View
4
Download
0
Category
Preview:
Citation preview
© Peter Kaznacheev
Presentation at the Society of Petroleum Engineers, London 25 September 2018
Peter KaznacheevSenior Research Fellow, King’s College London, European Centre for
Energy and Recourse Security (EUCERS)
OIL AND GAS ARTIFICIAL INTELLIGENCE – OGAI 2.0
The Application of Geospatial Information and Big Data Analytics
© Peter Kaznacheev
The structure of this presentation
© Peter Kaznacheev
Part One. OGAI 1.0. Operational applications of AI - an overview
- Upstream - Midstream - Downstream
Part Two. OGAI 2.0. Market analysis and geospatial information
- Who is collecting geospatial information - Who is using geospatial information - Market analysis based on regional data - Market analysis based on leading indicators - Market analysis based on proxy data - Integrating various types of data
Conclusions
2
! Reservoir analysis ! Seismic data
acquisition ! Seismic data
processing ! 3-D seismic imagery
AI in Geological surveys AI in Exploratory
drillingAI in Appraisal
! Well planning ! Automation of
drilling rigs ! Monitoring of
drilling ! Real-time drilling
optimisation! Well testing using
intelligent sensors ! Prediction of
failure of equipment
! Management of geological and geophysical data
! Well log digitising ! Well log
interpretation ! Calculation of
appraisal drilling parameters
! Reservoir modeling ! Reservoir
characterisation
Operational applications of AI Upstream — exploration
© Peter Kaznacheev
3
! Optimisation of production throughout its various stages
! Flow measurement in wellheads
! Automated production control
! Reservoir pressure maintenance
! Enhancing recovery ! Operational troubleshooting ! Safety control ! Forecasting behavior of a
layer depending on its current state
AI in field development AI in production
! Field surveillance ! Placement of well pads ! Optimisation of well
location and design ! Wells controlling ! Access to field equipment
and diagnostics ! Improving the rate of
penetration ! Automated Manufacturing
Execution System ! Wireless Sensor
Networks
Operational applications of AI Upstream — field development and production
4
" Determining the optimal path for pipeline construction
" Determining locations of terminals and pipeline connections
AI
" Compressor station automation
" Early warning about malfunctions
" Corrosion monitoring and diagnostics
" Calculating the most cost effective route
" Autopilot and autodocking systems " Tanker monitoring
" Analysing economic conditions and weather patterns to forecast demand
" Storage optimisation
Operational applications of AI Midstream
© Peter Kaznacheev
5
# Predictive analytics and predictive control algorithms
# Reduced costs of operating storage facilities
# Efficient equipment maintenance
# Automation and asset optimisation of refining processes
# Use of sensors and micro-controllers for infrastructure maintenance and early warnings
# Distant monitoring of storage facilities and equipment
# Leakage detection
# Detection of equipment deformation
Storage Management Petrochemical Refining Security and Environment
Cost reduction
Improved efficiency
Increased safety
Enhanced output
Environmental compliance
Operational applications of AI Downstream
© Peter Kaznacheev
6
Who is collecting geospatial information Selected companies
© Peter Kaznacheev
# Real-'medatafromland,seaandsatellitemonitors.PartoftheDailyMailTrust.
# Shippingandstoragedata.Specialfocus:drillingac'vityandrigcounts.
# Combiningcommodityshipmentsdatawithsatellitefeedsandgovernmentaldatabases.
# Inventorydataderivedfromsatelliteimagesofexternalfloa'ngrooftanks(EFRT).
# Real'meshippingdata,includingtankers.Globalmari'memap.
# Satelliteimageryanalysisofmari'meac'vity.Forecas'ngtonnageavailability.
# Datafrompublicandcommercialgeospa'alimageryproviders.
# Wellmonitoring,crudestorageandrefineriesdatafromsatellitefeeds.
Maritime Portal
7
ORBITALINSIGHT
© Peter Kaznacheev
# Hedgefunds# Traders# Refining&marke'ng
# Chiefeconomistsandstrategydivisionsofoilcompanies
# Shippingcompanies
# Shipbrokingservicefirms
# Governmentsofoilexpor'ngcountries
# Thinktanksandresearchcentres
# Interna'onalorganisa'ons
SHELL SCENARIOS
8 Who is using geospatial information
Market analysis based on regional data Monitoring tank farms and refineries
© Peter Kaznacheev
# Targetaudience:tradersanddownstreamanalysts.
# Casestudy:OrbitalInsightusessatelliteimagesofexternalfloa'ngrooftanks(EFRT).
# Thelevelofliquidcanbemeasuredbasedonthethicknessoftheshadowontheroof.
# UsingAIalgorithmstomeasurestocksinspecifictankfarms,regions,countriesorglobally.
9
Satellite images of external floating roof tanks — Orbital Insight
Types of oil storage
Market analysis based on regional data Using inventory data as a proxy
© Peter Kaznacheev
# Inventorydatacanbeusedforthepurposesofdownstreamcompe'torintelligence.
# Anotherapplica'on:iden'fytrendsincrudesupply.
# Changesinstocksserveasproxiesforoilproduc?ontrendsinkey“triggercountries”i.e.Venezuela,Iran,Iraqetc.
# Officialdatareflec'ngcrudeproduc'oninthosecountriesisunreliable.
Data source: OPEC
Crude production in thousand barrels per day
# Example:TheextentofVenezuela’scrashinoilproduc'onwaslargelyunforeseenbyanalysts.
# 0.5mlnb/dwastakenoffthemarketin6months.
# Oneofthemainfactorsbehindtheoilpricerisein2017-2018.
10
Market analysis based on leading indicators Drilling activity, rig counts, car counts, construction
© Peter Kaznacheev
# Someexamplesofleadingindicators:
- Offshorestorageintankers
- Drillingac?vity(especiallyintheUSlower48)
- Pipelinecapacityandthroughput(especiallyinthePermianbasin).
- Construc?onac?vity- Cellphonedata- Carcountsaroundoilandgassites.
# Casestudy:carcountsdataisavaluabletoolforretailanaly'cs.
# Alsoaleadindicatorforoildemandonaregionalorgloballevel.
11
Market analysis based on proxy data Renewables and EVs
© Peter Kaznacheev
# TrendsinrenewableenergyandEVscontributetopeakoildemand.
# Evensmallchangesinthosefactorsaretrendindicators.Theyhaveanimpactontoday’soilprice.
# Mostmodelsarecurrentlyrelianton3-dpartyforecastsofrenewablesandEVs.
# Needfordatasourcedfromrealobserva'ons(e.g.solarfarms).
# Example:Bloombergforecaststhatelectriccarswillmakeup35%ofnewvehiclessalesby2040.
# SeveralOECDcountriesareplanningtobanthesaleofnewpetrolanddieselcarsby2040.
# Thiscouldreduceglobaloildemandby8mlnb/d.
12
FUTURE STEPS IN COMMODITY
MARKET ANALYSIS
SUPPLEMENTING ECONOMETRIC MODELS
• 3-d party expert forecasts
• Price volatility (GARCH models)
• Market sentiment data
RESEARCHING MARKET SENTIMENT
• Semantic analysis of news events and other open sources
• Commodity market sentiment and «appetite for risk»
ARTIFICIAL INTELLIGENCE METHODS
• Neural networks, fuzzy logic etc.
• Combining econometric models and AI
SCENARIO AND INTERVAL FORECASTING
• Interval forecasting in addition to “point forecasts”.
• Price movement scenarios
10
Integrating various types of data Future steps. Working with various analytical tools
13
© Peter Kaznacheev
Conclusions
© Peter Kaznacheev
# OGAI1.0focusedonopera'onalapplica'ons:upstream(bothexplora'onandproduc'on),midstreamanddownstream.
# OGAI2.0willfocusonmarketresearch,corporatestrategyandplanning.
# Twoapproachesinmarketanaly'cs:nichedatavs.en'revaluechain.
# Subscribers:hedgefunds;traders;IOCsandNOCs;shipping/shipbrokingfirms;governmentresearchcentres;interna'onalorganisa'ons.
# Threemajoranaly?calroutes:1)Marketanalysisbasedonregionaldata2)Marketanalysisbasedonleadingindicators3)Marketanalysisbasedonproxydata.
# Integra?ngvarioustypesofdata:1)Mul'-factorforecas'ngmodels2)Mapsofpricedrivers.
14
© Peter Kaznacheev 15
Disclaimer:© Peter Kaznacheev. All Rights Reserved.No duplica'on or redistribu'on of thisdocument is permimedwithout wrimen consent. Peter Kaznacheevmakes no representa'onabout the completeness or accuracy of the data, calcula'ons, informa'on or opinionscontainedinthisreport.Viewsexpressedhereinhavenotbeeninfluencedbyanyrela'onshiptoanyen'tyreferredtohereinnortoanyclientofPeterKaznacheev.
Peter KaznacheevS. 56, 95 Wilton Road, Victoria, London, SW1V 1BZ, United Kingdom�
khaznah@khaznah.co.uk | www.kaznacheev.com
Recommended