© 2006 Gregory Finn 1 Advanced Oilfield Operations with Remote Visualization and Control ---------- Moving from PDAs to Intelligent Cooperative Assistants

© 2006 Gregory Finn1

Advanced Oilfield Operationswith Remote Visualization and Control

----------

Moving from PDAs to Intelligent Cooperative Assistants

4 April 2006

Gregory Finn

PTE 589


Lecture Plan

• Background & motivation

• Wireless sensors– overview & potential applications

• Wearable computers– overview & potential applications


Sensor material taken from:– John Heidemann & Wei Ye [USC/ISI]

• “An Overview of Embedded Sensor Networks”,ISI TR-2004-594, Heidemann and Govindan


Drives Toward Smarter Facilities

• Remote control …– hurricanes, unsafe activities

• Enhance capabilities– visualization, agility

• Increase productivity– greater production efficiency

– reduced downtime

• Improve safety


Smart Area Characteristics• Communication rich

– Pervasive networking– Sensors, devices, processes & people in the net

• Compute rich– Computers scattered throughout the space

• Autonomous monitoring

• Visualization rich– Streams of real-time sensor, device & process data

• Control rich– Remote access to devices, processes & people


Smart Areas & Visualization

Smart areas generate lots of data …– Who can watch it all?

– Autonomous monitoring is a practical necessary

Monitoring is visualization …– Uses sensors + network + computer to ‘see’

• structure and device health

• staff & objects

• pipeline & pump activity

• dangerous areas

• dangerous actions


Staff as Part of Smart Area

Smart area is networked …– Devices, sensors & processes communicate via networks

– But people do not ‘network’

Solution: Make staff part of the smart area– Provide staff with personal wireless hosts

– Host act as eyes and ears in the area

– Host autonomously monitors smart area


Why Do That?

Several reasons …– improve productivity

• bring data where it’s needed (manuals/procedures/info on demand)video via display/camera

– enforce/aid safe practices• know what sensors see

• know what to do & not to do

• know where to be & not to be

– improve oversight• know where staff are

• improve dispatch/team forming

• prevent unauthorized operation


Why Possible Now?

Moore’s Law (Gordon Moore – Intel 1965)

– roughly true, should continue ~20 years

– produces exponential rate of change

– #transistors/chip doubles in ~two years

Computers became exponentially …– smaller

– faster

– less expensive

– less power hungry


PDP-10 mainframe Dell Precision 670

Space 60 sq meters 0.25 sq meter

Power 40,000 watts 650 watts

Weight 1,800 kg 19 kg

Memory 1 MB 16,000 MB

Speed 1 Mips 4,000 Mips

Disk 80 MB 1,200,000 MB

Cost (2005 $) $2,000,000 $15,000

1970 vs 2005


Portables & Sensors Today


Example: Telos-B Mote

Characteristics …

• Powered via USB or battery

• Internal antenna

• Controllable xmit power

Flexibility …

• Sensors – Light/IR/Humidity/Temp

• CPU + TinyOS + 48KB Flash + 10KB RAM


Future Sensor Platforms?Smart dust (Kris Pister, UC Berkeley)• nodes smaller than 1mm3

• prices less than $0.05/each

Nok

ia

super cell-phonesor wearable computers

mote-size and price,but 32-bit CPU power

or


Why Else Is It Possible Now?

[courtesy of UCSD’s caida.org]

1969

1980

2000

… the Internet


Back in the Old Days...

1920s telephony:circuits---a physical wire from one end to the other

wire

the “router”(Aunt Mable)


Multiplexing: Splitting a Shared Channel

Frequency Division Multiplexing

Code Division Multiplexinga a a a a a a a a a a

Time Division Multiplexing


Logical View of the Telephone Network

Fixed size pipe from source to destination perfect for voice reliable conversations (QoS) provisioning, good engineering dumb & cheap end points, smart network evolved for 100 years (analog to digital)


Packet Switching (Internet)

Differences: packets as low-level component multiple kinds of traffic smart edges, (dumb network)

But: guarantees are much harder end-points are more expensive


Characteristics of the Internet

• Packet switched

• Freely available standards (IETF)

• End-to-end– intelligence and control in the end-points (dumb middle)

– critical to allowing deployment of new services

• Distributed (no central point of control)

• But security becomes harder


Commercial Activity Today

Development of …• networks (wireless and wired)

• low-power CPUs

• sensors

• applications

– centralized or stand-alone

• smart devices

– wireless monitor/control

• intrinsically safe hardware

Stage is set, however …smart area development is still research.


Missing: Software Infrastructure

Domain representations– facility & process models

Spatial directories– people, sensors, devices and objects

Standards & protocols– data interchange and naming

Security

Applications– processes and procedures


Still … We Can Assume

In a few years …… pervasive communication & computation

• wireless networking– 802.11, UWB, sensors

• pocket computers– 1 GHz, 1600x1200 head-mounted displays

– ‘spatial awareness’

• ‘smart’ areas– sensors, staff, devices, processes

in continuous contact


Lecture Plan





Why Sensors?

To know what is happening …

To visualize what is happening …– Measurement: temp., pressure, flow rate, mixture

– Safety: monitor hazards … H2S, forbidden states

– Structure: stress/corrosion in downhole & surface equipment, pipelines, refineries

– Reservoir: geology, current status of reserves, etc.

– Security: monitor position and intrusion


Sensor Challenges

Cost– Use thousands of sensors of many types

• RFID … passive ~ $0.10 … active ~ $5

• Motes ~ $50

Power consumption– solar + battery may be expensive or inappropriate

– battery life needed > 1 year

Networking– provide robust data interchange

– allow low power consumption


Why Network?

Networking:Ability of computers to exchange data …

Communicate what is known elsewhere …– distance education

– remote control

– share information and files• distributed management

– enable autonomous operation


ChallengesSecurity

– still largely ignored

Power consumption– given distributed computers & sensors with limited power

it’s very important

Bandwidth (speed)

Interaction– hardware / protocols: work pretty well today– software & cooperation

• danger of application Balkanization• no standards or competing standards• need agreements …ex: POSC and XML for oilfield data


Why Sensor Networking? Want to ‘see’ everywhere ‘visualize’ everything

– wellhead, surface facilities, control rooms– enable autonomous monitoring

Want to combine data– different information from different places reveals things– ex: ability to see bigger picture can make a big difference– reveal previously unobserved phenomena

Decrease cost– better information, more precise control

Increase safety– prevent dangerous actions– detect dangerous situations


ResultLots of computers interacting within the world

– physically distributed, sensing, different perspectives

Lots of computers interacting within the world– enough that they’re near what’s sensed, 100s-1000s– enough that some can be off and overall system still runs

Lots of computers interacting within the world– intelligent: able to decide what’s important, collaborate

Lots of computers interacting within the world– sensing, responding, acting– make the area smart


What’s New About Sensor Nets?

Many devices => treat devices as interchangeable– generic vs. dedicated to specific task– benefits: trade density for robustness, longevity, accuracy

Small wireless devices => resource constraints– limited energy, low bandwidth, higher latency– benefits: low price means sensors can be everywhere

challenges spur new technical approaches


Current Sensor Nets: SCADA Systems

SCADA: Supervisory Control and Data Acquisition– remote control of equipment– since 1980s

General focus:– dumb instruments (vs. being able to compute in field)– often custom networks– data sent to central computer or database

Very important today!– remote control and monitoring


Comparing SCADA and Sensor NetsSCADA vs sensor networks

mainframes vs PCs(expensive, centralized, inflexible) vs cheap, distributed, versatile

Where is the data?– SCADA typically moves raw data to a central site– sensor nets focus on keeping and processing raw data at smart sensor

Where is the control?– SCADA typically leaves control decisions to central site– sensor nets focus on shifting control to smart edges

Who defines them?– SCADA systems are often proprietary protocols– sensor networks are today typically research protocols

Probably both areas will converge.


Military: vehicle tracking(ISI at DARPA SensIT SITEX)

Government: vehicle monitoring(USC/SPPD & ISI)

Scientific: micro-habitat monitoring(UCLA/CENS at James Reserve)

Industry: equipment monitoring and control

Applications of Sensor Nets


Structural Health Monitoring• Goal: Design sensor networks for

improving the safety of structures (buildings, bridges, ships, aircraft, spacecraft)

• Research focuses:– Local excitation-based damage

identification– System components for fine-

grain structural monitoring

• Multi-disciplinary effort:– John Caffrey (CE), Ramesh

Govindan (CS), Erik Johnson (CE), Bhaskar Krishnamachari (EE), Sami Masri (CE), Gaurav Sukhatme (CS)


Oilfield Safety MonitoringUse sensor net to detectand warn about leaks.

Challenges:– long-lived– easy deployment– self-configuration– condition-based maint.


Downhole Sensors for Control

Goal:see what’s happening downhole sensors monitor return mixturecut off side-wells at sign of water

Technical challenge:severe operating environment, communication and control


Virtual Reality?

Virtualizing operations …– sufficient timely data

– models of operation

… allows video-game like treatment– remote observation

– remote participation

– ‘game’ the operation

Fanciful? Perhaps.– requires lots of development


Lecture Plan





Recap: Staff & Smart AreasSmart area is networked …

Smart area has lots of sensors …– Devices, sensors & processes communicate via networks

– But people cannot

Solution: Provide staff with wearable, wireless computer– Acts as eyes and ears in smart areas

– Monitors smart area for its wearer


Wearable Computer

Hands-free use– In pocket or on hip

– Normally on (PDA is normally off)

Common examples:– PDA, Cell phone, iPod

– Dedicated application, not general-purpose


Ideal Wearable Characteristics

Mediates between smart area and user– Unmonopolizing– Unrestrictive– Observable– Controllable– Attentive– Communicative

Courtesy: Steve Mann, Univ. Toronto


PDAs Today

Blackberry 8700g• CPU: 312 MHz

• Wireless: GSM cell phone, Bluetooth

• Memory: 64 MB flash/16 MB ram

• Display: 320x240 pixels


Video iPod

• CPU: 200 MHz

• Storage: 60 GB hard drive

• Display: 320 x 240 pixels

• Battery: 3 hrs playback


PDA of Today• CPU: 300 MHz• Memory: 1 GB• Storage: 60 GB

• Display: 320 x 240• Communication• Interaction

High-end workstation in 1998

Weaknesses


Addressing Weaknesses

Display– Resolution too low

– Screen too small

– Power hungry

ApproachHead-mounts

– 800 x 400 (DVD-quality)

– 3m view seen from 1.5m


EyetapComputer modifies what you see

– camera at eye position

– display over eye

– image to eye + computer

– superimposed feedback Steve Mann –Univ. Toronto, wearable pioneer



Communication– Cell phone network is low speed

– Expensive infrastructure

ApproachUse more attractive alternatives: 802.11 …

– Higher speed (up to 24 Mb/s)

– Inexpensive infrastructure



Interaction– PDA or ‘phone-call’ model

– Blind to surroundings & non-collaborative

ApproachMove toward wearable ideal

– Multiple wireless interfaces (near/far)

– Monitor surroundings

– Collaborate with other hosts in area


Application – TrackingAssume …

– wireless sensors• uniformly at known positions around facility

• announce every 2 seconds

– wearables• monitor sensors

• reports its ID & sensor IDs recently heard

Receiver can know where wearables are …– precision determined by sensor/receiver range

– history provides tracking & heading


Scenario: Tracking

receiver

?updates

sensor zones


Application – AvoidanceAssume same plus …

– administration• announces/withdraws dangerous area descriptions

– wearables• possess map: sensor ID position

• monitor area announcements

Wearable knows what areas to avoid …– monitors its location/heading

– warns wearer when approaching danger


Scenario: Approach Warning

S2

S1

S3

S5

S4

warningregion


Application – Virtual Gang LockAssume same plus …

– monitor• receives task description, announces task area

• monitors task member positions

• controls device state

– wearables• task members announce position to monitor

Monitor ensures safe practices …– controls entry/exit

– controls shutdown/restart


Scenario 1: Gang Lock

monitor

work area

device


Scenario 2: Gang Lock

work area

device


Collaboration: Buddy System

If you rely on a wearable for safety, it betterbe operating …

• wearables monitor each other’s health– heartbeat protocol

• health implies functionality

Lack of health implies trouble …– associated individual OK?

– need to suspend affected activity?

– need to find another buddy?


Looking Under the Hood

Much of this is in its infancy.

Serious work to be done on …– Resource discovery

– Scenario description & communication

– Security


Resource DiscoveryRouting finds hosts by their address.

How do wearables find resources?– ex: buddy, device, process method

Three approaches:– directory service (central or distributed)

– diffusion (broadcast or multicast)

– area search


XML Scenario Description• <!ELEMENT warning EMPTY >

<!ATTLIST warning command CDATA #REQUIRED message CDATA #REQUIRED>

• <!ELEMENT group ( buddy* ) >

• <!ELEMENT buddy EMPTY ><!ATTLIST buddy ident CDATA #REQUIRED>

• <!ELEMENT region ( coord_system, sphere+ ) >

• <!ELEMENT coord_system EMPTY ><!ATTLIST coord_system units CDATA #IMPLIED coord_ref CDATA #IMPLIED>

• <!ELEMENT sphere EMPTY ><!ATTLIST sphere x CDATA #REQUIRED y CDATA #REQUIRED z CDATA #REQUIRED r CDATA #REQUIRED>

• <!ELEMENT action ( warn_group ) >

• <!ELEMENT warn_group ( group ) >


Security

Downside of remote control …– attacks

• denial of service

• unauthorized access/use

• Eavesdropping

Encryption– public key or traditional

Authentication– biometric

– public key encryption

Documents

© 2006 Gregory Finn 1 Advanced Oilfield Operations with Remote Visualization and Control ---------- Moving from PDAs to Intelligent Cooperative Assistants