US007 172037B2

(12) United States Patent (10) Patent No.: US 7,172,037 B2

Dashevskiy et al. (45) Date of Patent: Feb. 6, 2007
(54) REAL-TIME DRILLING OPTIMIZATION 6,408,953 B1* 6/2002 Goldman et al. ............. 175/39
BASED ON MWD DYNAMIC
MEASUREMENTS

(75) Inventors: Dmitriy Dashevskiy, Celle (DE); John (Continued)

D. Macpherson, Celle (DE): Vladimir FOREIGN PATENT DOCUMENTS
Dubinsky, Houston, TX (US); Pat
McGinley, Sugar Land, TX (US) DE 1994.1197 9, 1998
(73) Assignee: Baker Hughes Incorporated, Houston,
TX (US
(US) (Continued)
(*) Notice: Subject to any disclaimer, the term of this OTHER PUBLICATIONS
patent is extended or adjusted under 35
U.S.C. 154(b) by 284 days. Lateral Drillstring Vibrations in Extended-Reach Wells, G. Heisig,
SPE, Baker Hughes INTEQ, M. Neubert, SPE, Baker Hughes
(21) Appl. No.: 10/814,918 INTEQ GmbH, paper presented at the 2000 IADC/SPE Drilling
Conference held in New Orleans, LA, Feb. 23-25, 2000.
(22) Filed: Mar. 31, 2004
(Continued)
(65) Prior Publication Data Primary Examiner Frank S. Tsay
US 2004/O256152 A1 Dec. 23, 2004 (74) Attorney, Agent, or Firm—Madan, Mossman & Sriram,
P.C.
Related U.S. Application Data
(60) Provisional application No. 60/459,283, filed on Mar. (57) ABSTRACT
31, 2003.
(51) Int. Cl A drilling control system provides, in one aspect, advisory
EiB 35/16 (2006.01) actions for optimal drilling. Such a system or model utilizes
E2IB 9/08 (2006.015 downhole dynamics data and Surface drilling parameters, to
(52) U.S. Cl 175/25: 175/27 produce drilling models used to provide to a human operator
(58) Field of Classification search- - - - - - - - s 175/57 with recommended drilling parameters for optimized per
formance. In another aspect, the output of the drilling
175/61, 62, 26, 27, 38 control system is directly linked with rig instrumentation
See application file for complete search history. systems so as to provide a closed-loop automated drilling
(56) References Cited control system that optimizes drilling while taking into
account the downhole dynamic behavior and Surface param
U.S. PATENT DOCUMENTS eters. The drilling models can be either static or dynamic. In
4,739,841 A 4, 1988 Das ............................. is one embodiment, the simulation of the drilling process uses
5,947,213 A 9, 1999 Angle et al. .... 175/24 neural networks to estimate Some nonlinear function using
6,012,015 A 9. .............
1/2000 Tubel TO2/6 the examples of input-output relations produced by the
6,021,377 A 2/2000 Dubinsky et al. .............. 7O2/9 drilling process.
6,026,911 A 2/2000 Angle et al. .................. 175/24
6,109,368 A * 8/2000 Goldman et al. ............. 175/39 26 Claims, 11 Drawing Sheets

-12
-> a E. at R(t)
i E. e Adkind-ta) drilling G() Drilling downhole responses
"delayed equipment controls Process Rs)
\-s Computer Supported action surface downhile
ManualCOrt Hook load response Basurement
Advanced
S- 14 Processing Alerts, 'ite iller E() - s |al
A: 46 : Recommendations : | MW
:S: DWhole data ;
:----------------
Gdcs(tm) Rs) telemetry
3. "A" Fly Automated conso corand is o surface
- to surface ... Filly automated data. Ades(n) Syster Ritt)
processing and rig control advice decimated and delayed
M-10 downhole responses

Controls (c) Responsas (R)

W Hook Load Drilling sROP
Rotary Speed Process Downhole WOBRPM
* Mud Flow Rate * Drillstring and BHA
x Mudproperties wbration

Hardware & wellbore characteristics (H) Environment (E)

x Geometry and mechanical parameters of the BHA x Formation lithology
X Characteristics of the drilloit, notor X Mechanical Properties
x Wellbore geometry and well profile of the fortation
 US 7,172,037 B2
Page 2

U.S. PATENT DOCUMENTS Improving Drilling Performance by Applying Advanced Dynamics

Models, M.W. Dykstra, SPE. Hughes Christensen, M. Neubert,
6,772,134 B1 8/2004 Jacubasch SPE, Baker Hughes, INTEQ, J.M. Hanson, SPE, Consultant, M.J.
6,944,547 B2* 9/2005 Womer et al. Meiners, SPE. Hughes Christensen, presented at the SPE/IADC
6,968,909 B2 * 1 1/2005 Aldred et al. ................. 175/26 DrillingConference held in Amsterdam. The Netherlands, Feb.
2002/O120401 A1 8, 2002 Macdonald 27-Mar. 1, 2001.
2002fO24652 A1 9, 2002 Schultz Field Experiences with Computer Controlled Drilling, Brett, J.F.
FOREIGN PATENT DOCUMENTS Warren, T.M., Wait, D.E. (Paper SPE 20107).
Measurement of BHA Vibration Using MWD, Close, D.A., Owens,
EP 0595033 A2 5, 1994 S.C. and Macpherson J.D., SPE/IADC 17273, 1988.
EP O718641 A2 6, 1996 Downhole Diagnosis of Drillingldynamics Provides New Level
EP 1126129 A1 8, 2001 Drilling Process Control to Driller, Heisig, G., Sancho, J.,
EP O881357 5, 2002 Macpherson J.D., SPE 49206, 1998.
GB 2340944. A 3, 2000 An Interactive Drilling Dynamics Simulator for Drilling Optimiza
GB 2352046. A 1, 2001 tion and Training, Dubinsky, V.S. Baecker, D. R., Paper SPE 49205,
JP 634-6448 A 12/1994 1998.
WO WO97/31175 8, 1997 A New Approach to Interpreting Rock Drillability, Bingham, M.G.,
WO WO98, 17894 4f1998 Petroleum Publishing Company, 1965.
WO WOOO,50728 8, 2000 Application of Drilling Performance Data to Overpressure Detec
WO WOO1 6 1140 8, 2001 tion, Jordan, J.R. and Shirley, O.J., 1966, JPT, No. 11.
WO WOO1 6 1140 A1 8, 2001 AGIP Deep Drilling Technology—2, Bellotti P., and Giacca D.,
WO WOO2,38915 A2 5, 2002 OGJ, vol. 76, No. 35, pp. 148.
Drilling Model for Sofi-Formation Bits, Warren T.M., 1981, JPT,
OTHER PUBLICATIONS vol. 33, No. 6, pp. 963.
Real-time Drilling Optimization Based on MWD Dynamic Mea Roller Bit Model with Rock Ductility and Cone Offset, Warren T.M.,
and Oniya E.C., 1987, SPE 16696.
surements-Field Test Results, Dashevskiy D.", Macpherson J.D.'. The Application of a New Drilling Model for Evaluating Formation
Dubinsky V., McGinley, P., Baker Hughes, presented at the and Downhole Drilling Conditions, Jogi P.N., and Zoeller W.A.,
AADE 2003 National Technology Conference “Practical Solutions 1992, SPE 24452.
for Drilling Challengers', held at the Radisson Astrodome Houston,
Texas Apr. 1-3, 2003. * cited by examiner
U.S. Patent Feb. 6, 2007 Sheet 1 of 11 US 7,172,037 B2

=•~•
-
~
U.S. Patent Feb. 6, 2007 US 7,172,037 B2
U.S. Patent Feb. 6, 2007 Sheet 3 of 11

s(ºuH0ds)98. •
d0}}

fiu| |0 S 0 Id
U.S. Patent Feb. 6, 2007 Sheet 4 of 11 US 7,172,037 B2

Cn
r

En E
-n -- 1 FOrmation
R. - R
? - -- 1 Responses
time
A B C D E

COntrols

Formation

Responses
time
U.S. Patent Feb. 6, 2007 Sheet 5 of 11 US 7,172,037 B2

-measured data
-Simulated data
-desired COntrols

Controls: C &
Surface responses. Rs 3.x.
downhole responses: RD &
U.S. Patent Feb. 6, 2007 Sheet 6 of 11 US 7,172,037 B2

COSt COSt s
unction unction
reCeIVing generation

100

Drilling
PrOCeSS

FIG. 6
U.S. Patent US 7,172,037 B2
U.S. Patent Feb. 6, 2007 Sheet 8 of 11 US 7,172,037 B2

-measured data
-Simulated data
-measured data USed
aS future COntrolS

0 delays
s so sao 1 delays

---- 3 delays
- - 5 delays
- - 7 delays

O 1 2 3 4
Prediction horizon (min)
FIG. 9
U.S. Patent Feb. 6, 2007 Sheet 9 of 11 US 7,172,037 B2

pe for distribution of Els error for ROP

- 5 SeC
-- 10 SeC
---- 15 SeC
-- - - - - - - 20 SeC

Prediction horizon (min)

FIG. 10
U.S. Patent US 7,172,037 B2

00Z
U.S. Patent

ROP
 US 7,172,037 B2
1. 2
REAL-TIME DRILLING OPTIMIZATION maintain steady drilling conditions by referencing param
BASED ON MIWD DYNAMIC eters such as WOB, RPM, Delta Standpipe Pressure and
MEASUREMENTS Torque. These systems were capable of Swapping between
the primary controlling parameter as conditions varied.
CROSS-REFERENCE TO RELATED However, they still lacked the important link to definitive
APPLICATIONS downhole dynamic measurements.
The early 1990's also saw the introduction of the first
This application takes priority from U.S. Provisional reliable downhole dynamics measurements. Such measure
Application No. 60/459.283, filed Mar. 31, 2003. ments are described in Close, D. A., Owens, S. C. and
10 Macpherson J. D., “Measurement of BHA Vibration Using
FIELD OF THE INVENTION MWD, SPE/IADC 17273, 1988 and Heisig, G., Sancho, J.,
and Macpherson J. D., “Downhole Diagnosis of Drilling
This invention relates generally to drilling of wellbores Dynamics Provides New Level Drilling Process Control to
and more particularly to real-time drilling based on down Driller”, SPE 49206, 1998, both of which are incorporated
hole dynamic measurements and interactive models that 15 herein by reference for all purposes. Earlier work carried out
allow real-time corrective actions and provide predictive on Surface based measurement systems had proven the need
behavior. for definitive downhole measurements. The cause and effect
of dysfunctional dynamics was now understood. One of the
BACKGROUND OF THE INVENTION last remaining hurdles to a viable drilling control system was
low telemetry rate between the downhole dynamic stools
Real-time drilling optimization that relies primarily on and the Surface systems, which currently are typically 2–10
Surface data has proven ineffective because it does not take bps. Early attempts at using Surface simulators to extrapolate
into account downhole dynamics, such as the behavior of a anticipated downhole dynamics behavior, as discussed in
bottomhole assembly (BHA) within the wellbore. Surface Dubinsky, V. S. Baecker, D. R. “An interactive Drilling
controlled parameters such as weight-on-bit and rotary 25 Dynamics Simulator for Drilling Optimization and Train
speed optimized for maximum penetration rate are of little ing.” Paper SPE 49205, 1998, which is incorporated herein
use if they induce severe downhole vibration that results in by reference for all purposes, in order to provide advice on
costly damage to the BHA. A measurement-while-drilling drilling parameter selection, were somewhat successful, but
(“MWD) dynamics measurement tool is, therefore, a very highlighted the complexity and non linear nature of the
useful component of a closed-loop-drilling control system 30 dynamics problem.
(DCS). For the last couple of decades a variety of mathematical
Early control systems either ignored the downhole models, usually termed drilling models, have been devel
dynamics component or recommended very broad actions, oped to describe the relationship between applied forces and
Such as the practice of avoiding predefined bands of rotary motions (for example, weight-on-bit and rotary speed), and
speed. These early attempts at automated control were 35 the obtained rate of penetration. Both analytical and numeri
further hindered by the state of existent rig instrumentation cal approaches have been Suggested to describe the very
and control systems, and the available computing power. complex three-dimensional movement of the BHA. In many
Several early systems included some form of expert-system, of these empirical models the relationship was in terms of a
typically a rule-based system overlaying a knowledge base. “bulk” formation related parameter, such as the formation
The disadvantage of Such systems was their inability to 40 constants of Bingham's early work. One of these constants
cover all or Substantially all potential Scenarios, and they was later related to formation pore pressure by Jordan and
quickly lost the confidence of the end-user. Shirley and the use of drilling models as pore pressure
In 1990, Brett, Warren and Wait documented the most “predictors' was initiated. Several models followed, such as
serious effort up to that point in time in Brett, J. F. Warren, Wardlaws analytic model Belloti and Gacia's sigma-factor
T. M., Wait, D. E., “Field Experiences with Computer 45 Warren's drilling models, and Jogi's drillability equation, all
Controlled Drilling” (Paper SPE 20107), which is incorpo attempting to describe the relationship between control
rated herein by reference for all purposes. The paper Sug parameters and observed rate-of-penetration with varying
gested that computer based drilling control systems were degrees of complexity. The following herein are incorpo
possible and capable of achieving meaningful results. How rated by reference for all purposes: 12. Bingham, M.G., “A
ever, they stated that achieving an economically viable 50 New Approach to Interpreting Rock Drillability”, Petroleum
system was not a simple task primarily due to the cost of the Publishing Company, 1965; 13. Jordan, J. R and O. J.
improved rig instrumentation and control infrastructure Shirley, 1966, “Application of Drilling Performance Data to
required. It was postulated that this was the main issue Overpressure Detection” JPT, No 11: 14. Wardlaw, H. W. R.,
underlying the failed emergence of a commercial system. It 1972, “Optimization of Rotary Drilling Parameters' PhD
should be pointed out that even in the early 1990s the efforts 55 Thesis, University of Texas; 15. Bellotti P., and Giacca D.
to develop DCS systems still paid little attention to down “AGIP Deep Drilling Technology 2, OGJ, Vol 76, No.35,
hole dynamics components of the control equation, thus pp. 148; 16. Warren T. M., 1981, “Drilling Model for
were limited in their capabilities. Soft-Formation Bits”, JPT, vol 33, no. 6, pp. 963; 17. Warren
The early 1990s saw the introduction of improvements to T.M., and Oniya E. C., 1987, “Roller Bit Model with Rock
rig instrumentation systems that represented a step change in 60 Ductility and Cone Offset, SPE 16696; 18. Jogi P.N., and
the drilling control process. Rig instrumentation networks, Zoeller W. A., 1992, “The Application of a New Drilling
the majority running on some form of Profibus System, now Model for Evaluating Formation and Downhole Drilling
had high-speed access to upwards of 2,500 rig sensors. The Conditions, SPE 24452.
replacement of the old style band brake draw works with new During the past 20 years the high-profile technology
hydraulic based systems allowed for dynamic control of 65 developments within the energy industry have focused pri
WOB, both positively and negatively. New and smarter marily on production, this being driven by the move to
“Automated Drillers' were introduced. Systems that could deepwater and other challenging environments. Develop
 US 7,172,037 B2
3 4
ment of downhole and Surface drilling technology has, to a parameter Such as rate-of-penetration ("optimized param
great degree, been left to the service companies and drilling eter) or hole clearing. The controller is also programmed
contractors. The high spread-costs of deepwater exploration with one or more constraints that can be considered user
has resulted in the drive for improved drilling performance defined norms (e.g., a value that is an operating set-point, a
in harsh and expensive environments, coupled with a range, a minimum, a maximum, etc.) for one or more control
demand for greater reliability from increasingly more com parameters. The control parameters include, but are not
plex downhole MWD tools. limited to, weight-on-bit, RPM of the drill string, RPM of
These goals are not exclusive, but rather are interdepen the drill bit, hook load, drilling fluid flow rate, and drilling
dent, as it is frequently unacceptable to optimize one per fluid properties. During operation, the controller uses on or
formance parameter to the detriment of the other. Hence, the 10 more models for predicting drilling system behavior, the
need for a system that takes a combination of Surface and measured responses and the selected parameters to deter
downhole data inputs, and recommends drilling parameters mine a value for an advice parameter that is predicted to
selected so as to optimize rate-of-penetration (ROP) while at produce the optimized drilling parameter while keeping
the same time allowing the BHA to behave within accept drilling within the specified constraints. In certain embodi
able limits. 15 ments, the controller uses a neural network. The advice
The present invention addresses some of the above-noted parameters include, but are not limited to, drilling fluid flow
deficiencies of prior systems and provides drilling systems rate; drilling fluid density, weight-on-bit, drill bit RPM, and
that utilize downhole drilling dynamics, Surface parameters bottomhole pressure.
and predictive neural network models for controlling drilling Suitable embodiments of the model used by the controller
operations and to predict optimal drilling. include “historical data relating to the characteristics of the
SUMMARY OF THE INVENTION
formation being drilled and the past behavior of the drilling
system. For instance, the model can include data relating
geometry of the BHA, mechanical parameters of the BHA,
This invention provides a control system that in one characteristics of a drill bit carried by the BHA, character
aspect uses a neural network for predictive control for 25 istics of a drilling motor in the BHA, wellbore geometry,
drilling optimization. The system can operate on-line during well profile, lithology of the subterranean formation being
drilling of wellbores. The system acquires surface and drilled, mechanical properties of the subterranean formation
downhole data and generates quantitative advice for drilling being drilled, lithology data obtained of an offset well, and
parameters (optimal, weight-on-bit, rotary speed, etc.) for formation mechanical property data obtained from an offset
the driller or for automated-closed-loop drilling. The system 30 well. In certain embodiments, the controller includes a
may utilize a real-time telemetry link between an MWD sub plurality of model modules, each of which are associated a
and the surface to transfer data or the data may be stored different system response. In addition to determining a
downhole of later use. Data from offset wells can be used response based on measured data, a model module calcu
successfully to describe the characteristics of the formation lates a cost for the response. In one embodiment, the
being drilled and the upcoming formation. The relationship 35 controller normalizes the costs of the several responses in
between these formation parameters and the dynamic mea determining the advice parameter. Also, in several embodi
Surements may be utilized in real-time or investigated off ments wherein real-time drilling data is dynamically
line, once the dynamics information is retrieved at the updated, the controller updates one or more models in
surface. Such a scenario may be likely, when there is real-time using an error calculation between a measured
substantial time-delay in getting MWD information to sur 40 value for a drilling system response and a predicted value for
face. The data can be processed downhole with models the drilling system response.
stored in the MWD and used in real-time, to alter, at least In another embodiment, the controller provides closed
Some of the drilling parameters. loop control for the drilling system wherein the determined
In another aspect, the present invention provides advice advice parameter is used to issue appropriate command
and/or intelligent control for a drilling system for forming a 45 signals to the drilling system.
wellbore in a Subterranean formation. An exemplary drilling Examples of the more important features of the invention
system includes a rig positioned at a surface location and a have been summarized (albeit rather broadly) in order that
drill string conveyed into the wellbore by the rig. The drill the detailed description thereof that follows may be better
string has a bottomhole assembly (BHA) attached at an end understood and in order that the contributions they represent
thereof. A plurality of sensors distributed throughout the 50
to the art may be appreciated. There are, of course, addi
drilling system for measure surface responses and downhole tional features of the invention that will be described here
responses of the drilling system during drilling. Exemplary
Surface responses include oscillations of torque, Surface inafter and which will form the subject of the claims
torque, hook load, oscillations of hook load, RPM of the drill appended hereto.
string, and rate-of-penetration. Exemplary downhole 55
BRIEF DESCRIPTION OF THE DRAWING
responses include drill string vibration, BHA vibration,
weight-on-bit, RPM of the drill bit, drill bit RPM variations,
and torque at the drill bit. In some arrangements, the For detailed understanding of the present invention, ref
measured downhole responses are preprocessed and deci erences should be made to the following detailed description
mated by a downhole tool (e.g., MWD tool or downhole 60 of the preferred embodiment, taken in conjunction with the
processor and transmitted uphole via a Suitable telemetry accompanying drawings, in which like elements have been
system. given like numerals and wherein:
In one embodiment, a controller (or “Advisor”) for con FIG. 1A shows an embodiment of a simplified data flow
trolling the drilling system uses the sensor measurements diagram according to the present invention for use in drilling
(i.e., the Surface and downhole responses) to generate a 65 of wellbores;
value or values for one or more drilling parameters (“advice FIG. 1B shows another embodiment of a data flow
parameter) that, if used, is predicted to optimize a selected diagram according to the present invention.
 US 7,172,037 B2
5 6
FIG. 1C shows exemplary parameters that affect a drilling defined end points (operating norms). Minimum and maxi
process that are considered in developing one embodiment mum acceptable values for WOB, RPM and Torque, and for
of a system of the present invention; various types of vibration (lateral, axial and tosional) are
FIG. 2 graphically illustrates the response of an exem specified. Tolerance of highly undesirable occurrences, such
plary drilling system to changes in selected parameters; as whirl, bit bounce, stick-slip and, to Some degree, torsional
FIG. 3 shows a graphical representation of use of certain oscillation, are set at a number approaching Zero.
available data to predict system responses. In one aspect, this invention aims at obtaining the opti
FIG. 4 shows a block diagram of an exemplary embodi mum drilling parameters (for example weight-on-bit
ment of a drilling control system made in accordance with (WOB), drillbit rotation per minute (RPM), fluid flow rate,
the present invention; 10 fluid density, bottom hole pressure, etc.) to produce the
FIG. 5 shows a simplified block diagram of one embodi optimum rate-of-penetration while drilling. The optimum
ment of a drilling Advisor made in accordance with the rate-of-penetration may be less than the maximum rate-of
present invention; penetration when damaging vibrations occur or due to other
FIG. 6 shows a block diagram for adapting one embodi constraints placed on the system, such as a set MWD
ment of a neural network to current drilling conditions. 15 logging speed.
FIG. 7 graphically illustrates a comparison between actual Once a model has described the relationship between the
and estimated gamma ray measurements; system input and output Sufficiently well, then the model can
FIG. 8 shows the use of measured, simulated, and mea be used to answer certain inverse questions, such as: “What
Sured data used a future controls during modeling; is the weight-on-bit and rotary speed to obtain the optimum
FIG. 9 shows accuracy of prediction for various modeling rate-of penetration?” In other words, these models may be
step sizes; used in a drilling control system whose goal is to optimize
FIG. 10 graphically illustrates accuracy of prediction for the rate-of-penetration. However, cursory inspection reveals
modeling steps of different durations; that a more complete question that may be asked is: “Given
FIG. 11 shows prediction at thirty-six steps ahead of rate a certain size and type of bit, on the end of a certain selected
of penetration by a model using five (5) second intervals; 25 drillstring, at a certain depth, drilling with certain mud
and properties and flow rates in a certain lithology, what is the
FIG. 12 graphically illustrates the improvement in pre weight-on-bit and rotary speed to obtain the optimum rate-of
diction accuracy when look ahead information is used. penetration? Unfortunately this question is so complex,
involving the interaction of so many different components
DESCRIPTION OF THE PREFERRED 30 (only a few of which are listed), that it is difficult to utilize
EMBODIMENT(S) the developed drilling models to obtain an answer. In
addition, the developed drilling models are linear while the
In one aspect, the present invention describes a system drilling process contains non-linearities (the intersection of
that provides advisory actions for optimal drilling. Such a a bed boundary by the drill bit is an example), and the
system is referred to herein as an “Advisor.” The “Advisor 35 achievement of an optimized rate-of penetration may result
system utilizes downhole dynamics data and Surface drilling in destruction of the BHA, because most models do not deal
parameters, to produce drilling models that provide a human with drillstring dynamics.
operator (or "Driller”) with recommended drilling param In certain embodiments, the model used in a control
eters for optimized performance. In another aspect, the system accounts for dynamics of the drillstring. Applying a
present invention provided a system and method wherein the 40 certain set of control parameters results not only in a certain
output of an “Advisor system is directly linked with rig rate-of-penetration, but also in certain motions and forces in
instrumentation systems so as to provide a closed-loop the BHA, which must be measured downhole while drilling.
automated drilling control system (“DCS”), that optimizes As discussed above, there are several possible options for
drilling while taking into account the downhole dynamic a mathematical description of the drilling process as a
behavior and surface parameters. Preferably, the drilling 45 complex system with many influencing parameters. In one
control system has close interaction with a drilling contrac embodiment, this invention treats the drilling process as a
tor and a rig instrumentation provider (e.g., the development dynamic system.
of a “man safe' system with well understood failure behav Dynamic systems can be viewed in two ways: the internal
ioral modes). Also, links are provided to hole cleaning and view or the external view. The internal view attempts to
annular pressure calculations so as to ensure an annulus of 50 describe the internal workings of the system and it originates
the well is not overloaded with cuttings. Thus, embodiments from classical mechanics. A classical problem is discussed
made in accordance with the present invention can, in one in literature is the problem to describe the motion of the
mode, help an operator or driller optimize the performance planets. For this problem, it seemed natural to give a
of a rig and, in another mode, be self-controlling with an complete characterization of the motion of all planets. The
override by the Driller. 55 other view on dynamic systems originated in electrical
Referring to FIG. 1A, there is shown in flow chart form engineering. The prototype problem discussed is to describe
the control and data flow for a drilling control system 10 electronic amplifiers. In Such a case, it was thought natural
made in accordance with the present invention. A rig 12 at to view an amplifier as a device that transforms input
the surface and a bottomhole assembly (BHA) 14 in a well Voltages to output Voltages and to disregard the internal
16 are provided with sensors (not shown) that measure 60 detail of the amplifier. This resulted in the input-output view
selected parameters of interest. These measurements are of systems. Such models are often referred to as input output
transmitted via a suitable telemetry system to the drilling models or “black box” models.
control system 10. In an exemplary deployment, a system In application where there is relatively little real-time
engineer or a Driller or an operator (“operator') inputs or information about the internal state of the whole drilling
dials acceptable vibration levels into the Drilling Control 65 system, it is preferred that a “black box' approach be used
System 10 and requests the system 10 to keep control for modeling of the drilling process although other
parameters within optimal ranges that fall within user approaches may be equally suitable in certain applications.
 US 7,172,037 B2
7 8
Referring to FIGS. 1B and 1C, there are shown in The response of the system usually remains stable when
flowchart form one approach wherein the drilling process controls and environment do not change. Changes in con
can be thought of as one that is affected by the following trols (C) and/or environment (E) tend to disturb the system.
exemplary categories: (i) controls comprising Hook Load, But when the controls and environment stabilize, the system
Rotary Speed, and Mud Flow Rate (drilling parameters 5 response stabilizes as well. Experiments have shown that the
referred to with numeral C(t)); environment, including, for stabilization time is about two minutes. Thus, if At-Ts (i.e.,
example, lithology and mechanical properties of the forma modeling time step is greater than the stabilization time) the
tion, etc. (formation parameters referred to with numeral dynamic behavior of the system cannot be traced. In Such a
E(t); and hardware, which consists of BHA (Bottom Hole 10
case, the drilling process may be considered as composed of
Assembly), drill bit, wellbore geometry, etc. (drill string and a set of “drilling steps” as shown in FIG. 2. Each step is a
BHA parameters referred to with numeral H(t)). transition from one stable state (C, E, R) to another stable
Controls (C) and Environment (E) change continuously state (C, E, R). However, the duration of each of
while drilling. Hardware changes from run to run, but it is these steps might be different.
known and can be considered as a set of constants for 15
In one aspect, it can be assumed that there are only two
particular bit run. In certain applications, environment is reasons why transitions may occur: change in the values of
unknown. In other applications, environment is known the bottomhole pressure controls and/or environment.
approximately and partially from offset wells. Under the In this case R (the new values of the responses) depend
influence of these inputs (C, E, H) the drilling process on: (i) new values of controls (C) and environment (E
); (ii) previous stable state (C, E, R.); and (iii) transition
generates responses, i.e. outputs of the “blackbox”. Some of path or stage (stage BD).
these inputs can be measured at the Surface (Surface In certain instances, the transition state BD may be
responses—Rs), e.g. ROP, Surface torque, oscillations of difficult to formalize (e.g., when the Driller makes the
hook load and drill string RPM, etc., while others are changes, because, even the same Driller may have different
preferably measured downhole (downhole responses—R), 25
ways of changing the control values). In those instances, this
e.g. actual WOB, bit RPM variations, torque at the bit and factor may not be very detrimental because preliminary field
other parameters characterizing drill bit and BHA dynamics. tests showed that, when formation does not change (i.e.
In one embodiment, responses measured downhole are pre E. const), the system response (R) in the stable state
processed and decimated by a multi-channel MWD drilling depends primarily on the control values (C). So, the fol
dynamics tool that reduce the amount of data to be trans 30
lowing assumption can be used as a working hypothesis:
mitted to the surface via a telemetry. In certain embodi considering H being a constant, and that controls C, and
ments, an MWD telemetry system can be used to transmit environment E, adequately define R:
data from the BHA and drillstring to the surface. If an MWD Rn+1=F(C, En+1) (1)
telemetry system is used then the downhole data are sig
nificantly delayed, and thus further decimated. Additionally, 35
As previously mentioned, the dynamic model of the
the downhole BHA may include further processing capabil drilling process applies when the modeling time step is
ity that processes the downhole data and determines advice much less than the system stabilization time. The herein
or actions that need to be taken and also to provide predic used approach to nonlinear system identification is to embed
tions. Such a data processing reduces the downhole data to the measured input-output variables in a higher dimensional
a manageable level for transmission. 40
space built just with current values of controls and responses
In one embodiment, the Drilling Control System may use (C (t), R(t)), and also transforms of C. R (for example their
all available data to generate advice parameters for the numerical derivatives). Other Suitable approaches may also
Driller and acts as a Drillers’ Advisor. In a separate embodi be used. Practically, the behavior of the drilling process can
ment, the Drilling Control System can deliver a command be described by embedding both the inputs and outputs in
directly to the drilling control equipment to provide a Closed 45
the form:
Loop Drilling Control System. In both cases, the DCS
operates as a discrete system, on a time step-by-step basis.
This time step. At (modeling time step), is bounded by a where N is the number of time delays. FIG. 3 illustrates a
minimum value: T, ZXt. This lower boundary (T,) is simple example of a neural net model that uses available
determined by the availability of the “fastest data and the 50 data to predict system response. In FIG. 3, the numeral 31
speed at which the data can be processed at each time-step. identifies measured data for controls C. Surface responses R.
For example, T, may be a short time interval (e.g., five and downhole responses R over time t. The numeral 33
seconds). identifies simulated data over time for C. R. and R, and
Experiments have generally shown that it takes about two numeral 35 identifies desired controls for such parameters.
to three minutes for the drilling process to stabilize. The 55 The simple model of FIG. 3 (with just one delay) may use
magnitude of the stabilization time (Ts) can be used to the current control values of WOB (to) and RPM (t), the
determine the manner in which the drilling process may be current surface response of torque (to), the current response
simulated. If T is significantly smaller than T. and a small of ROP (t), and the future controls of WOB (to-At), and
At can be chosen, then the control system can trace the RPM (to-At) to produce an estimate for the future ROP (to
dynamics of the drilling process, i.e., how the responses 60 At) and torque (tO+At) responses. In other embodiments of
change from one time step to the next. Otherwise, it may be the present invention, more Sophisticated models can use
preferable to consider drilling as a sequence of "drilling more delays, larger sets of controls and responses as well as
steps. Each step being a transition from one stable state to environmental data as inputs.
another stable state. The duration of each step is not neces These embedded models can be faithful to the dynamics
sarily fixed, but is determined by the events when changes 65 of the original system. In particular, deterministic prediction
in controls or information occur. Such a case would be static can be obtained from an embedded model with a sufficient
drilling models. number of delays. Thus, embedding opens the way towards
 US 7,172,037 B2
10
a general solution for extracting “black box” models of the optimal drilling efficiency will be obtained by substantially
observable dynamics of nonlinear systems directly from reducing ROP in view of unwanted vibrations or in view of
input-output time-series data relating to a drilling system. It other relevant parameters.
can solve the fundamental existence problem for a class of During the real-time operation of the Advisor, models can
nonlinear system-identification problems. 5
be adapted using recent real-time drilling data when found
In the above-described embodiments, the simulation of necessary. FIG. 6 shows one manner of Such an adaptation.
the drilling process can estimate some nonlinear function The error 80 between the recent real time data and the
using the examples of input-output relations produced by the predicted values can be used for updating models 84 for the
drilling process. In one embodiment, neural networks can be 10
drilling process 100. This improves accuracy of the local
used for this task due to their known “universal approxima prediction, both time- and State-wise, and increases stability
tion' property. Neural networks with at least a single hidden of the control procedure.
layer have been shown to be able to approximate any Usually, it is not practical to have historical data for all
arbitrary function (with a finite number of discontinuities) if combinations of parameters affecting drilling. Thus, models
there are a sufficient number of basis functions (hidden 15 based on input-output data typically do some interpolation
neurons). By changing the structure of the neural network, and extrapolation.
its capacity and generalization properties can be varied. A controlled field experiment was performed to test the
A model created on the basis of “historical data is above described system and to estimate the accuracy of the
applicable in situations similar to those observed in the data underlying neural network models. This test was carried out
used for the construction of the model. In one embodiment, at the BETA (Baker Hughes Experimental Test Area) facility
drilling performance over the entire range of operational located near Tulsa, Okla. A battery powered MWD drilling
parameters is optimized by using models created with data dynamics tool was used for downhole measurements. That
from more than one well. Referring now to FIG. 4, there is multi-sensor tool acquired and processed a number of
shown one strategy in implementing and using a controller dynamic measurements downhole, and calculated diagnostic
or Advisor 45. The term “controller should be construed in 25 parameters which quantified the severity of the drilling
a generalized sense as a single or plurality of devices vibrations. These diagnostics were then transmitted to the
configured to receive data, process data, output results surface via MWD telemetry and/or stored into the tool
and/or issue appropriate instructions, etc. Data 50 collected memory.
from different wells 52 are merged and stored in a data During the field test, the detailed data stored in the tool
storage device 54 associated with a data server. After a new 30 memory during drilling were dumped to the Surface com
well 64 has been planned and information about the BHA puter on a periodic basis. Information about the formation at
66, drill bit 68, and other components of the drill string is BETA facility was also available from offset wells. A PDC
available, a request is made for the relevant data model. bit used in the test is presented in FIG. 7.
Using this information, models 60 are created or extracted As downhole data became available it was processed to
from the pool of available models. The system may be 35 create models. Although training of the NN model (when
programmed to select the most appropriate model from a
pool of models or it may create an appropriate model from data are prepared and structure of NN is defined) does not
the data stored or provided to the system. Thereafter, one or require human interaction, it can be a time consuming
more of these models are used on the new well 64 for drilling process, especially for big data sets.
optimization. 40
It was decided to use static models, which have fewer
To make the system more robust, generic and easily inputs and hence can be trained much faster. This allowed a
extendable to future MWD tools, certain embodiments of test of the majority of the Advisor Software package and to
the controller or Advisor have a modular structure. An view some “action' in real-time during the test. Further data
example of a modular structure is shown in FIG. 5. Each processing, as well as comprehensive analysis of the
module 100 is associated with some system response and the 45
dynamic models, was carried out after the field test.
Advisor 102 uses sets of selected modules to generate This test was conducted by drilling at various values of
recommendations. Modules 100 comply with a predefined WOB and RPM and through different formations, in order to
external interface, but no constraints are preferably imposed collect a diverse data set. This diverse data set was then used
on module implementation. The modules are preferably for the following offline study. Mud properties, flow rate and
based on Neural Network models, but other types of math 50 BHA/bit were kept constant through the entire testing to
ematical models may also be utilized. minimize the number of factors affecting the drilling pro
CCSS,
Each module 100 takes control parameters as inputs and
produces a cost associated with the predicted value of the During the test, the real-time computed true vertical depth
future response. Costs produced by different modules are (TVD) was used as a reference to determine formation
normalized. This allows comparison of various responses, 55 properties at the corresponding depth from offset well data.
even if they are quite different in their nature (e.g. whirl vs. Then these values together with surface, surface RPM (all
bit bounce). The system 102 can look at various compari averaged on one-minute intervals) were used as inputs to the
Sons and determine the overall impact of these multiple and NN models to estimate ROP and downhole diagnostics.
often divergent responses to determine the overall impact on Computed values of ROP were compared to those actually
the drilling efficiency. The set of responses considered for 60 observed. As FIG. 7 illustrates they are in good agreement.
optimization, and the corresponding cost functions associ Estimation of the formation at the bit may be very useful
ated with them, define the overall optimization strategy. In not only for the DCS but for other applications swell. It is
the present system, parameters relating to the operating cost feasible to evaluate the properties of the formation at the bit
of a rig can be also considered. The weight assigned to Such using dynamic data. For this purpose neural networks were
operating costs can vary from rig to rig. For example, 65 created; they used the current values of WOB, RPM, ROP
offshore rigs cost substantially more for each hour of down and downhole diagnostics as inputs. FIG. 8 illustrates that
time compared to land rigs. The Advisor may determine that Such straight forward attempts to estimate formation prop
 US 7,172,037 B2
11 12
erties did not yield very good results. A more complex simulating the drilling process were created. FIG. 12 shows
approach will be desirable to design NN predictions for such actual ROP against predicted ROP.
a purpose. During the simulation (prediction three minutes ahead in
Testing of dynamic models was performed offline using this example) actual controls measured during the field test
data collected during the field test. Various parameters that were used as future inputs. Actual responses were used to
affect the creation of a NN model and influence its perfor initialize simulation of drilling dynamics. No actually mea
mance (i.e., how well it simulates the dynamic system) were Sured responses were used when simulation had started. The
evaluated in these tests. The testing included an assessment dynamic model, tested in Such a way, cannot accommodate
of the particular inputs used for NN training, the number of for formation changes which happen within three minutes of
neurons utilized in NN, duration of the modeling step, and 10 simulation. Nevertheless, the model showed good results
SO. O. when formation did not change Substantially.
If information about the formation to be drilled is avail
For each test, 60% of the available data were used for able, then it may be used to a great benefit in dynamic
building a model. Each model was trained to predict certain models. Another model of the drilling process which utilizes
responses one time-step ahead. Trained models were then 15 look-ahead formation information to make predictions was
tested on the remaining 40% of the data. A set of models was created using data from an offset-well. FIG. 13 shows the
used to simulate the future responses several time-steps measured and simulated ROP for the part of the test that
ahead. Controls that were actually observed during the field drilled through a section with fast formation changes.
test were used as future controls as shown in FIG. 9.
Clearly, models using formation data as inputs perform
To evaluate the accuracy of Such a multi-step prediction, better in this complex situation.
the computed values of the responses were compared to the In Summary, the structure of the drilling process has been
actual responses measured the same number of steps ahead, studied to create a design of a “Drilling Advisor that
and a percentage full scale (% FS) error was computed. It provides recommendations regarding which drilling controls
was found that errors computed during each test have a to adjust, and when to adjust such controls. Neural network
distribution which is approximated by the following func 25 models, along with an optimization strategy, were designed
tion: to fit this concept and implemented and tested.
For the model development a pseudo-statistical approach
was employed as an alternative to traditional analytical and
f(x)
(X) 1
2,3
exp(- fe ) numerical approaches. This approach is based on long-term
30 accumulation of practical field knowledge and utilization of
this knowledge for overall improvement of the model and
Value of was computed in each test to produce the best implementation of self-learning and self-adjusting capabili
fit of function (3) to the test error distribution. This “effec ties during drilling. Neural network models can predict
tive' prediction error () allowed a consistent comparison of development of the drilling process accurately enough when
the accuracy of different models investigated in different used on wells drilled through similar lithology with the same
tests and was used to determine optimal values of param BHA and bit. Better accuracy may be achieved, especially
eters that affected the creation of the NN model and influ for long term prediction, when information about the for
ence its performance. mation along the well path is available (for example, from
One parameter that was evaluated is the amount of delays offset wells).
at the neural network input. Although feed forward neural
40 The benefits of a closed loop Drilling Control System are
networks are essentially static, their usage may be extended many, and touch several aspects of the drilling and evalua
to solve dynamic problems by utilizing delay lines. In other tion process. The benefits Relating to Performance Drilling
words by using data from a number of previous time steps. utilizing DCS include Improved ROP. longer bit runs, more
FIG. 10 shows how the accuracy of models that use the same sections drilled in a single run, in gauge hole (Less forma
45 tion drilled), reduced downhole vibration, less wasted
inputs depends on the number of delays. Duration of the energy downhole, less trips due to MWD failure, reduced
time step in these tests was five seconds. BHA failure, steady state drilling, consistent start up after
Prediction error grows with an increase in the prediction connections. The benefits relating to formation evaluation
horizon. However, as FIG. 10 illustrates, a larger number of measurements include: improved quality of measurement, in
time delays improves accuracy. The same behavior was 50 gauge hole, reduced time between drilling and measurement,
observed for models that use different sets of inputs and for less vibration effects on measurements, improved MWD
different durations of the modeling step. More time delays data transmission, less noise due to vibration.
mean more inputs into the NN, resulting in a larger problem The foregoing description is directed to particular
to be solved to train the model. This in turn increases time
to train the NN model. embodiments of the present invention for the purpose of
55 illustration and explanation. It will be apparent, however, to
Another example of a parameter that influences the per one skilled in the art that many modifications and changes to
formance of the dynamic neural network models is the the embodiment set forth above are possible without depart
duration of the time step. The minimum duration of the time ing from the scope and the spirit of the invention. It is
step feasible for the particular data acquired during the field intended that the following claims be interpreted to embrace
test was five seconds. For longer intervals, the value of each 60 all Such modifications and changes.
mnemonic was computed by averaging the available data
over the time step. FIG. 11 shows accuracy of prediction for Nomenclature
modeling steps of different durations. It is observed that BHA-bottomhole assembly
although the models operating on shorter time steps would C, control parameters at n-th time step
require more steps to estimate value of responses for the 65 DCS-drilling control system
same time horizon, they produce better results. Based on E, environment properties at n-th time step
optimal values of these and other parameters, NN models MWD-measurement while drilling
 US 7,172,037 B2
13 14
NN=neural network error calculation between a measured value for a drilling
ROP-drilling rate of penetration system response and a predicted value for the drilling system
RPM-rotations per minute response.
R, responses at n-th time step 10. The system according to claim (1) wherein the
Rs Surface measured responses selected drilling response includes a measured downhole
R, downhole measured responses response that is preprocessed and decimated by a downhole
TVD=true vertical depth tool; and further comprising a telemetry system for trans
WOB-weight on bit mitting the decimated and preprocessed measured downhole
% FS percent of full scale error response to the controller.
What is claimed is: 10 11. The system according to clam (1) wherein the con
1. A system for forming a wellbore in a Subterranean troller utilizes hole cleaning and annular pressure calcula
formation, comprising: tions to determine whether an annulus of the wellbore is
(a) a drilling system including a rig positioned at a surface overloaded with cuttings formed during drilling.
location, a drill string conveyed into the wellbore by the 12. The system according to claim (1) wherein the con
rig, the drill string having a bottomhole assembly 15 troller provides closed-loop control for the drilling system
(BHA) attached at an end thereof, and a plurality of wherein the determined advice parameter is used to issue
sensors associated with the drilling system for measur appropriate command signals to the drilling system.
ing Surface responses and downhole responses of the 13. A system for forming a wellbore in a subterranean
drilling system during drilling; and formation, comprising:
(b) a controller operatively coupled to the drilling system (a) a drilling system including a rig positioned at a surface
and including at least one model for predicting behav location, a drill string conveyed into the wellbore by the
ior of the drilling system, the controller utilizing the at rig, the drill String having a bottomhole assembly
least one model, the measured surface and downhole (BHA) attached at an end thereof, and a plurality of
responses and at least one selected control parameter to sensors associated with the drilling system for measur
predict behavior of the drilling system and to determine 25 ing Surface responses and downhole responses of the
at least one advice parameter that produces at least one drilling system during drilling; and
Selected optimized drilling parameter while satisfying (b) a controller operatively coupled to the drilling system
at least one selected constraint. and including at least one model for predicting behav
2. The system according to claim (1) wherein the at least ior of the drilling system, the controller utilizing the at
one selected control parameter is one of (I) weight-on-bit, 30 least one model, the measured surface and downhole
(ii) RPM of the drill string, (iii) RPM of the drill bit; (iv) responses and at least one selected control parameter to
hook load, (v) drilling fluid flow rate, and (vi) drilling fluid predict behavior of the drilling system and to determine
properties. at least one advice parameter that produces at least one
3. The system according to claim (1) wherein the surface Selected optimized drilling parameter while satisfying
responses are one of (i) Surface torque, (ii) oscillations of 35 at less on selected constraint; wherein the controller
hook load, (iii) and rate-of-penetration, and (iv) oscillation includes a neural network.
of torque. 14. A method for forming a wellbore in a subterranean
4. The system according to claim (1) wherein the down formation, comprising:
hole responses are one of (i) drill string vibration, (ii) BHA (a) providing a drilling system including a rig positioned
vibration, (iii) weight-on-bit, (iv) RPM of the drill bit, (v) 40 at a surface location, a drill string conveyed into the
drill bit RPM variations, and (vi) torque at the drill bit. wellbore by the rig, the drill string having a bottomhole
5. The system according to claim (1) wherein the at least assembly (BHA) attached at an end thereof,
one advice parameter is one of (i) drilling fluid flow rate; (ii) (b) measuring Surface responses and downhole responses
drilling fluid density, (iii) weight-on-bit, (iv) drill bit RPM, of the drilling system during drilling using a plurality of
and (v) bottombole pressure. 45 sensors; and
6. The system according to claim (1) wherein the at least (c) determining at least one advice parameter that pro
one selected optimized drilling parameter is one of: (i) duces at least one selected optimized drilling parameter
rate-of-penetration, (ii) hole cleaning, and (iii) annular pres while satisfying at least one selected constraint using a
SUC. controller, the controller making the determination
7. The system according to claim (1) wherein the at least 50 using at least one model for predicting behavior of the
one model utilizes data relating to one of: (i) geometry of the drilling system, at least one selected control parameter,
BHA, (ii) mechanical parameters of the BHA, (iii) charac and the measured surface and downhole responses.
teristics of a drill bit carried by the BHA, (iv) characteristics 15. The method according to claim (14) wherein the at
of a drilling motor in the BHA, (v) wellbore geometry, (vi) least one selected control parameter is one of: (i) weight
well profile; (vii) lithology of the subterranean formation 55 on-bit, (ii) RPM of the drill string, (iii) RPM of the drill bit:
being drilled; (viii) mechanical properties of the subterra (iv) hook load, (v) drilling fluid flow rate, and (vi) drilling
nean formation being drilled; (iv) lithology data obtained of fluid properties.
an offset well; and (viii) formation mechanical property data 16. The method according to claim (14) wherein the
obtained from an offset well. Surface responses are one of (i) Surface torque, (ii) oscilla
8. The system according to claim (1) wherein the con 60 tions of hook load, (iii) and rate-of-penetration, and (iv)
troller includes a plurality of model modules, each the model oscillation of torque.
module producing a predicted value of a future response and 17. The method according to claim (14) wherein the
cost associated with the future response, the controller downhole responses are one of (i) drill string vibration, (ii)
utilizing the plurality of model modules to evaluate drilling BHA vibration, (iii) weight-on-bit, (iv) RPM of the drill bit,
efficiency. 65 (v) drill bit RPM variations, and (vi) torque at the drill bit.
9. The system according to claim (1) wherein the con 18. The method according to claim (14) wherein the at
troller updates the at least one model in real-time using an least one advice parameter is one of (i) drilling fluid flow
 US 7,172,037 B2
15 16
rate; (ii) drilling fluid density, (iii) weight-on-bit, (iv) drill bit response that is preprocessed and decimated by a downhole
RPM, and (v) bottomhole pressure. tool; and further transmitting the decimated and prepro
19. The method according to claim (14) wherein the at cessed measured downhole response to the controller with a
least one selected optimized drilling parameter is one of: (i) telemetry system.
rate-of-penetration, (ii) hole cleaning, and (iii) annular pres 24. The method according to claim (14) wherein the
SUC. controller utilizes hole cleaning and annular pressure calcu
20. The method according to claim (14) wherein the lations to determine whether an annulus of the wellbore is
controller is provided with at least one model used to overloaded with cuttings formed during drilling.
determine the advice parameter, the at least one model 25. The method according to claim (14) wherein the
utilizing data relating to one of: (i) geometry of the BHA, (ii 10 controller provides closed-loop control for the drilling sys
mechanical parameters of trio BHA, (iii) characteristics of a tem, wherein the determined advice parameter is used to
drill bit carried by the BHA, (iv) characteristics of a drilling issue appropriate command signals to the drilling system.
motor in the BHA, (v) wellbore geometry, (vi) well profile; 26. A method for forming a wellbore in a subterranean
(vii) lithology of the subterranean formation being drilled; formation, comprising:
(viii) mechanical properties of the Subterranean formation 15 (a) providing a drilling system including a rig positioned
being drilled; (iv) lithology data obtained of an offset well; at a surface location, a drill string conveyed into the
and (viii) formation mechanical property data obtained from wellbore by the rig, the drill string having a bottomhole
an offset well. assembly (BHA) attached at an end thereof,
21. The method according to claim (14) wherein the (b) measuring Surface responses and downhole responses
controller includes a plurality of model modules, each model of the drilling system during drilling using a plurality of
module producing a predicted value of a future response and sensors; and
cost associated with the future response, the controller (c) determining at least one advice parameter that pro
utilizing the plurality of model modules to evaluate drilling duces at least one selected optimized drilling parameter
efficiency. while satisfying at least one selected constraint using a
22. The method according to claim (14) wherein the 25 controller, the controller making the determination
controller updates the at least one model in real-time using using at least one model for predicting behavior of the
an error calculation between a measured value for a drilling drilling system, at least one selected control parameter,
system response and a predicted value for the drilling system and the measured Surface and downhole responses,
response. wherein the controller includes a neural network.
23. The method according to claim (14) wherein the 30
selected drilling response includes a measured downhole