Karlene Hoo, PhD Jason Dykstra, PhDChemical Engineering Professor Chief Technical Advisor

Montana State University Halliburton Corporate Research

Open-Ended Control Challenges in the Oil

Service Industry

Global Megatrends and Drivers

Source: Population Reference Bureau

Population growth continues

Especially in developing countries

Global Megatrends and Drivers

Energy Usage For Next 2 Decades

Source: U.S. DOE/EIA International Energy Outlook 2011

July 4, 2016 Estimate of Oil Reserves

Rystad Energy is an independent research group in the oil and gas industry

Unconventional Resources

Heavy OilShale

World Energy Trends

Harder to Find

Harder to Access and Produce

In Smaller Accumulations

More Costly

Produced with Fewer Experienced Resources

The next trillion barrels will be…

Exploration ProductionDevelopment

Technology Challenges

Higher Pressure/Temperature1

Knowledge, Depth, Accessibility, Power 2

Hostile Downhole Environments3

Telemetry, Communication Systems4

Automation and Control Systems5

Well Life Cycle

Exploration Well Construction Completions Production Abandonment

Drilling : The problems are getting more challenging

and the solutions are getting more valuable

• Drilling a well faster and cheaper

• With a high quality wellbore

The Challenge in Building a Drilling Control System

Accelerometers: ax, ay, az

Gyroscopes: aθ, aφ, aψ

Strain gauges: σx, σy, σz

• Mud Pulse communication, travels at the

speed of sound in drilling mud at 2-30

bits/sec

• Large sensor array and the surface and

the wellbore end in the bottom hole

assembly, but nothing in-between

• Dynamic modeling is difficult, in particular

the rock/bit interaction which is the source

of the detrimental vibration

• Many different systems to consider, and

many different environments

Formation sensing: Gamma ray,

Resistivity, Acoustic calipers and

rock mechanics, Neutron, Nuclear

magnetic resonance, and others

• High Temperature High Pressure (HPHT) – Harsh

drilling environment

• Long time delay for mud pulse data transfer / high

cost for other methods

• High cost for well construction and increase in cost of

poor quality

• Drill pipe vibration – disturbance to system,

accelerate the fatigue of material, hard to measure is

deep wells

• Robust control hardware and software under HPHT

• Automated well health monitoring and fault detection

under limited data access with large time delay

• Distributed control to overcome limited communication

and large time delay

Challenges:

Opportunities:

Challenge of Deep water drilling

Directional Drilling Systems

Point the Bit directional control Bent Sub

Formation Sensing Downhole Motor

Directional Drilling Systems

WHIRL

Stick-Slip

BHA

model

Path optimization

path uncertainty

Path

BHA

control

predicted force and position

command

sensor

Rig

control

Drilling

Optimizer

Update

Override

Drilling Optimization

Dynamics

ROP

Bit Wear

Full Drilling Optimization

Other Constraints:

1. WOB < max

2. RPM < max

3. No bit bounce

4. No bit whirl

5. Minimal Balling

6. Bit Temp. < Temp. max

ROP Model

Where

Wear Model

ROP Model Wear Model

ROP Model Wear Model

maxwearwear

Bit model from wear estimator

Wear model updated from actual performance

Available Bit types

min cost

Tripping points becomes part of optimization

Changing tripping points change acceptable wear rates and cost

BIT

TARGET

RATE

RPM

WOB

Optimization produces command vector as a function of time. This includes tripping points and bit types.

Includes vibrational map in operational space

1896

1930

1947

1949

Hydraulic

FracturingUS patent on “matrix acidizing”

Dow Chemical Co discovers that

downhole fluid pressure can crack

and deform rock formation

First experimental well – Hugoton

Gas Field, Kansas

First commercial hydraulic fracture

Other application areas:• Tunnel and dam construction

• Water well development

• Environmental applications

Acid dissolves part of the fracture face to create

short fractures that are highly conductive channels

International Shale PlaysEurope Australia

AsiaMiddle East

Oil and Gas Production

• About 35 to 40% of all currently drilled wells are hydraulically fractured

• About 25 to 30% of total U.S. oil reserves have been made economically producible by the process.

• Tight gas sands and shale reserves

– Both oil and gas

– 503 to 1500 TCF

– Low permeability

– Requires fracturing

USA Shale Plays

Custom designedPurpose built

Transport/Store/Deliver

Hydraulic Fracturing Equipment

High-power pumps

Blender Control room

Data bases

Hydraulic Fracturing Process

• Plug (special designed shaped charges): placed at the desired stage in the well bore

• Perforating gun is lowered into the casing

• Electrical current sets off a charge that shoots small holes through the casing and cement.

• As materials are pumped downhole, fractures are created

Sliding sleeve:multi-stage fracture

efficient

MHF:

wing length 1000 ft

width at wellbore 0.5 in

Stage 1: Pad

• Clean fluid or pad (usually

water) is pumped into the

well at a rate greater that

the fluid loss rate to the

formation

– Pad provides sufficient

width for the proppant to

enter the fissures

– High pressure gradient:

formation breaks and early

fracture growth expose new

formation to the injected

fluid

Stage 2: Proppant

• Solid particles are pumped as a slurry or suspension to prop open the fracture once pumping ceases

• Cheap proppant: sand

• Fluid: water-based polymer solution (guar, hydroxypropyl guar), aqueous foams, gelled hydrocarbons, …

• Additives: emulsifiers, anti-foams, stabilizers, emulsifiers, pH control, surfactants, buffers, bacteria control, …

Multi-stage Fracturing

• More flexibility

– Manage production

– Complete the well

• Improve well performance

• Deliver more effective completions

Shut-in

Fracture walls close

on propping agent

Flow-back of proppant

to the wellbore

Clean-up of fluid

system to allow oil/gas

to flow

Pressure curves in hydraulic fracturing

Fracture Treatment Design

• Mini-fracture data

– In-situ stress profile

– Formation properties

– Fluid-loss characteristics

• Reservoir model & fracture propagation model

– Optimum economic benefit

Fracture Length

Treatment Volume

$ Cost

$ Revenue

Less $ Cost

Reservoir Model

Fra

ctu

re

Pro

pag

ati

on

Mo

del

Challenges

• Layers of rock providing dissimilar confining stresses

• Changes in magnitude and/or orientation of the in situ confining stresses

• Effects of shear & temperature on fluid rheology

• Transport of suspended proppant particles into the fracture

• Fracture recession and closure

• Rapid geometric changes in one region as fractures extend into other lower stress zones

• Leak-off of fracturing fluid from the fracture to the surrounding rock

More Challenges

Fluid system

• Non-damaging to the rock

• Able to transport proppant efficiently

• High flow capacity

• Long term fluid loss

– Viscosity/temperature

– Pressure differential

– Rock permeability, porosity

• Effective viscosity: controls the internal pressure

Proppant

• Particle size distribution

• Roundness & sphericity

• Acid solubility

• Turbidity

• Crush resistance

• Less dense than water

• Stronger than diamond

• Cheaper than dirt

• Readily available

Uncertainty

• Current Practice• Fixed fracturing plan• “Blind” operation

• Emerging technologies• Advanced instrumentation

(fiber optics, microseismic monitoring

• Real-time automation system• Measurement filtering• State/parameter estimation• Optimize fracturing plan• Automate treatment adjustment• Data analytics for job learning

Fiber optic

measurements

DAS, DTS

Microseismic

monitoring

Fracturing

plan

Decision

making

"Open-loop” Approach

• Well treatment schedule

– Planned using expert advice

– Physics-based models (computing intensive)

“Close-the-loop” Approach

Hydraulic Fracture Automation

• Dynamic modeling of surface equipment and fracturing processes with closed-loop correction

• Incorporate microseismic and fiber optic technologies

• Optimal decision making

• Self learning from previous data

Controller

Fracture and

sand model

Fiber optic

measurements

Microseismic

monitoring

Surface

equipment

model

DatabaseLearning

system

Fracture Model

• Aim

– Estimate the fracture state over time

– Predict final propped fracture conditions

• Components

– Fracture geometry propagation

– Fluid flow and leak-off rates

– Suspended proppant transport

– Bank formation due to proppant settling

– Shut-in process

Reservoir stimulation, 18.

Chichester: Wiley, 2000.

Fundamental Principles

• Continuity of mass

• Momentum conservation

• Linear elastic fracture mechanics

(LEFM)

Numerical Challenges

• Issue 1: Nonlinearity near fracture tip

– Solution: Smaller elements in this region

• Issue 2: Fracture length increases with time

– Solution: Moving mesh

Fracture Closure

Control of the Fracturing Process

Control of the Fracturing Process

State Estimation

Quadratic “Adaptive” Dynamic

Matrix Control

Subject to: physical limits (states, actuators)linear step response model

Control Framework

10L/s fluid loss, 300 s from start of pad

Fracture Length Fracture Width

Proppant bank height

Be

fore

Sh

ut-

in

Effective frac vol

Ref: 6.52

W/O control: 2.96

Control: 6.20

Microseismic Interpretation

Observation Well

Estimation of Fracture Geometry & Proppant

Distribution

Model Selection & Control 𝑝(𝑀2, 𝑡)𝑝(𝑀1, 𝑡)

TECHNOLOGIES APPLIEDMachine Learning State Estimation Expert Systems Model-Predictive Control Robust Optimization

Acknowledgements

• Graduate Students

Qiuying Gu Vikram Shabde

Daguang Zheng Alejandro Gonzalez-Flores

Yingying Chen Eric Vasbinder

Zhenhua Tian Stanislav Emets

• Texas Tech University

– Uzi Mann, Ph.D.

– Shameem Siddiqui, Ph.D.

– PCOC members

• Halliburton

Thank You