The well geometry and simulation parameters are provided in
Figure 3-9. The annular mud velocities are reduced near surface due to the casing design
72
and increased casing diameters. This velocity needs to be higher than the slip velocity
(Equation (3-8)) to ensure cuttings are transported upwards. If the annular velocity of the
liquid is lower than the slip velocity, the cuttings weight keeps them at the last casing shoe.
This increases the cuttings concentration and BHP and could lead to further hole cleaning
problems if the cuttings are not transported to surface. Since the slip velocity increases with
cuttings size, a relatively large cuttings size of 6.4 mm is used in the simulations to obtain
worst case estimates. Various flow rates are simulated to determine the minimum required
pump rate.
Well Geometry Fluid Properties
MD Density
4688 1102
(m) (Kg/m3)
Openhole Yield stress
216 5.3
diameter (mm) (Pa)
Average
6.4 Pipe RPM 80
diameter (mm)
Figure 3-9. Well inclination and simulation parameters used in case study 2.
Figure 3-10 shows the liquid and solid velocity profiles for four pump rates of 102
m3/h, 114 m3/h, 125 m3/h, and 126 m3/h. The slip velocity is similar for all cases because
73
of the similarity in cuttings and mud properties. The velocity profiles are mainly affected
by the annular area and the bed blockage, with drops observed at the casing shoe locations.
At the last casing shoe, the solid velocity is less than 0.3 m/s in all cases. For pump rates
below 91 m3/h, the solid velocity is negative, i.e. the lift force is less than the cuttings
Figure 3-10. Mud and cuttings velocity profiles for different pump rates (case 2).
Figure 3-11 shows the steady-state bed blockage, cuttings concentration, and
pressure profiles for different flow rates. The observed bed blockage in the horizontal
section is 21%, 17%, 13%, and 8%, corresponding to the four pump rates. Clean-up cycles
can further reduce the bed blockage. The cuttings concentration profile is affected by the
solid and liquid velocity profiles, where the cuttings concentration is increased within the
larger casing sizes. The maximum solid concentration is observed on surface at 5.5%,
74
4.3%, 3.4%, and 2.9% for the 102, 114, 125, and 136 m3/h cases, respectively. While the
frictional pressure drop is increased at higher pump rates, the reduced cuttings
concentration lowers the hydrostatic pressure, resulting in very similar pressure profiles for
Figure 3-11. Steady-state bed blockage, cuttings concentration, and annular pressure
profiles while drilling (case 2).
In case 3, the model is used to calculate the required clean-up time and estimate the
bottoms-up cycles needed to remove the cuttings from the well. Here, a horizontal well
was drilled to the target MD of 6690 m, followed by a reaming process for 3 hours with a
pump rate of 65 m3/h at 120 RPM. Upon POOH, more reaming was done at 5730 m and
3764 m for 20 minutes and 45 minutes, respectively. The presented model is used to study
the effect of ROP on the solid concentration and bed blockage profiles during the drilling
operation. Required bottoms-up circulations for complete hole cleaning are also calculated
using the model to optimize the time spent on cleaning cycles. Three different cutting sizes
were used in the simulation to show the effect of cuttings size in wellbore cleaning. The
Drillstring
114 Drilling Parameters
diameter (mm)
Pump rate
Cuttings Properties 65
(m3/h)
1.3 /
Average
4.4 / Pipe RPM 110
diameter (mm)
6.3
Figure 3-12. Well inclination and simulation parameters used in case studies 3 and 4.
Figure 3-13 shows the bed blockage, cuttings concentration in suspension, pressure,
and solid velocity profiles for three different ROP values: 9 m/h, 30 m/h, and 46 m/h. In
this case, the flow rate was high enough to avoid the formation of a cuttings bed, except
for a small bed (<3% blockage) at the end of the build section for the 30 m/h and 46 m/h
cases, which can be removed during clean-up cycles. The cuttings concentration in
suspension is increased with ROP. In the horizontal section, cuttings concentrations are
0.3%, 1.1% and 1.7% for the 9 m/h, 30 m/h and 46 m/h cases, respectively. The total
volume of the cuttings in the wellbore for each case is 0.3 m3, 1.1 m3, and 1.6 m3,
respectively. Due to the small cutting concentrations in the annulus, the pressure profiles
are similar and are primarily governed by the hydrostatic fluid pressure and the frictional
76
pressure loss, with small differences due to the weight of the suspended cuttings. The
bottomhole pressures during drilling are 57109 KPa, 57668 KPa and 58061 KPa for the 9
m/h, 30 m/h and 46 m/h cases, respectively. The solid velocity profiles are similar for all
cases because the flow rate through the wellbore is the same and the slip velocity is
Figure 3-13. Steady-state bed blockage, cuttings concentration, annular pressure, and
velocity profiles while drilling (case 3).
Next, the impact of cuttings size on hole cleaning is investigated. While the average
cuttings size is assumed to be 4.4 mm, cuttings are not of equal size, and a distribution of
cuttings particle sizes exists. Larger cuttings particles have a higher slip velocity due to
higher weight, and move uphole slower than the smaller cuttings (Equations (3-8 – 3-10)).
Three different cutting sizes were used to estimate the time required for the clean-up cycle:
small cuttings with a diameter of 0.6 mm, original cuttings size of 4.4 mm, and large
cuttings with a diameter of 6.3 mm. Figure 3-14 shows the cuttings mass out and the BHP
during the clean-up cycle for the three cases, which is followed after drilling at 9 m/h. Due
to the slower velocity of the large cuttings, it takes longer to clean these cuttings from the
77
wellbore. The required time periods to clean the well are 122, 130 and 136 minutes for the
small, original, and large cuttings sizes, respectively. These clean-up times translate to
1.45, 1.55, and 1.62 bottoms-ups cycles required to clean the wellbore for the different
cuttings particle sizes. These cleaning times are about 50 minutes less than the amount of
time used for reaming on bottom, and 115 minutes less than the total reaming time during
POOH. Clearly, clean-up cycles are a more efficient use of time than back-reaming in this
case.
Figure 3-14. Cuttings mass out and BHP during the clean-up cycle (case 3).
To show the capability of the model in simulating MPD operations, a drilling and
hole cleaning situation is simulated where the BHP is set to 59295 KPa using an automated
proportional integral (PI) controller. In this case, the wellbore geometry of case 3 is used.
In this case, the well is drilled with a pump rate of 45 m3/h, followed by a 65 m3/h clean-
up cycle to remove the remaining cuttings from the well. The MPD controller is used in
both the drilling stage and the clean-up cycle to maintain the BHP. Figure 3-15 shows the
78
steady-state bed blockage, cuttings concentration, pressure, and velocity profiles during
drilling. The BHP is kept at 59295 KPa using the automated choke control. Due to the
reduced pump rate (compared to that of case 3), a small portion of the annulus is blocked
by the bed (approximately 8%) in this case, with a small increase at the end of build. The
Figure 3-15. Steady-state bed blockage, cuttings concentration, annular pressure, and
velocity profiles while drilling (case 4).
A clean-up cycle with a pump rate of 65 m3/h is followed after the drilling to remove
the cuttings and clear the bed blockage. Figure 3-16 shows the cuttings concentration
profile at different times during the clean-up cycle. Due to the re-entrainment of the
cuttings from the bed, a peak cuttings concentration of 5% is observed which flows uphole
and reaches the surface at approximately 45 minutes into the clean-up cycle. Figure 3-17
shows the cuttings mass out, BHP, and the choke opening during the clean-up cycle. Due
to the increased pump rate, the cuttings mass out and BHP are increased when cleaning
starts. The automated choke is adjusted accordingly to reduce the BHP and maintain the
setpoint of 59295 KPa. The choke opening increases until approximately 45 minutes into
the hole cleaning to account for the elevated hydrostatic pressure of the mud and cuttings
79
mixture in the vertical section. This elevated pressure is caused by high cuttings
concentrations. After 45 minutes, fluid flow with a high cuttings concentration reaches the
surface and the cuttings are removed from the well with a temporary increase in the flow
out rate. The controller responds by first increasing and later on reducing the choke opening
when the cuttings are out of the well. The maximum cuttings mass out is 112 Kg/min,
which is observed at 70 minutes. During the entire process, the BHP is kept at 59295 KPa
with minor oscillations. The cuttings are removed from the wellbore at 130 minutes, at
which time the choke opening reaches a steady-state, while keeping the BHP at 59295 KPa.
Figure 3-16. Cuttings concentration profile at different times during the MPD-controlled
clean-up cycle (case 4).
80
Figure 3-17. Cuttings concentration profile at different times during the MPD-controlled
clean-up cycle (case 4).
A summary of the analyzed cases is listed in Table . Note that these case studies
show the capabilities and applications of the model over a wide range of hole cleaning
situations.
81
MPD choke control during Using an MPD- BHP is controlled while
Case 4
drilling and clean-up cycle controlled choke drilling and cleaning the hole
3.4 CONCLUSIONS
There is a need for comprehensive real-time hole cleaning models in the drilling
industry. Current hole cleaning models either lack robustness and stability, are overly
simulations. The new hole cleaning model presented in this chapter is based on a drift-flux
scheme allows for stable and robust simulations using small time-steps to capture fast
transients during drilling operations. This allows for integrating the model with automated
choke controllers for simulating MPD operations. The effects of mud compressibility, area
discontinuities, and non-Newtonian fluid rheology are also considered. Providing real-time
estimates of the bed blockage and suspended cuttings concentration, the proposed cuttings
transport model provides informative and comprehensive hole cleaning guidance to the
drilling crew.
Applying the developed modeling techniques to field cases shows that the model
can identify improper or insufficient hole cleaning practices, provide optimum drilling and
hole cleaning parameters, and help minimize hole cleaning-related NPT. The model can
estimate the cuttings bed height and concentration, pressure, and velocity profiles along
the wellbore to ensure proper hole cleaning and avoid exceeding fracture pressures. Case
studies show that the model can be used to improve the available hole cleaning practices
and optimize drilling parameters and fluid properties to avoid hole cleaning problems.
Moreover, the model can be used to optimize the required circulating time for clean-up
cycles, which will benefit operational efficiency by reducing invisible lost time. Finally,
82
an MPD-controlled hole cleaning scenario is simulated to show that the integration of hole
cleaning modeling and choke control enables proper, real-time BHP control during the
83
Chapter 4: Deep Closed-Loop Geothermal Systems3
4.1 INTRODUCTION
The integrated thermal and hydraulic model introduced in Chapter 2 is modified for
use in geothermal applications to demonstrate the feasibility of this concept. Different well
geometries and pump rates are simulated to estimate the power generation of DCLGS
wells. The effects of open-hole completion of the lateral and insulation of the return section
many of the issues of concern with EGS, such as potential contamination of the circulating
fluid and short-circuiting. Deep CLGS wells drilled in rock formations with in-situ
temperatures above 200 ℃ could in theory be drilled anywhere around the world, thereby
integrating a managed pressure operation (MPO) system with deep CLGS (DCLGS) is
presented in this chapter. The concept includes an open-hole completion of the lateral
section, while the automatically controlled MPO system maintains wellbore integrity and
avoids fluid contamination. To demonstrate the feasibility of this concept, the combined
thermal and hydraulic model developed in Chapter 2 is modified for single phase flow of
water in DCLGS wells. The conduction heat transfer model is improved and validated
against PDE solvers. Two different geometries are simulated to show the potential of
DCLGS design, as well as the capabilities of the model in transient simulations and real-
time control of the well. Similar cases are also investigated for fully-cased wellbores, where
3The work in this chapter is previously published [115]. Q. Gu, D. Chen, P. Ashok, E. van Oort, and M.
Holmes contributed to the model development and manuscript revision of the published article.
84
sealing the lateral section is required for well control. Finally, effect of pump rate on the
Figure 4-1 shows the concept of a DCLGS well with an integrated MPO system.
The geothermal well in this particular example is U-shaped consisting of two vertical
wellbores that are drilled to the desired temperature region (200+ ℃) and connected by a
long horizontal section (note that other DCLGS configurations are possible as well, but we
are limiting the discussion in this chapter to a U-shaped well). This lateral section is
contact and efficient heat transfer between the fluid and the high-temperature rock.
Moreover, an open-hole completion of the lateral section avoids practical problems with
running casing and cementing downhole at extreme temperatures. Due to the variable
energy demand experienced by DCLGS wells and the frequent temperature cycling
associated with operating, starting, and shutting down, thermal expansion and contraction
of the casing and maintaining integrity of the cement sheet may present major issues. Such
problems are eliminated for the lateral section when it remains open to the formation as
proposed by the DCLGS concept (note that temperature cycling will still be an issue for
the cased-hole sections of the well). For power generation, low-temperature working fluid
is pumped into the well and is heated as it flows through the inlet vertical section of the
well and the horizontal lateral section, absorbing energy from the surroundings. The outlet
vertical section of the well is preferably thermally insulated to avoid heat loss to the
formation when the high-temperature fluid returns to surface. An automatic MPO system,
analogous to a MPD system used for the construction of challenging oil and gas wells [94],
is used to manage the pressure of the fluid flowing out of the exiting well. Using the MPO
85
system to apply surface backpressure, it is possible to both control the pressure-volume-
temperature (PVT) dynamics of the circulating fluid, as well as add additional hydrostatic
Moreover, by maintaining the pressure profile within the open-hole lateral section, the
MPO system avoids reservoir influx into the wellbore (in case the reservoir rock is
permeable) and eliminates fluid contamination issues. Finally, the energy is extracted from
the high-temperature fluid in a turbine or heat exchanger, and the cooled fluid is re-injected
Figure 4-1. Schematic of the MPO-controlled deep geothermal well concept, including
(1) inlet section (cased), (2) lateral section (open-hole), (3) outlet section
(case or VIT-insulated), (4) MPO choke, and (5) turbine/heat exchanger.
economically viable power source and establish an analytical foundation for subsequent
86
MPO control design, it is essential to first develop a thorough understanding of the
hydraulic and thermal behavior of the fluid under the pertaining HPHT conditions of a
DCLGS well. There are a few thermal models for the estimation of temperature profiles in
U-shaped CLGSs in the published literature. Sun et al. [95] provides a literature review of
the U-shaped systems. Numerical models of Sun et al. [96] and Sun et al. [97] estimate the
steady-state temperature profile in U-shaped wells using supercritical CO2 as the working
fluid. Schulz [7], Oldenburg et al. [58], and Song et al. [98] have developed models for
simulating shallow U-shaped geothermal wells using water as the working fluid. These
models can provide estimation of the temperature profiles over the lifetime of low-
temperature CLGS wells. However, fast transients and short-term effects, such as those
caused by changing the pump rate, are not considered in these models. Moreover, since
capturing pressure wave behavior is required for dynamic well control, such models can
only simulate cases where casing and heat conducting cement are used to seal off the
horizontal section and wellbore stability is already ensured. In order to evaluate the
The integrated model introduced in Chapter 2 can address this need and evaluate
the feasibility of the proposed DCLGS concept and ensure real-time MPO control. Note
that the lifetime performance of the geothermal system is beyond the scope of this work.
The long-time thermal depletion around the wellbore and the consequent power reduction
can be estimated using external formation heat transfer models [99,100]. The main
contributions of this chapter can be summarized as introducing a deep CLGS well concept
that enables access to high temperature reservoirs globally by: (1) considering an open-
hole completion of the lateral section for improved heat transfer, (2) adopting automatic
MPO control to avoid adverse fluid PVT behavior, wellbore instability and reservoir influx,
87
and (3) real-time hydraulic and thermal modeling that considers the fast transients relevant
to MPO control.
Section 2.2. Note that in the proposed DCLGS, the MPO system controls the PVT behavior
of the working fluid and avoids reservoir influx, reducing the flow to a single-phase liquid
flow at all times, which simplifies the solution algorithm and conservation equations. Due
estimations, the models used to calculate the thermal properties of the working fluid are
improved. The MPO controller is also implemented to ensure the PVT behavior of the
working fluid.
In a first approach, we assume that water is used as the circulation fluid. This is not
a restriction of DCLGS wells, and other working fluids such as supercritical CO2 can also
be considered. Due to the extreme pressure and temperature conditions in the DCLGS well,
assuming constant properties will lead to large errors. Water viscosity, for example,
decreases by a factor of 18 from 0 ℃ to 300 ℃. Therefore, fitted polynomial models of
water properties are used to calculate the density, viscosity, specific heat capacity at
constant pressure and constant volume, and thermal conductivity [79,101]. Density is
88
where pressure is measured in Pa, temperature in ℃, and density in kg/m3.
Neglecting the effect of pressure, the viscosity, specific heat capacities, and thermal
𝜇=
Due to the high temperatures in the well, water will boil as it is being circulated
back to surface. Large differences between the densities of liquid water and water vapor
lead to a large drop in hydrostatic pressure. This change in pressure makes control over the
BHP and maintaining well integrity (in the case of an open-hole completion) challenging.
Reservoir fluids could also enter the well (through any exposed permeable formations and
fractures) and pollute the water that is circulated out of the well. Moreover, evaporation
could lead to excessive pressures on surface due to the expansion, resulting in safety
hazards and risks and compromising wellbore integrity control. To address these issues and
control the water’s PVT behavior, return flow is controlled using an automatic choke such
89
that the entire operation is under MPO control at any time. A polynomial fit of the pressure
increases with the temperature [79]. At any given temperature, water starts boiling if the
In Eq. (4-6), temperature is measured in ℃ and pressure in Pa. Deep in the wellbore
itself, the hydrostatic pressure of the water column applies sufficient pressure to avoid
boiling downhole, even for extreme temperature gradients and high absolute downhole
temperatures. The well location where boiling is most likely to occur is on surface at the
exit due to the reduced hydrostatic pressure. Therefore, a proportional integral (PI)
controller that enables MPO is designed to maintain a surface backpressure (SBP) above
the boiling pressure by adjusting the choke opening automatically. Note that the MPO
system plays a crucial role in the proposed DCLGS concept. The MPO system controls the
fluid PVT behavior, maintains pressure control across the open lateral, ensures well
integrity, and avoids reservoir influxes and fluid contamination. Simulating the automated
choke control behavior relies on capturing the pressure wave dynamics and the fast
transients associated with rapid choke adjustments. Here, we simulate choke opening
where 𝑝𝑠𝑢𝑟𝑓𝑎𝑐𝑒 is the SBP that is maintained by the choke, 𝑝𝑏𝑜𝑖𝑙 is the boiling pressure on
margin that may also account for excess pressure needed for maintaining wellbore stability
90
in exposed open-hole sections in the well (note that this term can be further delineated if
real-time estimation of pore pressure and the pressure requirements for wellbore stability
frequency are used in the model to calculate the boiling pressure and maintain the required
SBP. For the simulations presented here, the safety margin is set to 1 MPa. It is assumed
that no further backpressure is required, and wellbore stability is already ensured [103].
Geothermal energy production is determined by the heat transfer rock, and the
capability of the rock formation to conduct heat from distant radius to the well. Due to the
importance of rock temperature dynamics and the effect of accurate rock temperature
estimations on the power generation of the well, the rock thermal model in Section 2.2.5 is
improved. The results of the thermal model are compared against built-in MATLAB
partial differential equation (PDE) solvers using fine meshing and small time-step.
Independence of cell size and time-step in the discretized model is also verified.
Assuming that there exists no fractures and the subsequent convection / advection
heat transfer within the formation rock, heat transfer is only limited to the conduction mode
(which is the worst-case scenario, studied in this work). In this case, the rock temperature
at a given measured depth (MD) along the well can be calculated from the 1-D radial
conduction equation 4-8. Note that this equation is coupled with the hydraulic-thermal
model of wellbore hydraulics, and the corresponding rock temperature is a 2-D model,
91
where 𝜌𝑟 , 𝐶𝑟 , and 𝑘𝑟 are the density, specific heat capacity, and thermal conductivity of
the rock formation, respectively, 𝑇𝑟 is the rock temperature (which is a function of time
and radial distance), 𝑡 is time, and 𝑟 is radial distance from the centerline. This PDE is
temperature:
temperature:
the rock face is equal to the convection heat transfer in the circulating
water:
𝜕𝑇𝑟
𝑘𝑟 . (𝑡, 𝑟 = 𝑟𝑜,𝑤𝑒𝑙𝑙 ) = ℎ. (𝑇𝑟 (𝑡, 𝑟 = 𝑟𝑜,𝑤𝑒𝑙𝑙 ) − 𝑇𝑤 ) ·································· (4-11)
𝜕𝑟
To accurately solve the near-well formation temperature dynamics within the rock,
in Equation 4-10). After 𝑟𝑒𝑛𝑑 , the rock formation is not affected by wellbore hydraulics,
and temperatures are equal to the initial geothermal temperatures. It is found that a distance
92
of 4.5 m is applicable for the time windows of this study ( < 5 days), and the formation
temperature changes beyond 4.5 m away from the well is negligible. Surrounding rock
formation is then discretized using a geometric method, where the cell size (i.e., thickness)
where 𝑡ℎ𝑖 is the thickness of the ith formation cell, and 𝛼 is the geometric coefficient. In
the simulations of this work, a geometric coefficient of 1.12 is used. The geometric
discretization enables using small cells near the wellbore, where accurate temperature and
heat conduction estimations are necessary for reliable simulation, while using larger cells
away from the well to reduce computational expenses. Results are compared for different
Figure 4-2. Geometric discretization of rock formation for heat transfer calculations
(with bi-lateral symmetry).
93
To verify the radial rock temperatures, a built-in MATLAB PDE solver (pdepe
function) is used to solve Equation 4-8, subject to the conditions of Equations 4-9 through
4-11, are results of the PDE solver are compared against the model. The PDE solver uses
3000 nodes to discretize the rock formation from the well-face to the “disturbed” distance
of 4.5 m from the well. Geothermal rock temperature is 250 ℃, and water is flowing at
350 m3/h and 70 ℃ through a 31.115 cm open-hole section of the well. Rock density,
specific heat capacity, and thermal conductivity are 2700 kg/m3, 1000 J/kg.K, and 2.5
temperature profiles are compared at 1 day and 5 days into the simulation (Figure 4-3). It
is observed that the model has a perfect match with the PDE solver. As a result, the
Chapter 5.
Figure 4-3. Comparison of the radial rock temperature profiles at 1 day and 5 days into
the simulation.
94
4.4 SIMULATION SCENARIOS
To demonstrate the utility of the model in simulating the operation and well control
of DCLGS wells, two cases with different geometries are discussed here. In both cases,
water is pumped at 300 m3/h into a U-shaped well for 13 hours, followed by a 3-hour
pumping at 400 m3/h, and 4 hours of pumping at 350 m3/h. The simulated scenario shows
how the dynamic temperature and generated power change with the changes of the pump
rate. The first well consists of two 7 km deep vertical sections (which would be drilled
independently) that are connected through a 7 km horizontal section (which could be drilled
as two 3.5 km sections drilled from the two independent wells that intersect at their end-
points), referred to as the “7/7” geometry. The wellbore diameter in the open-hole lateral
section is assumed to be 31.115 cm (12.25 in, in line with industry standard production
hole diameters in oil and gas wells) and the build radius of the curve sections of the well is
assumed to be 200 m. The initial temperature of the water in the well is assumed to be
equal to the temperature profile of the surrounding rock, linearly increasing from 40 ℃ at
locations such as the Los Angeles metropolitan area [59]). The second well has a TVD of
temperature gradient with depth is assumed to be constant and the same for the two
geometries, leading to a bottomhole temperature of 340 ℃ for the “10/5” geometry. For
both geometries, the density, heat capacity, and thermal conductivity of the rock formation
are set to be 2700 kg/m3, 1000 J/kg.K, and 2.5 W/m.K, respectively. These values represent
average crust properties [104]. While these values depend on the rock formation, they do
not vary significantly among different rock types [105]. Hence, for different sets of
formation properties, the resulted output temperature and generated thermal power
water is pumped continuously for 20 hours. During the simulation, the automatic controller
is used to apply enough backpressure to avoid water evaporation and associated pressure
reduction in the wells. The inlet and outlet vertical sections use 33.973 cm (13.375 in) outer
diameter casing. Casing and cement thicknesses for the cased sections are set to be 1 cm
Table 4-1. Relevant geometrical, formation, fluid and geothermal parameters used in
the simulations.
Geometry Fluid Properties
Vertical depth 7000 / 10000 m Working fluid Water
Horizontal length 7000 / 5000 m Density Eq. (4-1)
Build radius 200 m Viscosity Eq. (4-2)
Hole size (open-hole) 311.15 mm Specific heat capacities Eq. (4-3) and (4-4)
Hole size (casing) 317.88 mm Thermal conductivity Eq. (4-5)
Figures 4-4 and 4-5 show the water, rock formation (first node), and initial
temperatures versus MD at different times into the simulation for the “7/7” and “10/5”
equilibrium with the rock formation (e.g., after an extended shut-in period of no
pumped out of the well and replaced by cold water from the inlet section of the well. After
continuing pumping for a few hours, a semi-steady-state is reached in which all the water
is displaced by the cold water from the inlet section of the well. During the process, the
96
surrounding rock temperature continues to decrease due to the heat transfer to the water.
(i.e., thermal depletion of the well). As the water is circulated up in the return section, the
temperature of the surrounding rock drops, causing a heat loss from the water to the cold
rock formation near surface. The maximum gradient in the temperature profile is observed
at the beginning of the horizontal section, where the temperature difference between the
water and un-cased formation is at a maximum. It is observed that the increased pump rate
between 13 and 16 hours into the simulation reduces the temperature profile along the well.
This is because with the increased pump rate and fluid velocities within the wellbore, water
Figure 4-4. Temperature profile at different times for the “7/7” geometry (with open-
hole lateral).
97
Figure 4-5. Temperature profile at different times for the “10/5” geometry (with open-
hole lateral).
In the lateral section, water temperature increases along the well direction due to
the heat transfer with the formation at 250 ℃ or 340 ℃ for the “7/7” and “10/5” geometries,
respectively. The temperature gradient along this section slowly decreases as the water and
temperature difference (cf. Figures 4-4 and 4-5). Theoretically, in an infinitely long lateral
section, water temperature will reach the maximum possible downhole temperature, which
is equal to the undisturbed in-situ formation temperature. However, as the lateral length
increases, the total cost and complexity of drilling increases, while the gain in power
generation declines. Therefore, there exists an optimum lateral length beyond which the
gain in power generation does not compensate for the extra cost of drilling. This optimum
length depends on other parameters as well, most importantly the TVD (or equivalently,
the maximum geothermal temperature reached), the pump rate, and the hole size. The
developed model can be used to determine the optimum well geometry and pump rate for
98
well design and efficient power generation for a given set of subsurface pressure and
temperature conditions.
Figure 4-6 shows the pressure profile at different times for both geometries. The
pressure profile is affected by the hydrostatic pressure, frictional pressure drop, and SBP.
While the pressure is mostly determined by the hydrostatic pressure, a small pressure drop
is observed in the horizontal section due to friction for both geometries. At around 100
minutes, hot water that was initially in the well reaches surface and SBP is increased due
to the higher outlet temperature, requiring a higher pressure to avoid boiling. At this time,
because of the high temperature difference between the inlet and outlet sections of the well,
the density and hydrostatic pressure in the outlet section of the well reduces, resulting in a
lower BHP. The higher hydrostatic pressure gradient in the inlet section of the well results
in a zero pumping pressure, a situation known as the thermosiphon effect [106], which
reduces the power demand of the pumps. The thermosiphon effect is reduced as the
temperature in the outlet section of the well decreases towards the steady-state temperature.
With the changes in pump rate (at 13 h and 16 h into the simulation), the frictional pressure
drops along the well changes. It is observed that the increased pump rate of 400 m 3/h at
950 min into the simulation results in a higher BHP compared to the BHP at 1200 min (350
m3/h) and 750 min (300 m3/h). The BHP changes need to be accounted for when
determining the required SBP for well control over the open-hole lateral section.
99
Figure 4-6. Pressure profile at different times for geometries (with open-hole lateral).
Figure 4-7 shows DCLGS outlet temperature, the net thermal power generation,
and the outlet flow rate for both geometries while the pumping rate begins at 300 m3/h,
increases to 450 m3/h between 780 and 960 minutes, and reduces to 350 m3/h after 960
minutes. It is noted that the power values do not include any losses at surface due to power
conversion in e.g., steam turbines: the numbers reflect the net thermal power that is
generated by the producing well only. While the pump flow rate is kept at 300 m3/h for the
first 780 minutes into the simulation, oscillations are observed in the outlet flow rate during
the early stages of power generation. These variations are mainly caused by the
compression and expansion of the water due to pressure changes in the well (Figure 4-6),
as well as the thermal expansion behavior of the water within the wellbore. During the first
50 minutes, the hydrostatic pressure decreases due to the elevated temperatures in the
return section of the well. This causes water expansion in the well and increases the outlet
flow rate. At 50 minutes, the MPO system starts to increase the surface backpressure to
avoid evaporation. Due to the increased backpressure and the resulting compression of the
water, the outlet flow rate is decreased after this time. As steady-state approaches, the outlet
flow rate converges to the pump rate value. When the inlet pump rate changes at 780 and
100
960 minutes, the outlet flow rate converges to the inlet pump rate and the wellbore reaches
steady-state again.
Figure 4-7. Outlet temperature, thermal power and flow out versus time for both
geometries (with open-hole lateral).
The generated thermal power depends on the outlet temperature and outlet flow
rate. During the initial stages of circulation, the high-temperature water that initially stays
at bottomhole is circulated towards surface, increasing the outlet temperatures and the
resulting thermal powers. Thereafter, the temperature and power slowly decrease with time
due to the cooling of the heat-supplying rock, with some variations in the generated power
caused by the varying outlet flow rate. At 780 minutes (300 m3/h), the outlet temperature
and net thermal power for the “7/7” geometry are 128 ℃ and 30.6 MW (168 ℃ and 44.0
MW for the “10/5” geometry). With the pump rate increasing at 780 minutes, the net
101
thermal power is increased to 40.0 and 60.2 MW for the “7/7” and “10/5” geometries,
respectively. However, as the higher pump rate is maintained, the outlet temperature slowly
decreases due to the reduced heat transfer time, thus reducing the generated thermal power.
At 960 minutes, the pump rate drops to 350 m3/h, which reduces the net thermal power.
However, as the circulation is continued, wellbore temperatures are recovered due to the
improved heat transfer time at lower fluid velocities, increasing the outlet temperature and
thermal power. At 1200 minutes, the outlet temperature is 109 ℃ and the resulting net
thermal power is 28.1 MW for the “7/7” geometry. (142 ℃ and 41.4 MW for the “10/5”
geometry).
Figure 4-8 shows the SBP and choke opening versus time. As the outlet temperature
increases between 0-170 min (Figure 4-7), the boiling pressure increases according to
Equation (4-6). The automated controller calculates the boiling pressure, adjusts the choke
opening, and applies sufficient SBP to avoid evaporation and maintain well control. A
maximum SBP of 2.5 MPa and 4.5 MPa is applied at 75 minutes for the “7/7” and “10/5”
geometries, respectively. As the temperature and boiling pressure decrease, the choke is
slowly opened to reduce the excessive SBP, while maintaining a safety margin of 1 MPa
(assuming in this case that no excess safety margin is needed for open-hole stability control
in the horizontal lateral). Varying the pump rate will affect the outlet temperature, and the
choke is automatically adjusted to maintain the 1 MPa safety margin while the boiling
pressures vary.
102
Figure 4-8. SBP and choke opening versus time for both geometries (with open-hole
lateral).
completion of the lateral, it may not always be feasible to drill and maintain long open-
hole lateral sections. In this case, the lateral section needs to be cased and cemented to
ensure wellbore integrity. Figures 4-9 through 4-11 show the temperature profiles and the
generated thermal power for cased lateral sections for both geometries. The outlet
temperatures at 1200 minutes (where the pump rate is 350 m3/h) are 110 ℃ and 142 ℃ for
the “7/7” and “10/5” geometries, respectively. The outlet temperatures result in net thermal
powers of 28.2 MW and 41.4 MW for the “7/7” and “10/5” geometries, respectively. These
values are similar to the cases with open-hole lateral, showing that the heat transfer is
103
Figure 4-9. Temperature profile at different times for the “7/7” geometry (with cased
lateral).
Figure 4-10. Temperature profile at different times for the “10/5” geometry (with cased
lateral).
104
Figure 4-11. Outlet temperature and thermal power versus time for both geometries (with
cased lateral).
For a given well geometry, the generated power can be controlled by adjusting the
pump rate. In this section, the effects of three different pump rates (i.e., 250, 350, and 450
m3/h) on the power generation of the “7/7” well geometry are compared. In each case, the
pump rate is maintained for 20 hours until steady-state is reached. Figure 4-12 shows the
water temperature profile of the three different pump rates. The higher flow rate for the
case of 450 m3/h pump rate results in a lower output temperature compared to the other
two cases with lower pump rates. This is because at higher pump rates, the water has less
time flowing through the inlet and lateral sections, thus less time to absorb the heat from
the surrounding formation. Figure 4-13 shows the outlet temperature and the net thermal
power during the first 20 hours of pumping. The maximum outlet temperature is reached
at 194, 132 and 99 minutes for the 250, 350 and 450 m3/h cases, respectively. The outlet
temperatures at 1200 minutes are 128, 109 and 96 ℃ for the three cases, correspondingly.
105
It should be noted that the generated power slightly increases with the rising pump rate
despite a lower outlet temperature. This is due to the increased mass flow rate through the
well which carries more thermal energy per unit time to the surface and compensates for
the lower output temperature. After 1200 minutes, the generated thermal powers of the
250, 350 and 450 m3/h cases are 25.5, 27.9 and 29.3 MW, respectively. Varying pump rate,
and associated SBP changes controlled by the MPO system, is therefore an effective way
to deal with an ever-changing commercial power demand, which can also reduce the
Figure 4-12. Comparison of the temperature profiles for different pump rates.
106
Figure 4-13. Comparison of the outlet temperature and the generated power for different
pump rates.
4.5 CONCLUSIONS
system with a deep closed-loop geothermal system (DCLGS) for scalable power generation
concept considers wells at large well depth to access formations at 200 ℃ - 350 ℃ in-situ
controlled choke placed at the well outlet to ensure real-time wellbore integrity over the
open-hole lateral region and control the phase behavior of the circulating fluid.
model (Chapter 2) is used to describe the transient well hydraulics during the operation and
production stages of the DCLGS. The rock conduction heat transfer model is improved
107
using a geometric gridding method for accurate estimations of temperature dynamics
within the formation rock. The gridding is validated against robust PDE solvers. Compared
to existing geothermal models in the literature, the proposed model uses a fully-transient
semi-implicit algorithm that can capture the pressure waves and fast transients, which are
necessary for MPO control. The model can predict the generated thermal power for
operation scenario with various true vertical well depth and horizontal length, open-hole
or cased-hole lateral completions, heat insulation or lack thereof of the return flow, and
Simulation results show that the proposed integrated DCLGS concept with wells
drilled in temperature environments above 200 ℃ can initially generate net thermal power
in the range of 25-50 MW for a wellbore of 31.115 cm diameter and a circulating flow rate
in the range of 300-400 m3/h. The MPO control will enable geothermal wells to support
the base load as well as displace the spin reserve. Casing and cementing of the lateral
section (e.g., when necessary for well control and to avoid influx) does not significantly
affect the thermal performance of the well, which is due to the relatively small thickness
of cement and casing layers. It was also observed that while increasing the pump rate
reduces the outlet temperature, the generated thermal power could be increased by 15%.
108
Chapter 5: Different Designs & Numerical Analysis of DCLGS
depths, and lateral lengths are investigated in high level. In this chapter, the DCLGS
shaped and J-shaped) are quantitatively compared. Moreover, the effect of various
parameters on the overall outlet temperature and power generation of the closed-loop
system after 20 hours of operation is studied. Results of this chapter can be used to study
the geothermal potential of different locations, and to determine the optimum design and
5.1 INTRODUCTION
working fluids, geological conditions, etc. While the MPO concept is introduced based on
a U-shaped design in Chapter 4, other designs are proposed in the literature as well. One
of the most common designs is the J-shaped CLGS (Figure 5-1). In the J-shaped design,
the inlet and outlet sections of the well are separated through a VIT-insulated pipe within
the wellbore. In the J-shaped design, it is best to pump the cold fluid down the annulus to
allow heat transfer to the fluid within the vertical sections of the well. The hot fluid is then
returned to surface through the pipe. Compared to the U-shaped design, the advantage of
the J-shaped wells is that there is no need to drill two wells from the surface, reducing the
cost of drilling. Moreover, many depleted oil & gas wells exist around the world that could
be repurposed for geothermal energy production. Some of these wells such as extended
reach wells that are drilling in HPHT environments could be immediately retrofitted for
109
drilling production due to their access to high-temperature reservoirs and extended lateral
sections. On the other hand, as the inlet and return flow are both within the same wellbore,
there is the risk of energy loss from the hot to cold fluid, reducing the performance of the
geothermal system. Moreover, due to the reduced area, the fluid velocities are higher in J-
shaped wells with similar outer diameter, increasing the frictional pressure drop and
required pump powers, as well as reducing the contact time with the high-temperature rock
for effective heat transfer. To avoid these issues, larger diameter wells are required for J-
shaped designs in general, which could lead to more challenges during the drilling of the
large boreholes.
Figure 5-1. J-shaped closed-loop design with the integrated MPO controller.
110
Wellbore geometry and operational parameters also affect the performance of the
geothermal system. The main 3 parameters that define the performance of the closed-loop
system are:
2. Lateral length: Most of the heat transfer within the geothermal system
the heat transfer time, allowing the fluid temperature to increase towards
model that trades off between the cost of drilling and increased power.
3. Pump rate: Increasing the pump rate would directly increase the power
generation due to the increased mass rate out of the well. However, at
higher rates, fluid velocities are increased, providing less heat transfer
Moreover, increasing the pump rate also increases the frictional pressure
hydraulic model, and the effect of these parameters on the outlet temperature and the
generated thermal power are quantified. Other than these parameters, the insulation of the
return section, hole size, and geothermal temperature gradient effects are investigated.
In this section, results of the numerical analysis are provided. The simulations
consist of continuous operation for 20 hours with constant pump rate. The outlet
temperature and generated thermal power after 20 hours of operation are used to determine
the performance of the DCLGS well and to study the effects of different parameters on the
performance.
Two closed-loop designs are compared here: (1) a U-shaped well with a hole
diameter of 31.115 cm (similar to the case presented in Chapter 4), and (2) a J-shaped well
with a hole diameter of 31.115 cm and a 17.78 cm VIT-insulated pipe. Both wells are 7 km
deep with a 7 km barefoot lateral. In both cases, water is pumped at 350 m3/h and 40 ℃
The bottomhole rock temperature is 250 ℃. The formation rock has a density of 2700
Kg/m3, a thermal conductivity of 2.5 W/m.K, and a specific heat capacity of 2500 J/Kg.K.
Both simulations are continued for 20 hours and the results are compared for both cases.
PVT phase behavior of the fluid is controlled in both cases using the MPO controller.
Figure 5-2 and Figure 5-3 show the temperature profile at different times within the
circulation for the U-shaped and J-shaped wells, respectively. In the U-shaped well, the
water temperature increases along the inlet vertical section and the open-hole lateral
4This section is previously published [1]. E. van Oort, D. Chen, and P. Ashok contributed to the simulation
analysis and manuscript revision of the published article.
112
section. The maximum temperature gradient is observed at the beginning of the open hole
lateral section, where the temperature difference between the water and surrounding rock
is maximum. In the return section, hot water is continuously heated by the surrounding
rock until the temperature of the hot water is in equilibrium to that of the surrounding rock.
From this location to the surface, to avoid heat loss, the well wall could potentially be
insulated using VIT, which is studied later in Section 5.2.3. Similar temperature profiles
are observed in the J-shaped well. The only difference is that the return section in this case
is through the VIT-insulated pipe inside the annulus. The increased velocity through the
VIT-insulated pipe (with an inside diameter of 13.97 cm) results in a temperature increase
along the return section. This temperature increase, however, comes at the price of
Figure 5-2. Temperature profile at different times for the U-shaped design.
113
Figure 5-3. Temperature profile at different times for the J-shaped design.
Figure 5-4 shows the outlet temperature, thermal power generation, choke opening
and surface backpressure for both wells. At early stages of the simulation, the high-
temperature fluid in the well is returned to surface, increasing the outlet temperature and
thermal power. The maximum outlet temperatures are 203 ℃ and 253 ℃ for the U-shaped
well and the J-shaped well, respectively. As the hot fluid in the well is replaced by the cool
water from pumping, the temperature drops and reaches a semi-steady-state. This outlet
temperature and thermal power, however, are perpetually decreasing due to the thermal
depletion of the rock, which is taken into account by the model through the discretization
of the surrounding rock. After 20 hours of circulation, the outlet temperature and thermal
power of the U-shaped well are 109 ℃ and 27.9 MW, respectively. These values for the J-
shaped design are 101 ℃ and 24.9 MW, respectively. In both cases, the choke opening is
above the evaporation pressure during the transient operation. It is assumed that additional
114
backpressure is not required to maintain wellbore stability, and that the required surface
Figure 5-4. Outlet temperature, thermal power, choke opening, and surface back
pressure for U-shaped and J-shaped designs.
5.2.2 Sensitivity Analysis (Effect of Vertical Depth, Lateral Length, and Pump
Rates)
Vertical depth, lateral length, and the pump rate values are varied for the U-shaped well
according to the matrix of Table 5-1. For the entire sensitivity analysis, water is pumped at
40 ℃, while the phase behavior is controlled using the MPO controller. The surrounding
rock has an initial temperature gradient of 0.03 ℃/m, with a density of 2700 Kg/m3, a
thermal conductivity of 2.5 W/m.K, and a specific heat capacity of 2500 J/Kg.K. The lateral
section of the well is completed in openhole, and the inlet and outlet sections are cased and
cemented. Simulations are continued for 20 h at a constant pump rate, and the final outlet
115
Table 5-1. Sensitivity analysis matrix (180 simulation cases).
Number of Cases Minimum Value Maximum Value Interval
Vertical Depth (m) 6 5000 10000 1000
Lateral Length (m) 6 2000 7000 1000
3
Pump Rate (m /h) 5 250 450 50
Figure 5-5 shows the outlet temperature and generated thermal power after 10 hours
of circulation against flow rate and total well length. Outlet temperature and thermal power
are both increased with an increase in well length, which is due to the access to rock
formations at higher temperatures (in the case of increased vertical depth, TVD), as well
as the increased heat transfer time along the lateral section (in the case of increased lateral
length). The effect of well length on the thermal power is more important at higher
circulating flow rates, where the fluid temperatures are reduced, increasing the heat transfer
and the generated thermal power. Increasing the pump rate has two effects: (1) the
increased fluid velocities decrease the heat transfer time, reducing the heat transfer and
fluid temperature gradient along the wellbore, which results in a lower outlet temperature;
(2) with the increased flow rate and the subsequent mass rate into and out of the geothermal
system, the generated thermal power is increased. The effect of increased mass transfer on
the generated power is found to be more important than that of the reduced outlet
temperature, resulting in a higher thermal power at higher flow rates. The effect of pump
rate on both temperature and thermal power is more pronounced for longer wells, where
the power can increase by ~25% when pump rate is increased from 250 to 450 m3/h. In
short wells (below 15 km total length), the effect of pump rate on thermal power is less
than 10%.
116
Figure 5-5. Temperature (left) and thermal power (right) at 10 h versus total well length
and pump rate.
Figure 5-6 shows the outlet temperature and generated power versus the vertical
depth (TVD) of the wellbore for different lateral lengths. The plots are presented for flow
rates of 250, 350, and 450 m3/h. Increasing the vertical depth increases the temperature of
the surroundings, enabling more heat transfer and generated power. At low pump rates
(250 m3/h), increasing the vertical depth from 5000 to 10000 m increases the outlet
temperature from 76 to 142 ℃ for a 2000 m lateral, and from 99 to 174 ℃ for a 7000 m
lateral. The corresponding thermal powers change from 10.3 to 29.4 MW for a 2000 m
lateral, and from 17.0 to 38.8 MW for a 7000 m lateral. The effect of vertical depth on the
thermal power generation is more pronounced at higher flow rates and lateral lengths,
where the temperature difference between the fluid and the rock is larger. Similarly,
increasing the lateral length from 2000 to 7000 m improves the power generation.
117
However, the effect of lateral length on the performance of the wellbore is less than vertical
depth due to the lower temperature of the surrounding rock formations. At higher pump
rates and vertical depths, the effect of increasing lateral length is slightly higher, which is
due to the increased temperature difference and heat transfer between the fluid and the
rock.
Figure 5-6. Outlet temperature (left) and thermal power (right) versus TVD for different
horizontal lengths and pump rates.
Figure 5-7 shows the outlet temperature and thermal power versus flow rate for 5
wells with different vertical depth (h) and lateral lengths (l): (1) l=2000 m and h=5000 m,
(2) l=2000 m and h=10000 m, (3) l=5000 m and h=8000 m, (4) l=7000 m and h=5000 m,
and (5) l=7000 m and h=10000 m. The results show that increasing pump rate reduces the
outlet temperature due to the increased fluid velocities and reduced heat transfer time. For
example, in a 5000 m deep well with a 2000 m lateral section, increasing the flow rate from
118
250 to 450 m3/h reduces the outlet temperature from 76 to 61 ℃. The effect of pump rate
is more noticeable in deeper and longer wellbores. For instance, in a 10000 m deep
wellbore with a 7000 m lateral section, the outlet temperature reduces from 168 to 128 ℃
when the flow rate is increased from 250 to 450 m3/h. Although the outlet temperature is
reduced at high flow rates, thermal power generation is still increased due to the increased
mass flow rate through the wellbore. In a 2000 m deep wellbore with a 7000 m lateral,
thermal power generation is increased from 10.3 to 11.1 MW, when pump rate is increased
from 250 to 450 m3/h. The small effect of pump rate is due to the lower temperature
difference between the fluid and the rock in shallow wells, where fluid temperature
gradients are low regardless of the pump rate, resulting in a similar heat transfer between
the fluid and the rock. In deeper and more extended wells, however, the temperature
difference is high. Therefore, higher pump rates lead to smaller fluid temperature gradients
due to the increased velocities. The smaller temperature gradients at high flow rates
increase the temperature difference between fluid and the rock, which further improves the
heat transfer and the generated thermal power. For example, in a 10000 m deep well with
a 7000 m lateral section, increasing the pump rate from 250 to 450 m3/h increases the power
119
Figure 5-7. Temperature (left) and thermal power (right) versus pump rate for different
geometries.
The outlet section of the proposed U-shaped CLGS design could be insulated using
VIT to avoid heat transfer from the hot fluid to the cool rock formations at shallow sections
of the well near surface. Looking at the temperature profiles of non-insulated wellbores
(e.g., Figure 5-2) shows that the hot fluid that enters the outlet section is initially in contact
with hot temperature rocks close to the bottomhole. As the fluid flows towards surface, the
temperature of the surrounding rock is reduced at shallower regions of the outlet section.
When the wellbore is not insulated, the temperature of the fluid would continue to increase
within the deeper parts of the outlet section where the rock temperature is higher than the
fluid temperature. The temperature gradient is zero at the point where the rock and fluid
5This section is previously published [115]. Q. Gu, D. Chen, P. Ashok, E. van Oort, and M. Holmes
contributed to the simulation analysis and manuscript revision of the published article.
120
temperatures are equal, and is negative above that point, i.e., heat is transferred to the low-
temperature surrounding rock at shallow sections. For a given geometry and operational
flow rate, the point of zero temperature gradient can be found using the model, and
insulation depth can be determined. Insulating deeper than this point would eliminate the
desired heat transfer at deeper parts of the outlet sections, reducing the generated power.
In this section, the effect of optimum insulation of the outlet section of the well is
studied. The wellbore in this case is 7 km deep with a 7 km lateral. Water is pumped at 40
℃ for the entire simulation. Pump rate is set to 350 m3/h for the entire 20-hour simulation.
In The surrounding rock has an initial temperature gradient of 0.03 ℃/m and a bottomhole
temperature of 250 ℃, with a density of 2700 Kg/m3, a thermal conductivity of 2.5 W/m.K,
and a specific heat capacity of 2500 J/Kg.K. Two cases are compared: (1) using a
conventional casing string and cement sheet similar to the inlet section of the well (referred
to as the “original case”) and (2) insulating the near surface parts of the return section of
the well where the rock temperature is lower than the fluid temperature (referred to as the
“insulated case”). In this case, the optimum insulated depth is 2 km, where the remaining
Figure 5-8 shows the temperature profile of the two cases at different times.
Temperature profiles are similar along the inlet and lateral sections of the well for both
cases due to the same external conditions. However, the heat loss to the near-surface
surrounding rock in the original case results in a temperature drop along the top 2 km of
the outlet section of the well. This effect is more pronounced at around 100 minutes due to
the larger temperature difference between the water and the rock formation behind the
cement sheet, resulting in more heat loss. At 1200 minutes, a temperature drop of about 2
℃ is observed in this case, while the temperature in the outlet section of the well near
121
surface for the insulated case remains relatively unchanged. This effect would be more
Figure 5-9 compares the outlet temperature and thermal power generation for the
two cases. At 1200 minutes, the thermal power of the insulated case is about 0.6 MW higher
than that of the non-insulated case. The maximum observed surface temperature is 221 ℃
and 203 ℃ for the insulated case and the original case, respectively.
122
Figure 5-9. Comparison of the outlet temperature and the generated power for
uninsulated versus insulated wellbores.
One of the parameters that affects the performance of the geothermal system is the
hole size. Larger holes are more challenging to drill and maintain, especially under HPHT
conditions in deep systems. Moreover, the increased hole size greatly increased the total
volume of the well and the amount of fluid required to circulate within the system. On the
other hand (at similar flow rates), the fluid velocities are reduced in large holes, increasing
the heat transfer time between the fluid and the rock and the temperature gradient in the
well. Moreover, the lower velocity leads to reduced friction and frictional pressure drop
In this section, 3 different hole sizes are compared: (1) 21.59 cm, (2) 25.08 cm, and
(3) 31.12 cm. In all cases the well is 7 km deep with a 7 km lateral section. Water is pumped
in at a temperature of 40 ℃ for the entire simulation. Pump rate is set to 350 m3/h for the
entire 20-hour simulation. The surrounding rock has an initial temperature gradient of 0.03
℃/m and a bottomhole temperature of 250 ℃, with a density of 2700 Kg/m3, a thermal
123
conductivity of 2.5 W/m.K, and a specific heat capacity of 2500 J/Kg.K. Figure 5-10 shows
the temperature profile at different times. In smaller hole sizes, the increased fluid velocity
leaves little time for heat transfer, reducing the temperature gradient along the well. As a
result, the outlet temperature is reduced, and the generated thermal power will be lower in
smaller hole sizes (with similar pump rates). Moreover, the increased velocities in smaller
hole sizes lead to higher frictional pressure drop, increasing the required pumping power.
This will further reduce the performance of the well. Note that if the desired outlet
temperature is fixed, larger holes allow for higher pump rates which increased the
generated power due to the increased mass rate across the well. As a result, larger holes are
always desirable, especially if technical issues and drilling conditions do not pose further
Figure 5-10. Fluid temperature profile at different times for 21.59, 25.08, and 31.12 cm
wellbores.
Figure 5-11 shows the outlet temperature and thermal power for different hole sizes
(21.59, 25.08, and 31.12 cm). In small hole sizes, due to the increased velocities, the hot
124
fluid that was initially at bottomhole will be circulated to surface faster. As a result, the
maximum outlet temperature is observed after 70, 90, and 130 minutes into the simulation,
for the 21.59, 25.08, and 31.12 cm wellbores, respectively. The maximum outlet
temperatures at these times are 207, 206 and 203 ℃, respectively. In 31.12 cm case, the
reduced velocities increase the heat transfer time, reducing the temperature of the initial
hot fluid (at 250 ℃) as it is circulated to surface. After 20 hours of circulation, the outlet
temperatures are 98, 102, and 109 ℃, for the 21.59, 25.08 and 31.12 cm wellbores,
respectively. These temperatures correspond to a thermal powers of 21.5, 25.0, and 27.9
MW, respectively.
Figure 5-11. Outlet temperature, thermal power, and flow out rate for 21.59, 25.08, and
31.12 cm wellbores.
observed, which directly affect the performance of DCLGS wells with given geometry and
125
operational conditions. Geothermal gradients of up to 0.2 ℃/m are observed in Iceland
[107]. These gradients, however, are extreme conditions that are not observed in many
regions in the world. Looking at the United States, for example, geothermal temperatures
of ~300-350 ℃ at 7.5 km depth are available within many regions of the Western US
(Figure 5-12). While these high geothermal temperatures are not globally available, in
regions with desirable geothermal gradients, the performance of geothermal systems would
be greatly improved. This paves way for more economical geothermal energy exploration
and production in these regions, where similar wells can produce more geothermal power.
Before global application, it is best to implement the proposed concept in regions with the
a U-shaped DCLGS with a 7000 m vertical section and a 7000 km barefoot lateral is
studied. Water is pumped in at a temperature of 40 ℃ for the entire simulation. Pump rate
126
is set to 350 m3/h for the entire 20-hour simulation. In The surrounding rock has a density
of 2700 Kg/m3, a thermal conductivity of 2.5 W/m.K, and a specific heat capacity of 2500
J/Kg.K. 3 different thermal gradients are explored: (1) 0.023 ℃/m, which leads to a
bottomhole temperature of 200 ℃ (at 7000 m TVD), (2) 0.030 ℃/m, which leads to a
bottomhole temperature of 250 ℃, and (3) 0.037 ℃/m, which leads to a bottomhole
temperature of 300 ℃. Transient temperature profiles, as well as the outlet temperature and
Figure 5-13 shows the temperature profiles at different times for the three cases. In
the cases with higher temperature gradients, fluid is in contact with higher-temperature
surrounding rocks. Due to the larger temperature difference between the fluid and the rock,
heat transfer rate is increased, leading to a higher temperature gradient across the entire
geothermal well, i.e., inlet, lateral, and vertical sections. Upon fluid return to surface, the
heat loss to the low-temperature rock near surface is greater in the case of higher
temperature gradients, which is due to the larger fluid temperatures. This shows that
optimized thermal insulation of the return section (which is explored in Section 5.2.3) is of
greater importance in high-temperature fields. The thermal insulation can therefore further
increase the performance of DCLGS wells in geothermal fields with higher temperature
gradients.
127
Figure 5-13. Fluid temperature profile at different times for 200, 250, and 300 ℃
bottomhole temperatures at 7000 m TVD.
Figure 5-14 shows the comparison of the outlet temperature and thermal power for
the 3 cases. The outlet temperature is always higher in the wells with higher temperature
gradient. During the first 130 minutes of circulation, hot fluid that was initially at high
A maximum outlet temperature of 165, 203, and 242 ℃ is observed at 130 minutes for the
bottomhole temperatures of 200, 250, and 300 ℃, respectively. After 130 minutes, a semi-
steady-state is reached, where all the fluid that was initially in the well is circulated out.
With continuous circulation, the temperature of the near-wellbore formation rock decreases
due to the heat transfer with the circulating water (i.e., thermal depletion of the rock
formation). This temperature drop reduces the heat transfer, slowly reducing the outlet
temperature and generated power. After 20 hours of circulation, the outlet temperatures are
93, 109, and 125 ℃, for the bottomhole temperatures of 200, 250, and 300 ℃, respectively.
128
Therefore, the generated thermal powers at this time are 21.4, 27.9, and 31.4 MW,
respectively.
Figure 5-14. Outlet temperature, thermal power, and flow out rate for 200, 250, and 300
℃ bottomhole temperatures at 7000 m TVD.
5.3 CONCLUSIONS
analyzed. Various designs and parameters that affect the performance of closed-loop wells
are studied. J-shaped and U-shaped closed-loop designs are compared. The effects of
lateral length, vertical depth, pump rate, optimum insulation of the return section, hole size,
and geothermal temperature gradient on the outlet temperature and generated power are
also studied.
Comparing J-shaped and U-shaped designs shows that for similar geometries, the
U-shaped design can produce 12% more power due to the lower fluid velocities and
increased heat transfer time. Moreover, the increased fluid velocities and frictions in the J-
shaped design increase the required pumping power, reducing the net power generation.
129
Optimum thermal insulation of the return section is found to avoid heat loss to the low
temperature rock formations near surface, thereby increasing the outlet temperature and
power generation. This effect is more important in deeper wells that produce higher outlet
temperatures. It is also found that increasing the hole size from 21.59 to 31.12 cm increase
the power generation by 30%, which is due to the increased heat transfer time. Drilling
wells in desirable fields with high geothermal temperature gradients was found to greatly
affect the power generation. Power generation is increased by 50% when the geothermal
Vertical depth, lateral length, and flow rate were analyzed as the key parameters
that affect the performance of DCLGS wells, influencing the maximum rock temperature,
heat transfer time, and mass flow rate, respectively. It is found that increasing the vertical
depth from 5000 to 10000 m can improve the power generation by 120-220%, with the
effect of vertical depth being more pronounced in higher lateral lengths and pump rates.
Increasing the lateral length from 2000 to 7000 m can also improve the power generation
by 30-70%, where the effect of lateral length is more pronounced at higher pump rates and
shallower wells. While the effect of pump rate in shallow wells is less than 10%, in deeper
wells (> 7000 m deep), increasing the pump rate from 250 to 450 m3/h can improve the
130
Chapter 6: Summary & Future Work
6.1 SUMMARY
single-phase liquid flow. The model solves the Navier-Stokes equations for multi-phase
1D pipe (and annulus) flow, considering detailed phenomena such as the fluid
dependent thermal properties, gas solubility in non-aqueous drilling fluids, solid settling in
non-vertical wellbores, area discontinuity, heat transfer within the adjacent formation rock,
etc., which govern the physics of multiphase flow under high-pressure high-temperature
numerical scheme, allowing real-time simulations under various conditions and well
control operations. The scheme is modified for optimum performance in each flow
barefoot sections of the well is necessary for optimum well control. This is achieved
through an automatically adjusted choke at the outlet of the well. The automated choke
controller, however, requires reliable estimates of the downhole conditions for accurate
pressure control over the openhole region, which is provided through the advanced
equations enables fast but stable simulations using small time-steps (on the order of 10-2
131
s). The use of small time-steps allows the model to capture pressure waves within the
wellbore, which is necessary for simulating the fast transients associated with rapid choke
adjustments during managed pressure drilling (MPD) and managed pressure operation
(MPO) practices.
1. (in Chapter 2) Detect and simulate gas kicks in HPHT wells, model dynamic
well control using automated bottomhole pressure (BHP) control, and estimate
the dissolved gas density, break-out point and free gas volume. This provides
the drilling crew with valuable information to detect any potential kick as soon
2. (in Chapter 3) Simulate hole cleaning operations, estimate the bed height and
and providing efficient and reliable clean-up cycles before pulling out of hole
(POOH).
geothermal wells, estimate the transient outlet temperature and generated power
well control during the operation of the geothermal system, compare different
closed-loop designs, and explore and quantify the effect of various parameters
Although the model is mainly developed for simulating drilling and operation of
(EGS) wells, oil & gas wells (which was explored in Chapters 2 & 3), and any other pipe
flow under HPHT conditions, where detailed pressure control and modeling of pressure
132
waves is necessary. Moreover, due to the comprehensive modeling, the developed model
could be used to generate training data for data-driven and real-time physics-based models,
and to train rig personnel. Finally, the model could be used to develop and test novel MPD
controllers [82,108–110].
In this section, future work directions are stated to improve and extend the
capabilities of the developed model. Since the use of the model can be divided into 3 main
directions (i.e., kick modeling (liquid-gas), cuttings transport modeling (liquid-solid), and
thermal power estimation (liquid)), the future work in each direction is explored separately:
Kicks simply occur when the wellbore pressure falls below the reservoir/pore
pressure at openhole sections. However, there are many factors that could lead to inaccurate
During the drilling operations, solid particles (i.e., cuttings) are always available
within the well. Due to the high density, accumulation of these cuttings could greatly affect
the downhole pressures and affect kick control. Moreover, the cuttings particles affect the
fluid rheology, influencing the friction factor and pressure drop during circulation.
Addition of a simplified cuttings transport model to the kick control could take the effects
of the cuttings on well control into account. In this case, the flow would be extended to a
fluctuations, which could lead to small kicks. After each stand is drilled, the flow is stopped
133
for a short while to make a connection and continue drilling, which happens at every ~30
and the choke controller needs to apply sufficient backpressure to maintain the downhole
pressure. Another issue is the instances when the drillstring is run into the hole/pulled out
of hole, known as the swab & surge effects. During swab, running into the hole temporarily
increases the BHP, which could fracture the formation. Similarly, during surge, pulling out
of hole reduces the BHP, which could lead to an influx known as swab kick. At small
drilling windows (where the pore pressure and fracture pressure are close to each other),
accurate and reliable pressure control is necessary, as small fluctuations in BHP could
fracture the formation or introduce kicks. For simulation of continuous drilling and well
control operations, the effects of making connections, swab, and surge on the pressure
Hole cleaning is concerned with accurate estimation of the bed blockage, cuttings
concentration, and the volume and location of the cuttings in the well. To gather more
information regarding the volume of cuttings in the well, real-time cuttings sensors are
developed and studied that can measure the volume of cuttings that are returned to surface,
providing the crew with accurate estimates of the amount of cuttings in the well [111]. The
real-time sensor also provides the crew with other valuable information such as the density
and average size of cuttings, which is an input to the physics-based model. Integrating a
real-time cuttings sensor with the developed model will greatly improve the accuracy and
reliability of downhole estimations, helping the crew members to take proper actions when
necessary.
134
Secondly, the cuttings that are generated at the bit are not of the same size. In fact,
there is a distribution of cuttings sizes in the well, varying from fine solid particles to coarse
cuttings (and sometimes cavings). Particles of different sizes behave differently as they are
transferred to the surface. Larger cuttings tend to move slower in the vertical section due
to their weight and require more time to clean the wellbore from them. Improving the model
such that the average and standard deviation of the cuttings sizes are used as an input (rather
than a deterministic average size) enables the model to provide more accurate estimates of
the cuttings bed and concentration, as well as information regarding the clean-up time
In deviated wells, or at high flow rates, the settled cuttings (i.e., cuttings bed) can
move up towards the surface with the flow or collapse when the flow is stopped (also
deviated section, which could lead to excessive pressure, lost circulation, and other issues.
cuttings in case of an avalanche so that the crew can take necessary actions to safely restart
the circulation and clean the hole. Moreover, simulating unstable beds improves the
accuracy of the model by taking the effects of moving beds into account.
Drilling fluids are non-Newtonian and are designed to have increased viscosity
when pumping is stopped during connections, known as the gel effect. This way, the
cuttings particles in the non-horizontal sections of the well remain in suspension and do
not settle down. The gelling effect is not investigated in this work. Adding the capability
of mud gelling during connections, and breaking the gel when circulation is continued is
Moreover, desired mud properties and gel strengths could be determined to ensure hole
constantly increased. The current developed model assumes that the wellbore length and
cycles and short drilling periods with high accuracy. However, to simulate extended
drilling operations where the wellbore increase cannot be ignored, the model needs to
constantly update the well length and geometry as the drilling is continued. Adding this
capability to the model allows for accurate simulations over long time periods, which do
not require separate simulations to analyze cuttings transport at different well lengths.
~30-year time windows. While short-term simulations in this work are necessary for well
control and transient power generation, long-term performance of the wellbore determines
comprehensive simulation of transient hydraulics and heat transfer, and even though
simulations are real-time (around 2 times as fast as the real operation), the model cannot
be used for long-term performance of the geothermal well over its life cycle. A faster, less
accurate estimations of the long-term generated power and outlet temperature under
source, or equivalently, the volume of the rock that is affected and cooled down by the
circulating flow. Due to the small thermal conductivity, the volume of the rock that is
cooled down is relatively small, reducing the potential performance of closed-loop wells.
One way to increase the rock volume is to drill multiple lateral sections that are distanced
136
from each other. This way, each lateral section will receive heat from the surrounding rock
up to a certain distance, and its effect of the formation rock further away is negligible. This
effective radial distance where the rock temperatures are reduced due to the presence of
the lateral section can be calculated using radial conduction models and used to determine
the optimum distance between the lateral sections for optimum heat extraction (i.e., the
lateral sections are distanced far enough that each rock zone is affected by a single lateral,
allowing other laterals to extract heat from a different, more intact rock zone).
Another way to improve the heat extraction from the rock is to improve the
effective thermal conduction of the formation rock. While dry rock is not thermally
conductive, the presence of natural or induced fractures could greatly improve the heat
transfer within the rock. It is observed that cooling of rock formations could lead to
of the closed-loop well since the near-wellbore rock formations are cooled from ~200-350
filled with the circulating fluid and act as heat convection paths, which is far more effective
than heat conduction through the solid rock. In some cases in the literature, the presence of
natural fractures and heat convection is found to improve the heat transfer by around 6
times compared to a similar rock formation where conduction is the dominant heat transfer
form [58]. To accurately quantify the effect of fractures on the performance of the proposed
deep closed-loop system, the rock heat transfer network needs to be extended to simulate
heat convection within the formation rock into account. Moreover, with a more
sophisticated heat transfer model, the effect of hydraulic fracturing can be quantified,
which provides investors with better information on the potential performance of high-
porosity geothermal fields, as well as the advantage versus cost analysis of fracturing the
here. In many fields, supercritical CO2 is being used due to the advantages such as reduced
viscosity and avoiding corrosion in surface equipment. In cases where the hydrostatic
pressure of the water column is not necessary to ensure wellbore stability over the open-
analysis of CO2 versus water is required to accurately determine the advantage of each
fluid, and quantitatively compare the power generation and performance of the closed-loop
system.
138
Glossary
DG = dissolved gas
ID = inner diameter
MD = measured depth
MPD = managed pressure drilling
OD = outer diameter
PI = proportional integral
PVT = pressure-volume-temperature
139
RDFM = reduced drift-flux model
ST = solubility threshold
140
References
142
single and multi-phase flow operations during drilling, completion, well control
and intervention. IADC/SPE Asia Pacific Drill. Technol. Conf., 2008, p. 242–53.
https://doi.org/10.2118/114688-ms.
[26] Sun B, Sun X, Wang Z, Chen Y. Effects of phase transition on gas kick migration
in deepwater horizontal drilling. J Nat Gas Sci Eng 2017;46:710–29.
https://doi.org/10.1016/j.jngse.2017.09.001.
[27] Xu Z, Song X, Li G, Zhu Z, Zhu B. Gas kick simulation in oil-based drilling fluids
with the gas solubility effect during high-temperature and high-pressure well
drilling. Appl Therm Eng 2019;149:1080–97.
https://doi.org/10.1016/j.applthermaleng.2018.12.110.
[28] Ambrus A, Aarsnes UJF, Karimi Vajargah A, Akbari B, van Oort E, Aamo OM.
Real-time estimation of reservoir influx rate and pore pressure using a simplified
transient two-phase flow model. J Nat Gas Sci Eng 2016;32:439–52.
https://doi.org/10.1016/j.jngse.2016.04.036.
[29] Gu Q, Fallah A, Ambrus A, Chen D, Ashok P, van Oort E. Higher precision
automated managed pressure drilling control achieved through the addition of a
thermal model. Int. Pet. Technol. Conf., 2019. https://doi.org/10.2523/iptc-19326-
ms.
[30] Ma Z, Karimi Vajargah A, Ambrus A, Ashok P, Chen D, van Oort E, et al. Multi-
phase well control analysis during managed pressure drilling operations. Proc. -
SPE Annu. Tech. Conf. Exhib., Society of Petroleum Engineers; 2016.
https://doi.org/10.2118/181672-ms.
[31] Ma Z, Vajargah AK, Ambrus A, Ashok P, Chen D, van Oort E, et al. A
Comprehensive Hydraulic Software Package for Drilling Operations. AADE Natl.
Tech. Conf. Exhib., 2017.
[32] Ma Z, Karimi Vajargah A, Chen D, van Oort E, May R, MacPherson JD, et al. Gas
kicks in non-aqueous drilling fluids: A well control challenge. IADC/SPE Drill.
Conf. Exhib., vol. 2018- March, Society of Petroleum Engineers; 2018.
https://doi.org/10.2118/189606-ms.
[33] Ramadan A, Skalle P, Johansen ST, Svein J, Saasen A. Mechanistic model for
cuttings removal from solid bed in inclined channels. J Pet Sci Eng 2001;30:129–
41. https://doi.org/10.1016/S0920-4105(01)00108-5.
[34] Larsen TI, Pilehvari AA, Azar JJ. Development of a New Cuttings-Transport
Model for High-Angle Wellbores Including Horizontal Wells. SPE Drill Complet
1997:129–35.
[35] Adari RB, Miska S, Kuru E, Bern P, Saasen A. Selecting drilling fluid properties
and flow rates for effective hole cleaning in high-angle and horizontal wells. SPE
Reserv Eng 2000:273–82. https://doi.org/10.2118/63050-MS.
143
[36] Shah SN, El Fadili Y, Chhabra RP. New model for single spherical particle settling
velocity in power law (visco-inelastic) fluids. Int J Multiph Flow 2007;33:51–66.
https://doi.org/10.1016/j.ijmultiphaseflow.2006.06.006.
[37] Kelessidis VC, Mpandelis G. Measurements and prediction of terminal velocity of
solid spheres falling through stagnant pseudoplastic liquids. Powder Technol
2004;147:117–25. https://doi.org/10.1016/j.powtec.2004.09.034.
[38] Luo Y, Bern PA, Chambers BD. Flow-rate predictions for cleaning deviated wells.
IADC/SPE Drill. Conf., 1992, p. 367–76.
[39] Ozbayoglu ME, Saasen A, Sorgun M, Svanes K. Effect of pipe rotation on hole
cleaning for water-based drilling fluids in horizontal and deviated wells.
IADC/SPE Asia Pacific Drill Technol Conf 2008:332–42.
[40] Rubiandini R. Equation for Estimating Mud Minimum Rate for Cuttings Transport
in an Inclined-Until-Horizontal Well. SPE/IADC Middle East Drill. Technol.
Conf., 1999.
[41] Xiaofeng S, Kelin W, Tie Y, Yang Z, Shuai S, Shizhu L. Review of Hole Cleaning
in Complex Structural Wells. Open Pet Eng J 2013;6:25–32.
https://doi.org/10.2174/1874834101306010025.
[42] Akhshik S, Behzad M, Rajabi M. CFD-DEM approach to investigate the effect of
drill pipe rotation on cuttings transport behavior. J Pet Sci Technol 2015;127:229–
44. https://doi.org/10.1017/CBO9781107415324.004.
[43] Erge O, van Oort E. Modeling the effects of drillstring eccentricity, pipe rotation
and annular blockage on cuttings transport in deviated wells. J Nat Gas Sci Eng
2020;79:103221. https://doi.org/10.1016/j.jngse.2020.103221.
[44] Nazari T, Hareland G, Azar JJ. Review of cuttings transport in directional well
drilling: Systematic approach. SPE West North Am Reg Meet 2010;1:108–22.
https://doi.org/10.2118/132372-ms.
[45] Naganawa S, Sato R, Ishikawa M. Cuttings-transport simulation combined with
large-scale-flow-loop experimental results and logging-while-drilling data for
hole-cleaning evaluation in directional drilling. SPE Drill Complet 2017;32:194–
207. https://doi.org/10.2118/171740-PA.
[46] Cayeux E, Mesagan T, Tanripada S, Zidan M, Fjelde KK. Real-time evaluation of
hole-cleaning conditions with a transient cuttings-transport model. SPE Drill
Complet 2014;29:5–21. https://doi.org/10.2118/163492-PA.
[47] Watson A. Geothermal Engineering. Springer-Verlag New York; 2016.
[48] Tester J, Anderson B, Batchelor A, Blackwell D, DiPippo R, Drake R, et al. The
Future of Geothermal Energy. Massachusetts Inst Technol 2006;358.
[49] Rybach L. “The Future of Geothermal Energy” and Its Challenges. World
144
Geotherm Congr 2010:1–4.
[50] Polsky Y, Blankenship D, Mansure AJ, Swanson RJ, Capuano LE. Enhanced
geothermal systems well construction technology evaluation. vol. 28. 2009.
https://doi.org/10.1190/1.3255790.
[51] Lu SM. A global review of enhanced geothermal system (EGS). Renew Sustain
Energy Rev 2018;81:2902–21. https://doi.org/10.1016/j.rser.2017.06.097.
[52] Olasolo P, Juárez MC, Morales MP, Damico S, Liarte IA. Enhanced geothermal
systems (EGS): A review. Renew Sustain Energy Rev 2016;56:133–44.
https://doi.org/10.1016/j.rser.2015.11.031.
[53] Pan SY, Gao M, Shah KJ, Zheng J, Pei SL, Chiang PC. Establishment of enhanced
geothermal energy utilization plans: Barriers and strategies. Renew Energy
2019;132:19–32. https://doi.org/10.1016/j.renene.2018.07.126.
[54] Hu Z, Xu T, Feng B, Yuan Y, Li F, Feng G, et al. Thermal and fluid processes in a
closed-loop geothermal system using CO2 as a working fluid. Renew Energy
2020;154:351–67. https://doi.org/10.1016/j.renene.2020.02.096.
[55] Yu H, Li Q, Sun F. Numerical simulation of CO2 circulating in a retrofitted
geothermal well. J Pet Sci Eng 2019;172:217–27.
https://doi.org/10.1016/j.petrol.2018.09.057.
[56] Bu X, Ma W, Li H. Geothermal energy production utilizing abandoned oil and gas
wells. Renew Energy 2012;41:80–5. https://doi.org/10.1016/j.renene.2011.10.009.
[57] Ebrahimi M, Torshizi SEM. Optimization of power generation from a set of low-
temperature abandoned gas wells, using organic Rankine cycle. J Renew Sustain
Energy 2012;4. https://doi.org/10.1063/1.4768812.
[58] Oldenburg CM, Pan L, Muir MP, Eastman AD, Higgins BS. Numerical Simulation
of Critical Factors Controlling Heat Extraction from Geothermal Systems Using a
Closed-Loop Heat Exchange Method. 41st Work. Geotherm. Reserv. Eng., 2016,
p. 1–8.
[59] Blackwell D, Richards M, Frone Z, Batir J, Ruzo A, Dingwall R, et al.
Temperature-at-depth maps for the conterminous U. S. and geothermal resource
estimates. Trans - Geotherm Resour Counc 2011;35 2:1545–50.
[60] Gerner EJ, Holgate FL. OZTemp - Interpreted Temperature at 5km Depth Image.
Geosci Aust Canberra 2010. http://pid.geoscience.gov.au/dataset/ga/71143.
[61] Roberts B j. Non-CONUS U.S. Military Installations and Geothermal Features of
Europe. Natl Renew Energy Lab 2010. nrel.gov.
[62] Shadravan A, Amani M. HPHT 101-What Petroleum Engineers and Geoscientists
Should Know About High Pressure High Temperature Wells Environment. Energy
Sci Technol 2012;4:36–60. https://doi.org/10.3968/j.est.1923847920120402.635.
145
[63] Lunney I, Duriez A, Solbakk T, Skaug MB, Brian T. Harsh environment logging-
while-drilling tools enhance well performance in North Sea HPHT wells. SPE Eur.
Featur. 79th EAGE Conf. Exhib., 2017, p. 32–49. https://doi.org/10.2118/185866-
ms.
[64] Hall JE, Tiliakos N, Worst T, Foster RB, Pappas JM, Long R. HP/HT sensor for
real-time downhole density measurement of wellbore drilling mud. Proc. Annu.
Offshore Technol. Conf., vol. 2, 2017, p. 1119–25. https://doi.org/10.4043/27868-
ms.
[65] Kozlovsky Y. The World’s Deepest Well. Sci Am 1984;251:98–105.
[66] Emmermann R, Lauterjung J. The German Continental Deep Drilling Program
KTB: Overview and Major Results. J Geophys Res Solid Earth 1997;102.
https://doi.org/10.1111/j.1365-246X.1991.tb00816.x.
[67] Gupta VP, Yeap AHP, Fischer KM, Mathis RS, Egan MJ. Expanding the extended
reach envelope at Chayvo Field, Sakhalin Island (Russian). Soc Pet Eng - SPE
Annu Casp Tech Conf Exhib 2014;1:1–23.
[68] Ma T, Chen P, Zhao J. Overview on vertical and directional drilling technologies
for the exploration and exploitation of deep petroleum resources. Geomech
Geophys Geo-Energy Geo-Resources 2016;2:365–95.
https://doi.org/10.1007/s40948-016-0038-y.
[69] Rosneft. The World’s Longest Well Was Drilled in Sakhalin 2017.
[70] Fridleifsson G, Elders WA, Zierenberg RA, Stefánsson A, Fowler APG,
Weisenberger TB, et al. The Iceland Deep Drilling Project 4.5 km deep well,
IDDP-2, in the seawater-recharged Reykjanes geothermal field in SW Iceland has
successfully reached its supercritical target. Sci Drill 2017;23:1–12.
https://doi.org/10.5194/sd-23-1-2017.
[71] Evje S, Fjelde KK. Relaxation schemes for the calculation of two-phase flow in
pipes. Math Comput Model 2002;36:535–67. https://doi.org/10.1016/S0895-
7177(02)00182-6.
[72] Fallah A, Gu Q, Ma Z, Karimi Vajargah A, Chen D, Ashok P, et al. An integrated
thermal and multi-phase flow model for estimating transient temperature dynamics
during drilling operations. SPE/IADC Drill. Conf., vol. 2019- March, 2019.
https://doi.org/10.2118/194083-ms.
[73] Joshi SD. Heat transfer in in-tube flow of non-Newtonian fluids. Digital
Repository@ Iowa State University, http://lib. dr. iastate. edu/, 1978.
[74] Nellis G, Klein S. Heat Transfer. Cambridge University Press; 2009.
[75] Gul S, van Oort E, Mullin C, Ladendorf D. Automated Surface Measurements of
Drilling Fluid Properties: Field Application in the Permian Basin.
SPE/AAPG/SEG Unconv. Resour. Technol. Conf., Denver, Colorado, USA:
146
Unconventional Resources Technology Conference; 2019.
https://doi.org/10.15530/urtec-2019-964.
[76] Gul S, Johnson MD, Karimi Vajargah A, Ma Z, Hoxha BB, van Oort E. A Data
Driven Approach to Predict Frictional Pressure Losses in Polymer-Based Fluids.
SPE/IADC Int. Drill. Conf. Exhib., The Hague, The Netherlands: Society of
Petroleum Engineers; 2019. https://doi.org/10.2118/194132-MS.
[77] Nellis G, Klein S. Heat Transfer. Cambridge University Press; 2009.
[78] Himmelblau DM, Riggs JB. Basic principles and calculations in chemical
engineering. FT Press; 2012.
[79] Moran MJ, Shapiro HN, Boettner DD, Bailey MB. Fundamentals of engineering
thermodynamics. John Wiley & Sons; 2010.
[80] Chirinos JE, Smith JR, Bourgoyne D. A simplified method to estimate peak casing
pressure during MPD well control. SPE Annu. Tech. Conf. Exhib., vol. 6, 2011, p.
4589–600. https://doi.org/10.2118/147496-ms.
[81] Petersen J, Rommetveit R, Tarr BA. Kick with lost circulation simulator, a tool for
design of complex well control situations. SPE - Asia Pacific Oil Gas Conf., 1998,
p. 25–33. https://doi.org/10.2523/49956-ms.
[82] Gu Q, Fallah A, Ambrus A, Ma Z, Chen D, Ashok P, et al. A Switching MPD
Controller for Mitigating Riser Gas Unloading Events in Offshore Drilling. SPE
Drill Complet 2020;Preprint:17. https://doi.org/10.2118/194163-PA.
[83] Karimi Vajargah A, Sullivan G, Johnson M, van Oort E. Transitional and
Turbulent Flow of Drilling Fluids in Pipes : An Experimental Investigation. Am.
Assoc. Drill. Eng., 2017.
[84] Fallah A. Estimation of Temperature-Dependent Parameters using an Integrated
Thermal and Hydraulics Simulator for Drilling Applications. The University of
Texas at Austin, 2018.
[85] Adams J. Ocean currents. Microsoft Encarta 1999;2.
[86] Karimi Vajargah A, van Oort E. Determination of drilling fluid rheology under
downhole conditions by using real-time distributed pressure data. J Nat Gas Sci
Eng 2015;24:400–11. https://doi.org/10.1016/j.jngse.2015.04.004.
[87] Inc. DCS. Drillcool Resources 2015. https://drillcool.com/resources/resources/
(accessed August 8, 2019).
[88] Monteiro EN, Ribeiro PR, Lomba RFT. Study of the PVT properties of gas-
synthetic-drilling-fluid mixtures applied to well control. SPE Drill Complet
2010;25:45–52. https://doi.org/10.2118/116013-PA.
[89] Fallah A, Gu Q, Ma Z, Karimi Vajargah A, Chen D, Ashok P, et al. Temperature
Dynamics and Its Effects on Gas Influx Handling During Managed Pressure
147
Drilling Operations. J Nat Gas Sci Eng 2020.
https://doi.org/10.1016/j.jngse.2020.103614.
[90] Bourgoyne Jr AT, Millheim KK, Chenever ME, Young Jr. FS. Applied Drilling
Engineering. Volume 2. 1986.
[91] Ahmed R, Miska S. Advanced Wellbore Hydraulics. Advaned drilling and well
technology. USA Society of Petroleum Engineers. Ed. Bernt S. Aadnoy. SPE;
2009.
[92] Jalukar LS. Study of Hole Size Effect on Critical and Subcritical Drilling Fluid
Velocities in Cuttings Transport for Inclined Wellbores. The University of Tulsa,
1993.
[93] Bassal AA. The effect of Drillpipe Rotation on Cuttings Transport in Inclined
Wellbores. The University of Tulsa, 1995.
[94] Rehm B, Schubert J, Haghshenas A, Paknejad AS, Hughes J. Managed Pressure
Drilling. Elsevier; 2008.
[95] Sun F, Yao Y, Li G. Literature review on a U-shaped closed loop geothermal
energy development system. Energy Sources, Part A Recover Util Environ Eff
2020;42:2794–806. https://doi.org/10.1080/15567036.2019.1618990.
[96] Sun F, Yao Y, Li G, Li X. Geothermal energy development by circulating CO2 in
a U-shaped closed loop geothermal system. Energy Convers Manag
2018;174:971–82. https://doi.org/10.1016/j.enconman.2018.08.094.
[97] Sun X, Wang Z, Liao Y, Sun B, Gao Y. Geothermal energy production utilizing a
U-shaped well in combination with supercritical CO2 circulation. Appl Therm Eng
2019;151:523–35. https://doi.org/10.1016/j.applthermaleng.2019.02.048.
[98] Song X, Shi Y, Li G, Shen Z, Hu X, Lyu Z, et al. Numerical analysis of the heat
production performance of a closed loop geothermal system. Renew Energy
2018;120:365–78. https://doi.org/10.1016/j.renene.2017.12.065.
[99] Cheng WL, Huang YH, Lu DT, Yin HR. A novel analytical transient heat-
conduction time function for heat transfer in steam injection wells considering the
wellbore heat capacity. Energy 2011;36:4080–8.
https://doi.org/10.1016/j.energy.2011.04.039.
[100] Ramey HJ. Wellbore Heat Transmission. J Pet Technol 1962;14:427–35.
https://doi.org/10.2118/96-pa.
[101] Lemmon EW, McLinden MO, Friend DG. Thermophysical Properties of Fluid
Systems. NIST Chem. WebBook, NIST Stand. Ref. Database Number 69, Eds. P.J.
Linstrom W.G. Mallard, National Institute of Standards and Technology; 2018.
https://doi.org/https://doi.org/10.18434/T4D303.
[102] Zoback MD. Reservoir Geomechanics. 2010.
148
https://doi.org/10.1017/CBO9780511586477.
[103] Yu W, Bao-Lin L, Hai-Yan Z, Chuan-Liang Y, Zhi-Jun L, Zhi-Qiao W.
Thermophysical and mechanical properties of granite and its effects on borehole
stability in high temperature and three-dimensional stress. Sci World J 2014;2014.
https://doi.org/10.1155/2014/650683.
[104] Whittington AG, Hofmeister AM, Nabelek PI. Temperature-dependent thermal
diffusivity of the Earth’s crust and implications for magmatism. Nature
2009;458:319–21. https://doi.org/10.1038/nature07818.
[105] Eppelbaum L, Kutasov I, Pilchin A. Applied Geothermics. Springer Berlin
Heidelberg; 2014. https://doi.org/10.1007/978-3-642-34023-9.
[106] Adams BM, Kuehn TH, Bielicki JM, Randolph JB, Saar MO. On the importance
of the thermosiphon effect in CPG (CO2 plume geothermal) power systems.
Energy 2014;69:409–18. https://doi.org/10.1016/j.energy.2014.03.032.
[107] Foulger GR. The Hengill geothermal area, Iceland: Variation of temperature
gradients deduced from the maximum depth of seismogenesis. J Volcanol
Geotherm Res 1995;65:119–33. https://doi.org/10.1016/0377-0273(94)00088-X.
[108] Gu Q, Fallah AH, Ambrus A, Ma Z, Chen D, Ashok P, et al. A switching
controller for mitigating riser gas unloading hazards in offshore drilling.
SPE/IADC Drill. Conf., 2019. https://doi.org/10.2118/194163-pa.
[109] Gu Q, Fallah A, Gul S, Ashok P, Chen D, van Oort E, et al. A Novel Deepwater
Kick Control Strategy for Handling Riser Gas Unloading with Data-Driven
Parameter Estimation. SPE/IADC Manag. Press. Drill. Underbalanced Oper. Conf.
Exhib., 2020. https://doi.org/10.2118/200522-ms.
[110] Gu Q, Fallah A, Feng T, Bakshi S, Chen D, Ashok P, et al. A novel dilution
control strategy for gas kick handling and riser gas unloading mitigation in
deepwater drilling. J Pet Sci Eng 2020;196:107973.
https://doi.org/10.1016/j.petrol.2020.107973.
[111] Han R, Ashok P, Pryor M, van Oort E. Real-time 3D computer vision shape
analysis of cuttings and cavings. SPE Annu. Tech. Conf. Exhib., 2018, p. 24–6.
https://doi.org/10.2118/191634-ms.
[112] Enayatpour S, van Oort E, Patzek T. Thermal cooling to improve hydraulic
fracturing efficiency and hydrocarbon production in shales. J Nat Gas Sci Eng
2019;62:184–201. https://doi.org/10.1016/j.jngse.2018.12.008.
[113] Enayatpour S, van Oort E, Patzek T. Thermal shale fracturing simulation using the
Cohesive Zone Method (CZM). J Nat Gas Sci Eng 2018;55:476–94.
https://doi.org/10.1016/j.jngse.2018.05.014.
[114] Fallah AH, Gu Q, Saini G, Chen D, Ashok P, van Oort E, et al. Hole cleaning case
studies analyzed with a transient cuttings transport model. SPE Annu. Tech. Conf.
149
Exhib., vol. 2020- Octob, 2020. https://doi.org/10.2118/201461-ms.
[115] Fallah AH, Gu Q, Chen D, Ashok P, van Oort E. Globally scalable geothermal
energy production through managed pressure operation control of deep closed-
loop well systems. Energy Convers Manag 2021;236:114056.
https://doi.org/10.1016/j.enconman.2021.114056.
150