real-time data integration for operational … • the vision of “operational bi (business...

28
Real-Time Data Integration for Strategic & Operational Intelligence Strategic & Operational Intelligence Michael R Brulé PhD P E Michael R. Brulé, PhD, P.E. Principal, Technomation ® 5 th International Conference on Integrated Operations in the Petroleum Industry Trondheim, Norway September 29-30, 2009

Upload: phungkiet

Post on 11-Jun-2018

220 views

Category:

Documents


0 download

TRANSCRIPT

Real-Time Data Integration for Strategic & Operational IntelligenceStrategic & Operational Intelligence

Michael R Brulé PhD P EMichael R. Brulé, PhD, P.E.Principal, Technomation®

5th International Conference on Integrated Operations in the Petroleum IndustryTrondheim, NorwaySeptember 29-30, 2009

Michael R. Brulé, PhD, P.E. Principal, Technomation®

Speaker BioSpeaker Bio

Mike Brulé is principal of Technomation®, a consultancy providing research, advisory, and implementationservices for E&P information management, software development, and real-time systems integration. Hespent the first half of his thirty years in the energy business with Kerr-McGee and Shell working in NorthSea and Gulf of Mexico field development and operations, and in alternative fuels research. Brulé currentlyfocuses on E&P enterprise architecture, information management, integrated oilfield modeling and

ti i ti kfl t ti ti l BI MPP d t h i d di ti l ti foptimization, workflow automation, operational BI, MPP data warehousing, and predictive analytics forreal-time decision-making and E&P business improvement. He holds a PhD in chemical engineering andan MBA from the University of Oklahoma.

Contact: [email protected]

 

 

Agenda

• The Vision of “Operational BI (Business Intelligence)”:

RealReal--Time Data Integration for Operational IntelligenceTime Data Integration for Operational Intelligence

p ( g )Improve Production Surveillance & Optimization (PS&O) effectiveness and efficiency by reducing time-to-decision

• Findings and implications of SPE Digital Energy Study Group’s i d t t d li ti i t tisurvey on gaps in data management and application integration

• Waterflood and workover field examples: Looking at the business impact of having “data ready-to-go”Technology Enablers to achieve Operational BI:• Technology Enablers to achieve Operational BI: – E&P Data Standards– Integrated Asset Modeling

AI Data Mining– AI Data Mining• Remarkable examples outside the oil & gas industry• Conclusion

SPE Digital Energy Study Group Mission

Facilitate implementation of the digital oilfield through integration of information technology people and processes by:

Oilfield Integration / RealOilfield Integration / Real--Time Operations / Data Management & Standards Time Operations / Data Management & Standards

information technology, people, and processes by: Identifying opportunities for improvement and supporting the

development and implementation of information technology integration solutions, standards, and best practices spanning theg p p gbusiness, surface, and subsurface domains

Regulatory / Financial / Logistics

Facilities / Operations / Production

Efficient IntegrationReduce Cost

Facilities / Operations / Production

Drilling / Reservoir / Geoscience

Effective IntegrationProduce More Oil and Gas

Time-Based Operational Processes:Multi-Time-Scale & Multi-Subject A Fundamental Tenet: A Fundamental Tenet: DecisionsDecisions drive ActionsActions that create ValueValue

Unneland, T. and Hauser, M. 2005. “Real-Time Asset Management: From Vision to Engagement—An Operator’s Experience” Paper SPE 96390 presented g g p p p pat the Annual Technical Conference and Exhibition, Dallas, 9–12 October.

Brulé, M., Charalambous, Y., Crawford, M., and Crawley, C. 2009. “Reducing the Data Commute Heightens E&P Productivity” JPT September.

Contexts of Oilfield Integration

• People integration real time collaboration 24x7 around the globe

SPE subcommittees on RealSPE subcommittees on Real--Time Operations & Oilfield Integration study the following:Time Operations & Oilfield Integration study the following:

• People integration—real-time collaboration, 24x7, around the globe• Data integration—needs to be accessible to everyone, and consumable

in applications. • Operations Monitoring & Control integration—“RTOCS” operationsOperations Monitoring & Control integration RTOCS operations

centers, 24x7 ubiquitous monitoring portals• Disciplines integration—The leverage of experts across disciplines like

geoscience, drilling, reservoir, and production engineering• Integrated Asset Modeling— “composite” modeling and simulation,

integrating more operations that comprise the business• Process integration—Combining work processes for greater efficiency• SOA and industry standards applications integration integration for• SOA and industry standards—applications integration, integration for

interoperability

Data stand out as fundamental and foundational to all the other integration contextsintegration contexts.

SPE Digital Energy Survey Summary

Th “D t C t ” i C tl t th E&P I d tTh “D t C t ” i C tl t th E&P I d t

• Anecdote: “Engineers spend half their time chasing data”• Statistics: 44 percent of an E&P professional’s time for production surveillance

The “Data Commute” is Costly to the E&P Industry The “Data Commute” is Costly to the E&P Industry

p p pis spent gathering and preparing data for analysis. Only 14 percent of the surveyed people had more than half of their time available for doing analysis.

• People time and many millions of dollars are lost to nonproductive data-handling activitieshandling activities

• Big Crew Change: Retiring-expert knowledge must somehow transfer to newcomer and crossover employees.

• Data Explosion: With each instrumentation advance, more data must be integrated, at different timescales, and across many disciplines

• Data management gaps impact workflow automation, process transformation, and application rationalization.

With a hopping half the time lost on data chores that’s aWith a whopping half the time lost on data chores, that’s a 2x potential increase in the workforce…with the same people!

Sources: JPT, Sep 2009, “Reducing the Data Commute Heightens E&P Productivity” and SPE 116758 “Bridging the Gap between Real-Time Optimization and Information-Based Technologies.”SPE 110236 (SPE RTO TIG, CVX, HAL, SLB, others)SPE 97288 C ti i Ed ti N d f Di it l Oil Fi ld SPE 112152 XOM PS&OSPE 97288 Continuing Education Needs for Digital Oil Fields. SPE 112152 XOM on PS&O. World Oil Apr 2008 (Shell-Drilling Surveillance), also Nov 2007 “Improving E&P Performance with Right-Time Business Intelligence.” SPE Production & Operations, Nov 2006 (XOM, HAL, CVX, COP, RDS, BP, others)Digital Energy Journal, Sep 2009. “BP – using connectivity to drive productivity” and “Using massively parallel processing databases”

The “Big Data” Problem: Hundreds of silo’d data and systems

Enterprise (e.g., SAP)> Applications must retrieve

data from these systems

The Typical E&P Operator has data in many systems for many purposes.The Typical E&P Operator has data in many systems for many purposes.

p ( )data from these systems > Data types include

– Geophysics (structures, rock properties, migration patterns)R i / d ti i i– Reservoir/production engineering

– Platform (equipment)– Manual (Worker input, e.g.,

data capture by pumpers)E t i (SAP th )

Operational & Equipment Data

– Enterprise (SAP, others)

Manual Data CaptureExploration

Production

Though popular, “portal integration of data” provides only Visibility, not Transparency

Visual Portal for Monitoring Production KPIsVisual Portal for Monitoring Production KPIs——Good for surveillance, but Good for surveillance, but not for not for predictionprediction

Dashboards alert asset team to a

Asset team determines root cause, takes action,

Asset team monitors

Asset MgrReservoir/ Prod Engr

Field personnel

problem and reforecasts plan

Reservoir/ Prod EngrGeology/petrophysics

Opns Supvsr/Field Reservoir/ Prod Engr

Opns SupvsrFi ld l

monitors operations

See World Oil, “Improving E&P Performance with Right-Time Business Intelligence,” Nov2007.

Field personnelpersonnel

Planning analystField personnel

Let’s separate some of the effects and look only at the impact of making a high-quality decision faster

• We are:NOT changing the amount or nature of the scientific or

Making the most of our science and engineeringMaking the most of our science and engineering

– NOT changing the amount or nature of the scientific or engineering work

– NOT changing the advances in the SPE, AAPG, SEG, AIChE, or rest of the industry in field surveillance, geoscientific study, y , g y,and engineering best practices

– NOT exercising snap judgment or changing the quality of the decision

– NOT really doing anything differently than we would have otherwise

We are just looking at removing the lost non value adding time in• We are just looking at removing the lost, non-value-adding time in getting to the decision, amplifying the productive time of engineers and geoscientists

R i th t /l t k i t dRemoving the entropy/lost work associated with the “Data Commute”

Waterflood Case History: Effect of “Time to Decision”

Looking only at one effect: Looking only at one effect: making the decision fastermaking the decision faster

Spraberry Driver Unit, West Texas• Production after primary recovery: 2,620 bbls/day• Avg Production over waterflood period: 4,270

Mi i t fl d t d ti 8

p y

• Minimum waterflood study time: 8 mos.• Improvement with data integrated and accessible:

– “Time to Decision” cut in half to 4 mos.– 4 mos. ≈ 200,000 bbls accelerated production – Worth $2 million/month at today’s oil prices

C O G SL.F. Elkins et al., Field Case Histories, Oil and Gas Reservoirs, SPE Reprint 4a, Dallas 1975.

Bigger Benefits:Visibility vs. Transparency of the E&P BusinessSurveillance without Optimization is a little like the Surveillance without Optimization is a little like the “frog in boiling water”“frog in boiling water”

What about the “Optimization” in PS&O (Production Surveillance & Optimization)?

• Surveillance helps us see impending problems when they are happening or about to happen, through comparison of current trends with past trendsof current trends with past trends

• Optimization helps us see problems before they happen, through comparison of trends with predictions from first-principles models or “the new AI,” i.e., statistical and stochastic analytics

What about the “Optimization” in PS&O?What about the “Optimization” in PS&O?Opportunity Areas for Complete Oilfield IntegrationOpportunity Areas for Complete Oilfield Integration

Increase

“Accelerating” “Accelerating” and and “Uplifting” through complete integration“Uplifting” through complete integration

+ultimate recovery(optimization uplift)

WOGLO

0

Accelerate & maximize production

Lower operating& intervention

costs (time-shift acceleration) ul

ativ

e Fl

ow

GLOWF/EOR

0Earlier & Better Decisions

)

Cum

uC

ash

-

Project / Asset Lifecycle

Bigger Benefits:Transparency of the E&P BusinessMoving from the Moving from the “frog in boiling water” “frog in boiling water” to to “watching the asset like a hawk.”“watching the asset like a hawk.”

Other benefits from complete integration p gwill be realized that are much larger:

• With daily integrated optimization with data t tl f d t d l th d fconstantly fed to models, the need for a

waterflood study will have been realized earlier (not just the time that the study would have taken). )

• A study would have been initiated in, say, 2 years, instead of the traditional 3-5 yrs. Add at least another 3x plus to NPV.

• Surface Facilities would not have to be overdesigned to catch up on water injection.

• Safety and compliance issues would be greatly reducedwould be greatly reduced

Tale of Two Forums: Theoretical vs. Empirical

The Alignment Challenge: Aren’t we talking about the same goals? The Alignment Challenge: Aren’t we talking about the same goals?

Source: www.SPE.org

Digital Energy JournalSeptember 2009

• (p. 4):“A growing area is data analytics services. That's an area that will really take off in the next few years So for example we can look at our pipeline and how

There’s room for Empirical There’s room for Empirical plus plus TheoreticalTheoretical

in the next few years. So for example, we can look at our pipeline and how measurements of wall thickness over time and how corrosion takes place - and use that to make empirical physics calculations as opposed to theoreticalphysics calculations.”

• (p. 22):Look at what marquee companies outside of the oil & gas industry are doing very successfully. A multitude of opportunities exist to apply their mature approaches to Operational BI in E&P. Much work is being done in process transformation and p g papplication rationalization, but the low hanging fruit lies in improving data systems and processes.

( 19)• (p. 19):“One of the problems which happens too often in the field is literally the lack of implementation detail available for decision-making, hence an obstacle to new technology adoption. Instead, the scenario which unfolds is that the individual describing the new technology is typically unable – in lieu of a detailed implementation methodology – to providetechnology is typically unable in lieu of a detailed implementation methodology to provide end-users an accurate picture of (a) exactly how the technology will be integrated into day-to-day operations and (b) how all the various risks to business interruption will be addressed in detail.”

DEJ can be downloaded from www.digitalenergyjournal.com/issues/DEJ20web.pdf

Example Marquee Companies Successful at Real-Time Data Management, Predictive Analytics, & Optimization

• Boeing and the US Air Force use operational business intelligence for aircraft engine design, fleet-wide aircraft maintenance, reliability and parts tracking,

Looking Outside the Oil & Gas IndustryLooking Outside the Oil & Gas Industry

g g , , y p g,analysis, and 24x7 collaboration with service providers

• Caterpillar and Ford use them for supply chain and inventory analysis, early warning system, quality, & warranty analysis

• Wal Mart eBay and amazon com use them to take actions based on data• Wal-Mart, eBay, and amazon.com use them to take actions based on data combined across all enterprise subject areas, within seconds of a transaction anywhere in the world.

• Western Digital uses MPP database technology to mine millions of detailed data i t t i i l ti f i f di k d i f t ipoints streaming in real-time from many pieces of disk-drive manufacturing

equipment and also data from their supply chain. The data are processed with a systematized and automated “analytics factory” approach, with in-database Weibull or gamma-distribution analytics and Western Electric rules, based on

i i l f t ti ti l t lprincipals of statistical process control.

With each complex problem that is solved, it can be solved thousands of times again within a systematized and automated “analytics factory.”g y y y

Operational BI Maturity StagesStage 5

mpl

exity

Stage 4

Stage 5ACTIVE DECISIONINGMAKING it happen!

kloa

d C

om

gOPERATIONALIZINGWHAT Is Happening?

Stage 3PREDICTING

y an

d W

ork

Continuous Update & Ti S iti Q i

Event Initiated Actions Takes Hold, “Closed-Loop”

Stage 2ANALYZINGWHY did it happen?

WHY will it happen?

sing

Que

ry Time Sensitive Queries Become Important

pp

Stage 1REPORTINGWHAT happened? Analytical

ModelingGrows

Batch

KPIs, Dashboards, Ad Hoc

Complex AI Analytics, New Insights

Incr

eas

Increasing Data Detail Volume Integration & Schema Sophistication

Continuous Model Update plus Real-Time Queries

Complex Event-Based TriggeringPrimarily Batch

Increase in Ad Hoc Queries

Increasing Data Detail, Volume, Integration & Schema Sophistication

After Brobst, S. and Rarey, J. Proc. Data Warehousing (2002).

OPERATIONAL INTELLIGENCE

Make adjustments to smart wells to optimize production, lower costs, and increase profitability

Monitor and predict production trends monthly,

STRATEGIC INTELLIGENCE

p yp y,weekly, daily, hourly, minutes, seconds, sub-second(see Digital Energy J. article)

ACTIVATING

What caused the production decline in the

waterflood unit?(e.g., World Oil article)

Would a workover program for production wells result in

restored production?

ACTIVATINGMAKE it happen!O&G Example: What was

the production for a waterflood unit last month in a company’s West Texas field?

OPERATIONALIZINGWHAT IS

happening now?

REPORTING

ANALYZINGWHY

did it happen?

PREDICTINGWHAT WILL

happen?

Integrated Production Operations yields better forecasts that drive profitability through higher

REPORTINGWHAT

happened?

did it happen?

asset performance levels (increased production) and lower overall costs, including a reduction in safety and compliance events.

Chevron i-field “staircase,” Unneland & Hauser (2006)

Three High-Level Data Capabilities Crucial to Operational BI in E&P

1. Integrating data of different timescales:

MultiMulti--TimeTime--Scale/Multi Scale/Multi ––Subject “Models and Mining”Subject “Models and Mining”

g g• Drilling, reservoir, and production engineering would all benefit from

being able to combine historical (accumulated over years), tactical (weeks to months), and high-frequency data from historians (sub-

d t d t d )seconds to seconds to days) • To cover more of the enterprise, include data from other discipline-

oriented source systems, including the underlying data stores of shared-earth-modeling suites and financial data from ERP systemsshared-earth-modeling suites, and financial data from ERP systems such as SAP

1. Integrating data across disciplines and across multiple subject areas, i.e., Multi-Scale/Multi-Subject j

2. Implementing a systematized and sustainable data “factory” approach to augment the industry’s traditional full-physics modeling methods with statistical and stochastic methods,

i.e., “Models and Mining”

Integrating Data of Different Timescales• Most current production optimization suites are based on historians, which slip-stream a

sampling of their data to relational databases

Source: Reid et al., “Holistic Production Optimization Achieved One Workflow at a Time ” JPT April 2008

• Integration of all the data at different timescales> Reveals statistically important events> Provides more data for modeling of drilling and production processes

Achieved One Workflow at a Time, JPT, April 2008.

> Gives geoscientists, engineers, and operations professionals consistent and integrated data for more informed decision-making

> Brings the engineers closer to the operations people

Cross-Discipline Collaboration

Source: http://www.ted.com/talks/robert_full_learning_from_the_gecko_s_tail.htmlTED stands for Technology, Entertainment and Design, although it represents much, much more. Since its inception as a conference in 1984, it has emerged as a premier site for “ideas worth sharing.”

Multidisciplinary Collaboration and Multidisciplinary Collaboration and D t I t ti i E&PD t I t ti i E&PData Integration in E&PData Integration in E&P

Traditional Contemporary

GeophysicsGeophysics Petrophysics

PetrophysicsGeologyGeology

Reservoir Engineering

Reservoir Engineering g gEngineering

Reservoir ModelReservoir ModelAfter Cosentino, L. Integrated Reservoir Studies, IFP Publications, 2001.

Combining PVT with seismic to locate oil• Acoustic impedance depends on rock and fluids so take the effect of the rock out of theAcoustic impedance depends on rock and fluids, so take the effect of the rock out of the

seismic data.• Cannot tell the difference between all possible oil, gas, and water mixtures.• Simulation showed that o/g/w or g/w could not be flowing at the same place, at the same

ti (h / / b f ti )time (have o/w or o/g because of segregation)• If AIoverall>AIOil=>it’s wet. If AIoverall<AIOil => it’s O/G, good completion potential!

Water Oil

Gas

CompletionPotential WetPotential

Addy et al., “Determining the Location of Remaining Oil Using Acoustic Impedance: Poza Rica Field, Mexico,” SEG Workshop 2003.

Integrated Asset ModelingSource: Juell et al Model-Based Integration & Optimization SPE IE 2009 Amsterdam

• IAM provides holistic view of combined operations and of the business• Provides more insight than independent models and simulations

> In analogy with looking at all the data in AI data mining, the IAM sum could be greater than the individual model partsI d ll b ti di i li

Source: Juell et al., Model Based Integration & Optimization. SPE IE 2009, Amsterdam.

> Induces more collaboration among disciplines

• Implications of SPE Digital Energy Survey on IAM:> 83 percent of people surveyed perceive gaps in application integration that impede their daily work. > Tools are format-sensitive; constantly repackaging data from one application to the next> Tools cannot be run in automatic mode; require specialists> Tools cannot be run in automatic mode; require specialists> Commercial unified tools suites exist, but many people still resort to Excel.

• To be successful, IAM must have:> A pluggable and extendible integration framework that is sustainable

Leverage of industry standard XML protocols such as Energistics WITSML and PRODML– Leverage of industry-standard XML protocols such as Energistics WITSML and PRODML– Robust connectivity for applications, ensuring reliable material and energy balancing– Rank order the most important control variables– Bring together disciplines to compare variables on a common ground, e.g., business and economic goals

> Easily interchangeable models—detailed full-physics to simple approximations, with modelsEasily interchangeable models detailed full physics to simple approximations, with models changeable within time domains and need for prediction reliability

– E.g., for transient pipeline during facility start-up, switch to OLGA; after steady-state, GAP– Quick change-out from detailed compositional reservoir simulation to simple black oil model

> Easier-to-use interfaces with application interconversion and interoperability nearly automatic– Support of proxy models capable of representing more detailed simulation models– Choice of models, not dictated by one brand or company => able to embed internal know-how

Important BIOs* in Surveillance & Optimization

1. Predictive Equipment Maintainability & ReliabilityEmploys predictive methods of equipment reliability and

*Business Improvement Opportunities*Business Improvement Opportunities

– Employs predictive methods of equipment reliability and maintainability practiced with high success in other industries

2. Drilling Surveillance & OptimizationO C– Operations and Cost transparency in drilling operations

– Predict costly events such as stuck-pipe incidents

3. Well & Field Asset Surveillance– Operations and Cost transparency in oilfield operations– Predict operational problems such as overheating of equipment

4 Production Optimization4. Production Optimization– “Models & Mining”: full-physics models facilitated by cross-disciplines

integration and in-database statistical analytics

Summary: A Modern Approach to Reduce Time-to-Decision and Improve E&P Operations Optimization

– SPE survey showed Data to be a top problem for integrated operations• Data management must be standardized systematized and automated

Systematically Managing Data & Combining “Models and Mining”Systematically Managing Data & Combining “Models and Mining”

Data management must be standardized, systematized, and automated• Application interoperability would benefit from IAM systems that are flexible,

pluggable, and extendible• Looking at other industries, industry data standards and IT tools for BI (business

i t lli ) i h l f lintelligence) is helpful:– E&P industry standards such as Energistics WITSML/PRODML for XML Web

services protocols in an SOA– For scale and performance, MPP in-database analytics is a powerful capability

that enables systematized and automated management of data and tactical decisionthat enables systematized and automated management of data and tactical decision-making.

– Statistical and Stochastic in-database predictive analytics can be used when there are too many effects for full-physics models

Combine IAM full physics modeling with in database mining to augment the– Combine IAM full-physics modeling with in-database mining to augment the industry’s traditional use of models and simulation.

– People challenges include the need for management leadership to drive data integration directives which the leaders of companies outside the oil & gas industryintegration directives, which the leaders of companies outside the oil & gas industry have done with high success.

Questions?

Thank [email protected]