how to use analytics to improve...

Event sponsored by:

How do man and machine best work together?What can big data do for Scotland?1234most - Collaboration to make better production forecastsTeradata - How to get value from analyticsData Lab - Letting academia work on your data problemsHalliburton - data, models, workflow andchange managementJonathan Guthrie - how to get past 'no'

Event Report, How to use Analytics to improve Production, Sept 29, 2015, Aberdeen

Special report

How to use Analytics to improve ProductionSeptember 29, 2015, Aberdeen

KRAKEN is a product of

Centralize reservoir engineering data in a single environment

Reduce Post Simulation Model QC/QA and Data Analysis Time

Increase reservoir engineering productivity

Customize and automate reservoir engineering workflows

Import and analyse reservoir models from various simulators

Create multiple simulation scenarios comparison and sensitivity analysis

Implement efficient history matching

Explore Analytical Production Forecasting

Seamlessly couple with third-party engineering tools

Find out how you can:

Taking Large Reservoir Data Handling to New Frontiers

A cost efficient analytical toolsaving valuable engineering resources and time

Contact us now:[email protected]

www.enginsoft.comwww.kraken.com.br

Handling to New FrontiersTaking Large Reservoir Data



resources and timesaving valuable engineering A cost efficient analytical tool

Handling to New Frontiers






Data Analysis TimeReduce Post Simulation Model QC/QA and

single environmentCentralize reservoir engineering data in a


Data Analysis TimeReduce Post Simulation Model QC/QA and

single environmentCentralize reservoir engineering data in a

Reduce Post Simulation Model QC/QA and

Centralize reservoir engineering data in a

various simulatorsImport and analyse reservoir models from

engineering workflowsCustomize and automate reservoir


Data Analysis Time

various simulatorsImport and analyse reservoir models from

engineering workflowsCustomize and automate reservoir


Data Analysis Time

Import and analyse reservoir models from

Customize and automate reservoir


Seamlessly couple with third-party



comparison and sensitivity analysisCreate multiple simulation scenarios

various simulators





various simulators





www.kraken.com.brwww.enginsoft.com

[email protected] us now:

www.kraken.com.brwww.enginsoft.com

[email protected] us now:

engineering tools

AKEN is a product of KR


engineering tools

AKEN is a product of



This special edition of Digital Energy Journal is an Event Report from our forum in Aberdeen on September 29, 2015, "How to use Analytics to improve Production".

How to use Analytics to improve Production

3

Event websitewww.digitalenergyjournal.com/event/e66da.aspx

Report written by Karl Jeffery, editor of Digital Energy [email protected] Tel 44 208 150 5292

Sales manager Richard [email protected] Tel 44 208 150 5291

Conference produced by Karl Jeffery

Layout by Laura Jones, Very Vermilion Ltd

Cover art by Alexandra Mckenzie

Digital Energy Journalwww.d-e-j.com

Future Energy Publishing, 39-41 North Road, London, N7 9DP, UK www.fuenp.com

In the world of chess computing, “every chessplayer will agree that the combination of ahuman plus a chess computer program are ab-solutely unbeatable,” said Sergio Sama Rubio,Industry Solutions Advisor with Halliburton,speaking at the Digital Energy Journal confer-ence in Aberdeen on Sept 29, “Using Analyt-ics to Improve Production”.

It should be similar in the world of oil and gasproduction. If you can combine the human’sstrengths (strategy, designs and choices, ob-jectives and decisions) with the computer’sstrengths (repetitive tasks, running models,optimising and training) you can work out thebest way to get the most production and max-imum safety with your resources available.

Dr Duncan Irving of analytics data platformcompany Teradata said that oil company staffwill typically hand one data file onto the nextperson in the chain with no-one keeping trackof the algorithms or models used to create it.

By the time the data is used to make a deci-sion, sometimes the only version of the datais on a PowerPoint slide. “This is not the sameof pretty much any other industry Teradataworks in,” he said.Scott Harrison of Scottish Enterprise said hehad been speaking to a senior leader from anoil and gas maintenance company, whopointed out that he gets paid in terms of manhours from sending staff offshore to fixthings. It might be better for the client if hecould do predictive analytics and only fixthings which need fixing, but not better for hiscompany.

The conference was not all negative. TomFox, a former field development engineerwith Shell, explained how to bring togethercommercial, production and maintenancestaff together to work on a production fore-cast, all putting their respective expertise intodeveloping and accessing a model which theycan access through a web browser.

Duncan Hart of Data Lab offered the re-sources of Scotland’s academic community toassist with any data problem the industry has,with funding from the Scottish government topay for it.

Oludare Elebiju of Enginsoft explained howcompanies can set up a central data repositoryfor all reservoir data, using technology origi-nally developed for a very large Brazilian oiland gas company.

Halliburton’s Sergio Rubio explained howsetting up well structured ‘workflow’ systemsis a great way that oil companies can harnessthe expertise of retiring staff members.Jonathan Guthrie explained how you can startsmall with an analytics problem to prove itworks, and Duncan Irving explained how thegoal is to achieve a far more ‘sentient’ organ-isation – one which has a much better under-standing of what is going on.

It isn’t an easy time to be selling technologyor services – energy consultant JonathanGuthrie told a story of how he identified anopportunity for an oil company to accelerateand deliver a potential £15m of businessvalue. The initial analysis was made from theoil company’s own data. To validate and com-plete the analysis would require a short-termcommitment of £15k from the operator. Butwas then told to cut his consultancy fees by10 per cent because ‘that was company pol-icy’. He walked away from the business.

But the current oil price environment meansthat many companies have more time to eval-uate better data technologies and pressure tosee if they can deliver results. But can they getpast the senior management who prefer to“just do the same rubbish we’ve been doingbecause it feels safe”?

Scottish Enterprise – understanding big data

Scottish Enterprise has been looking into 'bigdata' since spring 2014 to find out what valueit can actually add, said Steve Harrison, proj-ect manager with Scottish Enterprise.

It wants to help oil companies separate hypefrom reality with analytics. If all the hype wastrue, "we'd find oil in a second, we'd be ableto maximise production and turn it into dol-lars," he said.

Senior management are asking for more 'evi-dence'. "We've been working with leadershipto understand what evidence they need," hesaid.

Another challenge Scottish Enterprise is look-ing at is improving the goal alignment be-tween operators and suppliers. For example,the incentives for predictive analytics are notas great as you might think. A senior leader ofa maintenance company once told him, ‘I getpaid for the man hours for people who go outand fix kit. No-one will pay me for tellingthem when it’s going to break.’ Mr Harrisonsaid.

Scottish Enterprise wants to help operatorsand service chain companies to help make theUKCS more competitive, and help the com-panies develop into international dominantplayers, he said. But also consider that themain objective of government, which paysScottish Enterprise's bills, is often jobs, notcompetitiveness.

Watch Steve's talk on video atwww.digitalenergyjournal.com/video/1739.aspx

How to use analytics to improve production Digital Energy Journal’s Aberdeen conference on Sept 29 2015 explored how man and machine (or woman andmachine) can work together to achieve the best results in production

Digital Energy Journal - Special report, How to use Analytics to improve Production, Sept 29 2015


4 Digital Energy Journal - Special report, How to use Analytics to improve Production, Sept 29 2015

Together we can improve production forecastingProduction forecasting is a critical task in the oil and gas industry, and is best if you have a mixture of complexmodelling and the combined expertise of people from many different disciplines. Tom Fox explained how todo it.

Oil and gas asset managers need reliableproduction forecast information so they cantake better decisions about how to max-imise value from every well. And makingbetter production forecasts needs a mixtureof analytics and human expertise.

“Analytics are the key to understandingpast production, but analytics alone fail tomake an optimal forecast,” said Tom Fox,managing director of production consul-tancy 1234most, and a former field devel-opment consultant with Shell, speaking atthe Digital Energy Journal Aberdeen con-ference on Sept 29 2015, “Using Analyticsto Improve Production.”

“A digital solution is an incomplete solu-tion. If we leave out the people, that digitalsolution is probably useless,” he said.

Production planning is complex because itmust reconcile three perspectives: commer-cial, production and maintenance. Typically, these are three separate analysisflows which involve different expertise anddiverse software, he said.

There are also uncertainties in both themodelling methods and the input data, socase studies are needed to assess a range offuture scenarios.

Big data techniques, such as machine learn-ing, offer a tempting shortcut to decisionsupport, he said.But if you skip the laws of engineering andphysics, you can end up overloaded withfalse alarms mixed with false positives.

Engineering and maintenance models, cal-ibrated by historical analysis, are essentialto make valid predictions.

DynamicForecaster © 1234most Ltd

Know sooner, decide better, act faster

5



Need to be integrated

The Piper Alpha disaster is a sobering re-minder of the human and commercial costsof management failing to make good deci-sions, due to weaknesses in the integrationof complex activities Mr Fox said.

More than twenty-five years later, with lowoil prices, we are once again under hugecommercial pressure to keep productionflowing, in order to defer asset abandon-ment costs. The resulting imperative is toto carry out maintenance activities at min-imal cost and with reduced downtime.

Complex decisions are especially prone topsychological decision traps. Sadly, we arenot as rational in practice as we claim to bein textbooks and management courses, MrFox said.

The best antidotes to these psychologicaltraps are collaborative decision-making andexplicit recognition of the various forms oftraps, he said, in some lessons taken fromthe 1999 book “Smart Choices” by HowardRaiffa and Ralph L Keeney.

Collaborative decisions

Collaboration deserves purposeful invest-ment of time, energy and money, Mr Foxsaid.

Everyone makes decisions, at all levels inthe organisation and in all locations. Whenwe involve more people in taking produc-tion planning decisions, the decisionprocess must evolve to handle the increasedcomplexity.

For example, instead of a binary yes or no,the gradients of agreement scale is a subtlesolution to encourage greater participationin the decisions, as well as more reliablecommitment to implementation."My vision is that we extend the decision-making and the analytics across the entireorganisation, from the operators in the fieldto the management board room; that every-one can get a vision of how the operationworks, how it can be optimised and thateveryone can be involved in making theplans that implement efficient and eco-nomic production,” said Mr Fox.

One way to implement this vision would beto develop a collaborative business processfor production forecasting, together withserver-hosted software which whole assetteams can use..

A production planning process

Mr Fox presented a production planningprocess based on a cycle of continuous im-provement, which integrates expertiseacross the organisation from field operatorsto senior managers.

This process combines and integrates sev-eral proven techniques: ‘Produce the Limit’workshops, multi-scenario planning, partic-ipative decision-making, delegation of re-sponsibility to well-informed operators andquantitative ‘After Action Reviews’.

With web enabled software, non-specialistscan access analytics and modelling systemsremotely. This way they can make a bettercontribution to the planning process andthey can even experiment with their ownideas.

Analytics and optimisation together

To demonstrate that there is high, incre-mental value accessible from analytics andoptimisation, Mr Fox presented a casestudy on the optimisation of gas lifted oil

wells.

Faced with uncertainty in the future rate ofgas lift supply, optimised field-wide solu-tions must be precomputed for a range ofscenarios. When compared with the base case (ofwhat would be expected from a skilled andexperienced operator) total oil productionis significantly increased by using an

advanced optimisation algorithm on numer-ous gas lift performance curves (obtainedby analytics).

DynamicForecaster

DynamicForecaster is a multi-user web ap-plication for collaboration on both produc-tion analytics and forecasting, developedby Mr Fox’s company, 1234most.

It helps asset managers in upstream oil andgas companies who want to improve the fi-nancial return on investment, by using theinsights from multiple users.

Asset teams can now make quicker andmore varied analyses of past and future

production, leading to better quality deci-sions and implementation. Further informa-tion is online at dynamicforecaster.com

View Mr Fox’s video and download slidesatwww.digitalenergyjournal.com/video/1619.aspx

DynamicForecaster 1234most Ltd

Operations coordination

meetings

workshop

Account for shortfalls and note

opportunities

Review and commit to plans

and forecasts

Collaborative analytics and forecasting

Work the Plan Use initiative

Review and Learn from the past (use data)

Plan and forecast multiple scenarios

Manage opportunities and risks

Identify Design

Execute Adjust

6

Teradata – how to get value from analyticsOther industries have managed to get value from analytics – it is time the oil and gas industry followed, saidTeradata’s oil and gas practise lead Dr Duncan Irving



In the oil and gas industry it is typical for data tobe handed from one expert to another along theworkflow, with nothing captured about where thedata has come from and whether it can be trusted.

“By the time it has got to the person who needsto make the decisions it’s got to PowerPoint, andthere's no understanding of the algorithms (usedto create it) or any sense of data lineage – time iswasted re-establishing trust in the data”, saidDuncan Irving, principal consultant, Teradata Oiland Gas Team.

“This is the opposite of pretty much any other in-dustry Teradata works in.”

Many industries operate on much thinner mar-gins than the oil and gas sector, but they stillthought it was worth spending money on analyt-ics, he said.

Mr Irving’s goal is “trying to explain to the oilindustry what this could do for them,” he said.

Analytics systems should “become a more robustand significant part of the IT infrastructure in theupstream domain.”

Oil companies are attaching sensors all over theplatform which generate data for the life of thefield, but “we are not getting maximum valuefrom it,” he said.

Many companies keep data siloed in different de-partments and domains, with very little data com-munication between them, and this makes itharder to implement new processes and work-

flows at scale, he said. “You have to understandthe challenges in your own architecture,” he said.

You could say that oil and gas companies have 3types of data – transactional, technical and sci-entific.

Transactional data is stuff in the Enterprise Re-source Planning (ERP) systems, SAP systems,operations and finance data. Technical data couldinclude designs of casing shoes and well trajec-tories. Scientific data can include flow simula-tions and seismic data.

All of this data “comes in different flavours,” hesaid.

Unconventionals and data

The North American unconventionals businesshas done much more with data.

For example, ConocoPhillips used analytics towork out the best way to reduce the number ofshut-ins (when drilling needs to be stopped, forexample due to high pressure gas entering thewell).

They managed to reduce the number of shut-insby 90 per cent, Mr Irving said.

ConocoPhillips is also using analytics to workout how to optimise maintenance on wells andtank emptying, he said.

It has improved its logistics, building on workdone by a US parcel carrier, which worked outthat (for example) it may be faster to do threeright turns around the block than one left turn (theUS drives on the right). They found certain rout-ings led to less wear on the vehicles.

The company has been tackling questions like,what is the optimum well spacing (how manywells per mile) - and what is the best way to buildthe substations, roads and other infrastructure.

The company uses analytics as a continual guideto operations, he said. It brings together an inte-grated view of “what is the most effective wayto develop my fields,” keeping OPEX as low aspossible.

The unconventional oil and gas business was ba-sically a re-invention of the oil and gas industry,which meant an opportunity to bring in manynew processes. The unconventionals business“looks a lot more like a factory than what we

have offshore,” he said.

Retail, manufacturing and refineries

Consider that Walmart has a supply chain sotightly integrated, that by the time you’ve loadedyour shopping into your car, replacement itemshave been loaded onto a truck at the depot.“That's how integrated their supply chain is,” hesaid.

Walmart has an integrated understanding of howto access the biggest ‘wallet share’ of a neigh-bourhood, with different prices, promotions andmarketing. “They understand the environment inwhich they are operating.”

It also sees its business as selling shelf space tosuppliers, as much as it is about selling every-thing to consumers, and uses analytics to max-imise what it can do with its shelf space.

Looking at car manufacturing, Daimler is doing190,000 types of component failure analysis akinto ‘decline curve analysis’ work every nightbased on data from the production line, to workout if there is something which should betweaked. “They understand the total cost andprofitability of every piece of plant on the pro-duction line,” he said.

Oil refineries are also getting good with analytics.At one refinery, the refinery managers have sattogether with the traders to design a mutual dash-board, so the products being produced can be ad-justed according to near real-time changes in themarkets.

Sentient

Mr Irving uses the word ‘sentient’ to describewhere he thinks oil and gas companies need togo.“This is a bit of a hard concept”, it basicallymeans to have a better feeling about what is hap-pening,” he said.

This is a “new style of industry, you have organicsharing of data, you have organic sharing of ex-perience,” he said. “You have infrastructurewhich allows everyone to add the value theyneed to using tools they are familiar with.”

If this was happening with offshore oil and gas,all the various technical domains should havemuch better insight on how their equipment is

7Digital Energy Journal - Special report, How to use Analytics to improve Production, Sept 29 2015


running today, what it is contributing to the busi-ness bottom line, how it is operating togetherwith other facilities, what should be developednext, what rewards are available, and does thislink together with understanding of the subsur-face.

People could quickly answer questions like, ‘howto optimise the scheduling of maintenance tasks,’‘which bit of the reservoir to drill into next,’ ‘arewe depressurising the reservoir too early.’

Operational areas

It gets much tougher implementing collaborationtools and analytics for operational areas of the oiland gas business, compared to strategic depart-ments, he said. In operational areas, people areunder constraints of cost and time, which makeimplementing software harder.

There is also more need to build bespoke soft-ware tools, because everybody works differently,and every team has a different mix of skill sets.For example, one company might want a special-ist software tool for monitoring a reservoir usinglive data from permanent reservoir monitoringequipment, he said.

Discovery analytics

The “gateway” class of analytics is ‘discovery’,which basically means looking through the datafor new insights that potentially could be put ontoan operational footing.

As an example, consider that auction websiteeBay has a large staff of data scientists continu-ally probing its live database to try to find pat-terns. About 60 per cent of the workload on itsmaster database is from internal data scientists,he said. As a result, they might discover that oneuser prefers to click on red buttons rather than

green ones, and only show them red buttons fromnow on.

In a similar spirit, Teradata has given a data sci-ence project to a Petroleum geoscience MSc stu-dent at the University of Manchester, to try tofind overlooked stratigraphic relationships bydoing analytics on a large amount of wells data,he said. This discovered badly interpreted faciesand unseen hot shales, both leading to a re-ap-praisal of the entire basin.

Enabling data management

Teradata sees its role as ‘enabling data manage-ment’ – or helping the people who get very closeto the data to help the company do more.

Many of the people employed by data managersare highly qualified, with PhDs in geology ordata physics. “They got into data managementbecause they were the ones who cared the most,”he said. “Once you get tarnished with that brushit is very hard to get away from it.”

“But it is a waste of talent, having someone thatclever who is responsible for knowing it’s on thisor that file system, instead of enabling value tobe extracted with the data.”

“That person has to make sure all the metadataand master data. Someone has to care enough todo that and be knowledgeable about it.”

The ‘big data’ label is not very helpful, he said.“To me, big data is anything where you take indata too quickly for your systems and processesto deal with. If your systems are Kingdom suiteand a load of spreadsheets to run a brown field -you've probably got a big data problem. If youneed to extract value from SCADA in your his-torians, and SAS analyses on your SAP logisticsreports across a regions assets, then that's defi-nitely a big data problem.”

Looking at subsurface data, managing 2D alonecan be a ‘very difficult computational problem,”when you get to 3D or 4D (with time) it gets dif-ficult.

You have to find a way to make this data avail-able for the life of the field, not just lock it awayin your favourite software – and you have tomanage security and safety.

SQL

Usually the best language for analytics is SQL,because that is the language which business peo-ple’s systems use. “If you want to show the CFO something - thathas to be something he understands – ideally afew numbers (small data) and a graph. That’sprobably the thing he's already got - and it’s prob-ably powered by SQL.”IT staff in different domains speak different lan-guages. IT staff serving the business tend to useStructured Query Language (SQL) for interro-gating databases. Engineers like languages likeMatlab, and some scientific people do other lan-guages. “I was talking to seismic people yester-day, they kept saying 'we do this in FORTRAN,”he said.

The challenge is to pull all of these IT domainstogether under one umbrella for the CFO’s inte-grated “widget”. It is data-driven analytics whichenable these new kinds of transformation insightand as other industries have seen, complement(rather than replace) the experience of the deci-sion makers who have access to such largeamounts of data and computational power.

Download Duncan's slides athttp://www.digitalenergyjournal.com/event/e66da.aspx

Basin re-appraisal in New ZealandThe MSc project at the University of Manchester was designed to assess the applicability of standard signal processing functions to well logs. 2500logs from over 150 wells in New Zealand’s Taranaki basin were loaded into a Teradata Aster database along with their interpreted formation logs.The project started by taking well logs from a ‘few hundred’ wells in the basin, putting them together, and ‘cleansing’ the data (removing obviously er-roneous data).It also did some manual work to simplify the text descriptions of the logs, reducing the 800+ words used to describe the formations and the facies to92 ‘official words’. By simple data modelling, the data could be analysed by well, by depth, or by formation, to get a better understanding of the basin.“Symbolic aggregate approximation” or SAX, more usually used for speech recognition was used to split data up into various bins based on charac-teristics. This was then used to identify 29 possible ‘hot shales’ (in a few seconds) that had been overlooked in the reports, Dr Irving said.The next task was more subtle, to try to see if it was possible to tell the difference between “sandstones and mudstones interbedded” and “sandstonesand siltstones interbedded”.Here, “you’re doing some pattern recognition which is a lot more subtle across 4 logs at the same time,” he said. This requires a lot more parallel computer processing, saying “go and find the things here which look like this,” he said.The computer matched a lot of patterns – so then the next question is whether the geological stories behind them fit. On further investigation, it wasclear that formations had been mis-assigned by performing the “spot the difference” analyses.A lot of insight and experience was gained from a simple six-week study for a fantastic MSc project.



EnginSoft – quality checking your reservoirmodelsHaving a centralised platform for data management can be very helpful for reservoir engineers to do qualitychecks of reservoir simulation models and perform analytical analysis, said Oludare Elebiju of EnginSoft

“If you have a single platform where youanalyse post simulation [reservoir] models, andintegrate all your data (reservoir, wells, pres-sures and rates), you could probably save a lotof time,” said Oludare Elebiju, consultantreservoir engineer with EnginSoft.

“You would be in a better position to maketimely decision,” he said, speaking at the Dig-ital Energy Journal Sept 29 Aberdeen confer-ence, “Using Analytics to ImproveProduction.”

With a central data repository, it is easy to cre-ate workflows that allow for various tasks to beperformed without compromising the qualityof the actual data, he said.

This approach operates within a single environ-ment integrating numerical and analyticalanalysis and comparing the results, for examplenumerical vs. analytical production forecast-ing.

A centralised system also helps to reduce theamount of time spent on quality control of thesame data.

This will enable several tasks to be performedby different engineers if the data quality has notbeen compromised, he said.

This will also reduce the amount of time reser-voir engineers spend on data management preand post simulation.

Typically, if a reservoir engineer has two weeksto do a job, which requires obtaining data fromdifferent sources and applications, it might takea great deal of time in evaluating the quality ofthe data (QA/QC).

If about 50-60 per cent of the total time is spenton data quality check and control, then the en-gineer might not have sufficient time to carryout the ‘real’ job, he said.

The reservoir engineer is expected to carry outdesignated tasks and make recommendationsbased on results achieved. The quality of theresult can be directly matched to the quality ofdata the engineer has to work with.

If the quality of data is poor and sufficient timeis not invested in QA/QC before embarking onspecific tasks (reservoir modelling/simulation,history matching and forecasting), then resultsand standards are easily flawed.

So a tool that provides an efficient platform toperform effective reservoir data and modelQC/QA will contribute a significant value inoptimising processes, reducing data QA/QCtime and maximising simulation, interpretationand reporting time, he said.

EnginSoft provides Computer-Aided Engi-neering (CAE) and intelligent Digital Prototyp-ing (iDP) tools, as well as process simulationand design chain optimisation.

EnginSoft is distributors of ‘Kraken’, a reser-voir data manipulation and integration plat-form, which can be used as a central repositoryfor all reservoir data, and used to support struc-tured workflows.

The software is developed by ESSS, a SouthAmerican engineering simulation solutionscompany, and was originally developed for alarge well-known Brazilian oil and gas com-pany.

Data accuracy

Managing reservoir data centrally will alsomake it easier to ensure accuracy. It is commonfor oil companies to manage production datausing spreadsheets only, and then pass thespreadsheets onto the reservoir engineers, hesaid

“For example, you might be working on anasset where you receive production data fromthe field (weekly/monthly) after allocations aredone. It is possible to spot inconsistent trendsin the data (well rates, pressures, well potential,etc.) which are mostly due to allocation errors.Such errors can be minor or significant andmay require time to correct.

Decision making

Kraken can make it easier for reservoir engi-neers to understand reservoir performance.“You want to know, how I am doing in termsof production, injection, over time.”

For example, “you want to see your productionby well, block, field. You want to see the wateroil ratios by well, by block, by field.”

Any decisions, for example about maintenance,water flooding and EOR, can be made usingthe full range of available reservoir data.

Kraken can also support structured workflowsand better multidisciplinary collaboration.

Optimising

Kraken can help engineers create customisedprocess templates that can easily be docu-mented and reused from time to time for im-plementing different process workflows. Thishelps to optimise the processes to ensure thatthe same project is done much more quicklyand efficiently, enabling better and timely de-cision making.

Companies have different approaches to reser-voir data analytics, architecture and integrationto specific reservoir/petroleum engineeringworkflows. It is important to critically evaluatethe integration process to ensure that resourcetime is optimally utilised during project execu-tion so that cost is reduced and value is max-imised, Mr Elebiju said.

Watch Oludare's talk on video and downloadslides athttp://www.digitalenergyjournal.com/video/1621.aspx

9

Data Lab – letting academia work on yourdata problemsData Lab is a Scottish government funded body which will help connect academic data experts withindustry’s data problems



The Scottish Funding Council, a Scottish bodywhich promotes higher education, has set up“The Data Lab”, an organisation to connect ac-ademic data scientists with the oil and gas andfinancial industries, solving industry problems.

Data Lab is based and administered from theUniversity of Edinburgh, but it has an office inAberdeen, based at Robert Gordon University(RGU). It works together with Scottish Enter-prise.

There is a similar organisation serving health-care, called the Digital Health Institute, alsofunded by the Scottish Funding Council.

Data Lab welcomes data challenges from boththe ‘private and public sector,’ he said DuncanHart, business development executive with theData Lab, based in Aberdeen.

“We try to take a problem and go out to univer-sities and say how can you, with academic part-ners, come up with an innovative solution thatwill make a difference,” he said. “We’re usingthe power of the academic minds to create mod-els and create algorithms.”

As an example, Data Lab connected oil majorREPSOL with a team at the University of Glas-gow.

“We can be involved in medium scale and largescale collaborative projects,” he said. “Thefunding is dependent on the project.”

The projects are usually short term, not four orfive years, which is typical for academic re-search projects. “We're looking at real world re-sults that can deliver real world value,” he said.

A further aim is to develop data science capa-bility in Scotland, and also develop technologywhich Scotland can export, he said.

“Open innovation” is a key philosophy behindData Lab. “No company can afford to developthings on their own anymore, it’s an impossibil-ity,” he said.

“Oil and gas companies want to maximise up-time and minimise downtime. If data analyticscan help improve production by 6-12 per cent,there’s a lot of people who will be very happy.”

“If you can take 1 per cent, 2 per cent off yourproduction cost, that's a significant saving. Ifyou can take two weeks out of your shutdownbecause of a decision that is around data analyt-ics, that is a lot of saving.”

Data science can also make it easier for peopleto access the right data, and data which is veri-fied as correct.

Value

Data Lab can also help work out the value of adata project. “We can help you scope the idea,work out the end result. That's better than tryingto rush headlong into a data science project,” hesaid.

Data projects can go wrong if you ask the wrongquestions, select the wrong users, or have dis-agreements, he said. They can also go wrong ifyou have data silos, like engineering depart-ments not talking to reservoir staff.

“There can be management resistance. Do man-agement understand what they are trying to do?If nothing happens at the top that affects the bot-tom.”

“There's a lot of academics who live and breathethis.

Download Duncan's slides athttp://www.digitalenergyjournal.com/event/e66da.aspx



Halliburton – data, models, workflow andchange managementTo get a production improvement system working requires a mix of data, models, workflow and changemanagement. Sergio Sama Rubio, Industry Solutions Advisor with Halliburton, explained how to do it

The four key components of devel-oping a production optimisation sys-tem are data, models, workflows andchange management, said SergioSama Rubio, Industry Solutions Advisor with Halliburton.

“If we only take care of three of them it doesn'twork.”

The ultimate aim is ‘best in class operational per-formance’,” he said. In most cases, companiescould see a one per cent improvement in produc-tion without additional capital investment.

If they could get a better understanding of thebehaviour of the processing facility, so theycould navigate the constraints better, then “wecould be producing more.”

Consider the US refining industry. Between1990 and 2000, the overall capacity of refiningstayed roughly constant, but the utilisation in-creased from 85 to 95 per cent.

“Refineries in the 90s got better and better atusing existing installed capacity,” he said. “Ithink it’s fair to recognise that these guys man-aged to squeeze the last drop of capacity.

“There is a lot we can learn from refineries. Tryto extract the lessons learned and apply the samelessons to our world.”

Data

Starting with data, there are many different as-pects of data to consider – the quantity of data,is it covering the right aspects, time dimension(for example data for a historian), space dimen-sion (which part of the asset or fields does it re-late to), is it the right type of data.

“There are significant savings in from makingsure we really capturing the data that will helpus act better,” he said. This means “capturingonly the data that is relevant for my decisionmaking, not capturing every single pressure,temperate and flowrate.”

We should ask, “Do I have the data, when I needit, in the time I have allocated?” he said.

Sometimes people in your organisations objectto a way of working, saying ‘you don’t touch mydata’, he said. “This could be solved with clearguidelines, for example saying ‘you can viewmy data but you cannot modify my data.’”

Data can be managed with a ‘federated’ system,

with the applications processing the data work-ing directly with the source data, or with an in-termediate ‘operational data store’, which all ofthe data is copied to, and made available to thevarious applications from there.

Models and workflows

Workflows mean having a structured way ofdoing certain tasks which are commonly done,such as well integrity management, steam flood,artificial lift optimisation, well performance, reg-ulatory production reporting.

Some workflows are more complex, such asfield development planning and artificial liftplanning.

These workflows are about understanding whathas happened in the past or what is happeningtoday, or what will happen in the future.

By developing a structured workflow, you canmake sure work is done in a standardised way,and also encapsulate the knowledge of the com-pany’s top experts, including people who haveleft the company.

A workflow will say, for example, after gentle-man A has done a certain thing it goes to gentle-man B.

Workflow is sometimes more structured or for-malised than others, For example it could be asystem where you decide to send a document toa colleague to see if she has anything to add.

A good workflow provides “a natural division oflabour, things that people do well and thingscomputers do well,” he said.

“Humans are good at strategy, designs andchoices, objectives and decisions. Machines aregood at repetitive tasks, running models, opti-mising, and training.”

“Humans are able to see trends, see in a morecreative manner. Computers will follow algo-rithms,” he said.

To take an example from the use of computersin chess, “every chess player will agree that thecombination of a human plus a chess computerprogram are absolutely unbeatable,” he said. “Itgives birth to a new discipline - advanced chess.”

Change management

The final step is change management, getting to

the point where people are comfortable and inthe habit of using the systems. A failure here hascaused the failure of many projects.

It is often helpful to do a ‘proof of concept’ first,demonstrating to people the value that the newsystem can provide, before starting to implementit, he said.

For example, you could build a workflow whichis used on just 10 wells. If the business can seea gain from it, the scope can be extended.

”Build a business case to convince managementthat this system will be able to increase uptimeby 3 per cent,” he said.

When implementing workflows it might be eas-iest to start with regulatory compliance, becausethese are tasks that the company needs to do.

Or you can look at areas critical to continued op-erations. “You say, ‘Mr CEO, do you want thecompany to keep operating? Then we need awell integrity system.’ But do it in a way that itis the first step in a longer journey.”

You might also want to continually revise yourworkflows. One Halliburton project has beenthrough nine iterations of workflow since thefirst release in July 2008.

The systems need to be able to be used by peoplewith a wide range of technical competence, hesaid. “We cannot expect that everybody in ourcompany has a PhD in geology or petroleum en-gineering.”

“We need to make these tools available for op-erators and engineers,” he said. “They have dif-ferent needs.”

A common problem with change management isorganisational fatigue. “Many companies launchinitiatives, they say, ‘we need to start with thedata, we are going to do it in the house in aproper way.’ By the time the three years havegone past, management has changed priorities.”

“Some companies are notorious in that the G+Gpeople, production people and surface peopledon't talk to each other,” he said. “This is one ofthe bottlenecks in our industry. Don’t let siloesget in the way.

Watch Sergios's talk on video and downloadslides athttp://www.digitalenergyjournal.com/video/1732.aspx



Jonathan Guthrie – how to get past “no”Senior management in oil and gas companies often avoid interesting analytics projects because they aremore comfortable doing things the way they always did them, said energy consultant Jonathan Guthrie

There is a lot of talk in Aberdeenabout trying to be more efficient,but “to be honest, I’m not seeing alot of new efficiencies being intro-duced,” said energy consultantJonathan Guthrie.

He was speaking at the Digital Energy Journalforum in Aberdeen on September 29 2015,“Using Analytics to Improve Production”

“I’ve been working in oil and gas since 1981,this is the sixth downturn I’ve seen, and behav-iours have not changed. It is so disappointing.Every other industry gets better, but this onecarries on, they batten down the hatches andcarry on with the same old story and excuses.”

In this climate senior management often justwant to keep the company on a conservativecourse, saying no, killing ideas, and demandingonerous reporting through bureaucratic systems,he said. “There isn’t anyone in charge of newgrowth ideas.”

The mantra seems to be, “let's give the appear-ance we want to improve, but let’s just continuethe same old workflow, the same old inefficientprocesses we've always been doing, it feelssafe.”

Mr Guthrie has looked at data from many E&Pcompanies in Aberdeen, and found that “onecompany is no better than another in how theycollect, manage and analyse data.”

As an example, Mr Guthrie recently worked outhow he could help an operator deliver £15m ofvalue to the business, which he could implementwith a few weeks of consulting work. But theoperator’s procurement staff insisted that com-pany rules were that all consultants should cuttheir rates by 10 per cent.

“I said, ‘no, why should I,’” Mr Guthrie said. “Ihad already offered them my lowest rate. Whyshould I as a small independent take a cut for ashort-term service engagement just becauseeverybody else is? They were only thinking ofcost and were blind to value.”

So as a piece of advice to the industry, “if youwant to get analytics going, you have to avoidthese people, you have to find the person withthe vision and enthusiasm to change behavioursand do something different,” he said.

Meanwhile most data is just treated like TV, thedata comes in and people look at it, it getsstored, but never analysed. It would be better ifthe data was analysed and the results used towork out ways to improve.

The talk about ‘big data’ is more hype, MrGuthrie said. E&P companies have always spentlarge amounts of money gathering data, thepoint is that they do nothing with it.

Instead of just talking about optimising ‘invest-ment dollars per BOE of production,’ compa-nies should be asking broader questions, suchas, ‘where is the best place to drill a new well,’or ‘are you using the best completion technol-ogy,’ he said.

“A lot of work we do is showing people whatthey don't know they don't know, or tell themabout things that aren't happening, which theythought were happening,” he said.

Operators still waste large amounts time and ef-fort requiring highly skilled engineers to spendtheir time ‘QCing’ data, or extracting andcrunching data in spreadsheets. “How sad isthat?” he asked.

Another common problem is that people onlyfocus on ‘their own’ data from a small segmentof operations. You arrive at a situation where‘their’ data says everything is fine, but theplant/equipment is still failing.

Start small with analytics

On the software side, you can start small, takingdata from different systems, mash it together inan analytic tool, quickly visualise what is goingon, get a better understanding of the flows andlook for actionable insights.

“Data is locked in our applications, experienceand knowledge are locked in the mind of ouremployees, analytics is the key to unlocking thevalue here” he said.

Emotions

Data attracts a lot of odd emotions.

“Many people despair that there’s too muchdata, and it is of poor quality and it needs to befixed before it can be used. The first step is tojust look at a small piece of it. You can't eat thewhole elephant at once.”

There is a lot of apathy around data. People say,“I've collected all this data and nobody everlooks at it, why should I bother,” he said.

There is also fear, with people worried aboutstating ownership of their data, in case they getblamed if someone else finds an error in it.

Another emotion is control, where the dataowner declares a right to tell others how the datashould be used and who is allowed to access it.“These are the barriers that need to be brokendown,” he said.

IT department

The IT department can prove a barrier to ana-lytics. “They have security, they have policies,they will decide how data will be stored,” hesaid.

They want to provide software in the traditionalway, ‘Give me a spec, tell me what you needand I'll deliver it to you,’ he said.

But the business user doesn’t usually know whatthey need, to be proactive they need unrestrictedaccess to the data and self-service tools to ex-plore data of all types, to discover insights anddeliver actions.

History and now

The real challenge for oil and gas experts is touse historical data together with the streamingdata about what is happening right now.

There are analytics tools which work throughhistorical data and real time data together tomake predictions, he said.

For example the analytics can say, if event ‘A’,‘B’ and ‘C’ happen, then there’s a failure on itsway.

The computer can automatically analyse the in-coming data stream, and see A and B have hap-pened and C is close to happening and raise thealarm.

The computer system can also automaticallyhelp you check you have the right spares avail-able, how much unplanned downtime you have,what are the real-time trends and how this is im-pacting profitability.



Tools

For analysis, Mr Guthrie has used Tableau, Qlikand TIBCO Spotfire (his analytic tool of choice,he was previously employed by Spotfire).

Tools like this can work simultaneously on datafrom multiple sources, without having to bringthe data into a central warehouse, he said.

If you want to do modelling on the data, then itis easier if the data is in one place, but “youdon’t have to start there,” he said.

When optimising, you want to get as big a pic-ture as you can of the process.

“The great thing about analytics is, it helps peo-ple very quickly visualise what is happeningover time,” he said. “You can drill and completebetter wells, optimise processes and routes.We're constantly searching for what we don'tknow we don’t know.”

As an example, there is discussion going onabout taking old seismic shoot data and re-pro-cessing it using new techniques.

Looking at Spotfire in particular, most oil com-panies have corporate licenses, which meansthat any company employee can use it. “For anengineer typically you just have to get a copyon your desktop to connect to the data and startanalysing.“

Oilfield services companies could use analyticsfor root cause analysis, drilling and completionoptimisation, analysis of revenue and profitabil-ity across different products and regions, HSEincident tracking, maintenance schedule optimi-sation and project management, he said.

Wood McKenzie currently delivers a tool to itscustomers using Spotfire to do decline curveanalysis on the production data history frommore than 1 million wells in North America, hesaid.

Out of the box Spotfire is a great tool both forquickly building simple dashboards, and fordoing complex data science, he said.

Middle East

Mr Guthrie was engaged over summer 2015working for a Middle East oil company, wherehe used Spotfire to help implement new analyticworkflows, to monitor and analyse how allo-cated production from 2,500 wells was declin-ing, the effects on the gathering centres andwells at risk.

Spotfire was also used to optimise rig andworkover activities.

“By delivering visibility of rig resources andworkover activities with Spotfire the team wereable to accelerate the delivery of 120,000 bopdof gain across all assets in 6 weeks just by pri-oritising and optimising the activities and re-sources,” he said.

The business environment for these companiesis quickly changing, as fields start to decline andwater management becomes a larger factor, theyneed to analyse more and more data to be pro-active in the decisions they are making for thefuture life of the reservoirs, fields and the assetsto deliver the production.

Case study

As one example, the company is combining

data from spreadsheets, files, Halliburton’s“OpenWorks’ and Schlumberger’s “Finder”data management tool to build new workflowsto support the decision making processes.

Spotfire acts as an ‘integration layer’, which is“nothing clever”, just bringing data into theanalysis from a number of underlying datasources/views, he said.

In ‘Eagle Ford’ multiple data sources providethe feed into a Teradata data warehouse, whichdoes the data governance work, and then the ad-vanced analytics is delivered using Spotfire.

The company does a wide range of analytics in-cluding predicting failures, optimisation, look-ing at well settings, recommending actions,electrical submersible pump monitoring.

Another customer is looking at Spotfire data tomonitor and analyse Electrical SubmersiblePump (ESP) failures and compare suppliers, hesaid.

“This is where I see the biggest value add, ana-lytics is going to help identify the right tech-nologies, the right tools for the job, and showwhat works what doesn't,” he said.

So analytics should help you get to first oilfaster, improve your production capability andbe more efficient, he said.

Watch Jonathan's talk on video and downloadslides athttp://www.digitalenergyjournal.com/video/1713.aspx

List of attendees 'How to use Analytics to improve Production' Aberdeen, Sept 29, 2015

What did you enjoy most about the event?

“ “The honesty of the talks with re-gard to the difficulties of imple-menting up-to-date analyticstechniques & associated sys-temsJohn Greenhough, U of Edinburgh

Learning experiencesof other companies.Understanding whatdoes, and (more im-portantly) what doesnot, work in practice

“ Breadth of expertise andknowledge at the event. Greatnetworking. Great value formoneyAlexander Petrie, Director, Left FieldAssociates Scotland Ltd

”” ”



“ “Interesting discussions withother participants

A morning ofpresentations +lunch is an efficient formatfor learning andnetworking.

“ The general presenta-tion quality was good,and the vendors didwell in providing educa-tional material whichwasn't solely about flog-ging products.

”” ”

“ Great insightsfrom a range ofvendors and oilcompanies.

”

Graeme Coghill, Managing Director, Aberdeen Group LtdSteve Eacott, Industry Solutions Lead,AccentureScott Petrie, Director & Principal Tech-nology Consultant, Analysis Logic Ltd.Ehud Reiter, CTO, ArriaGatsbyd Forsyth, Uk Operate Sub/Surf & PO Mng, BG GroupDan Kuszpit, Principal Consultant, Carlton Resource SolutionsOludare Elebiju, Consultant ReservoirEngineer, EnginsoftStephen Turner, Sales Manager, EnginsoftVinicius Girardi Silva, Business Development Manager, ESSSAvinga Pallangyo, Finance Admin, Finding PetroleumJonathan Guthrie, GreyCloud Limited - Analytics for Energy

Sergio Sama Rubio, Industry SolutionsAdvisor, HalliburtonMike Scott, Senior Managing Consultant,Intelligent Operations, HalliburtonGavin Robinson, Business DevelopmentManager, LandmarkAlexander Petrie, Director, Left Field Associates Scotland LtdAzuka Akeze, Reservoir Engineer,Maersk OilCindy Wood, Senior Consultant, New Digital BusinessSteven Taylor, Production Reporting Analyst, Nexen Petroleum (UK) LtdNick Pashley, Business Applications Analyst, Nexen Petroleum (UK) LtdChris May, Systems Director, OceaneeringTom Fox, Director, 1234mostKurt Prendergast, General Manager,Palantir Solutions

Deborah Murison, Production TechnicalSales, SchlumbergerSteve Harrison, project manager, Scottish EnterpriseFiona McDonald, Account Manager,Scottish EnterpriseVivek Jaiswal, Reservoir Engineer, Senergy EnergyPeter Brownsmith, Business Development Lead, Tata ConsultancyServices LtdDuncan Irving, Oil and Gas Practice Lead (EMEA/APJ), TeradataDuncan Hart, Business Developement,The Data LAbJohn Greenhough, Research Associate,University of EdinburghWim der Kinderen, Consultant, Well Optimation LtdAnagha Vellani, Business Analyst, Wipro UK Limited

Get the most from your investments in people and data. Timely, trusted, granular and integrated data on an open platform transforms the way oil and gas operators manage their resources.

What would you do if you knew?™

Discover more oil & gas industry solutions

Explore – More of Your Data. Drill – Deeper Into Your Data. Produce – Value From Your Data

teradata.com/oilandgas

how to use analytics to improve...

Documents