Quantifying uncertainty through the Roxar ‘Big Loop’

1 30 DEW JOURNAL February calculating of uncertaintywithin the reservoir sgeological structure is one of themost important factors foridentifying both in-place andrecoverable hydrocarbons in Asiatoday. Yet, while its significance iswell-known to practitioners, toooften such uncertainty time and resourceconstraints may often be the reasonfor this, in many cases the structureof the reservoir modeling workflowand its integration betweendifferent domains are reasons sequential focus of manyreservoir modeling workflows andreservoir simulation s limitations(where changing the geometry ofthe simulation grid is almost thesame as starting again)

2 MeansQuantifying uncertaintythrough the Roxar Big Loop This article has described how new technologies can help Asian asset teams achievea closer integration throughout the seismic to simulation workflow - between theclassical domains of geophysics, geology and reservoir foundation for this approach is the Roxar Big Loop workflow and it s newevergreen, ensemble workflows that can be refined, updated and used to test variousassumptions, as modelers and engineers together move through the different decisiongates. The result for Asian operators will be improved information for field developmentplanning and well placement and better investment returns from reservoirs - crucial ata time of low oil & gas ModelingEirik GudmundsenRegional Director Asia Pacific, Roxar Software Solutions,Emerson Process Managementthat that the structural model isoften locked in as soon as thestructure has been defined.

3 Thisrules out future updates evenwhen fast changing data arrives and can lead to a higher risk ofgeological inconsistenciesFebruary 2016 DEW JOURNAL in the simulationmodel(s).Linked to this sequentialapproach is the lack of integrationbetween different disciplines anddomains, where uncertainties aresustained within each domainrather than being captured andpropagated throughout the the current industrybackdrop of low oil prices and aneed to increase investmentreturns, this article will argue thatthere is a need for improving theapproach to reservoir modeling inAsia today - both to improveefficiency and to make article focuses on theimportance of the Roxar Big Loop approach to reservoir modeling andits three main consist of the evergreennature of the workflow throughwhich customers can keep theirreservoir models up to date.

4 Theensemble workflow where thesolution allows risk to be quantifiedacross the portfolio anduncertainties propagated fromwhere they belong to where theymatter; and the collaborativeapproach where - throughimplementing the Big Loop -customers can improve theefficiencies and integration of theirdifferent result will be a repeatableand automated workflow that iseasy to update, where uncertaintiescan be added at any level, and thatgenerates crucial information forfuture field development andplanning decisions in the Velocity &Structural ModelOne key part of the new workflow isthe automated updating of thestructural and velocity model basedon a Bayesian geostatisticalapproach to depth updating of reservoirmodels can often be costly andmanually intensive, particularlywhen including new well approaches wheredepth surfaces are conditioned tonew well observations can alsolead to unintended errors.

5 Such aszone pinch out and wells endingup in the wrong modeling and depthconversion also come with largeuncertainties, mainly becausethere is a lack of data to conditionto, particularly for estimatinganisotropy or directionaldependencies of seismic 1 illustrates how a newwell has been drilled and wherethere is a mismatch between thedepth information of the well andthe existing structure (a and b). Insuch cases, the traditionalhandling of this problem is tocondition the depth surfaces to thenew well observations (c).

6 This,however, can lead to unintendederrors (zone pinch out and wellsending up in the wrong zone) asalready 1d, however, shows analternative solution where the newdata is viewed in relation to all theexisting information, and implementing the Big Loop approach to reservoir modeling -customers can improve the efficienciesand integration of their different disciplinesReservoir Modeling32 DEW JOURNAL February changes are necessary toincorporate the new data in arobust way to ensure can, for example, be an updateof the velocity model, a localadjustment of the surfaces or acombination of both.

7 Depending onthe relative uncertainties allowingfor the technology advancesare leading to more efficient modelupdates of the reservoir structure( Stenerud et al. 2012)with thekey outcome being a robust andintegrated update of the velocity andstructural model through a reusingof the existing workflow (figures 1band d). This conditioningtechnique, combined with theexisting modeling framework,allows Asian operators to supportevergreen workflows consistentwith all the latest this in mind, Emerson hasdeveloped tools for the automatedincorporation of multiple structuralrealizations based on uncertaintiesin both the velocity model andseismic solution is based on aBayesian geostatistical approachto depth conversion (( et al.))

8 1991,Abrahamsen, 2005) that adds valueto existing velocity modelingpractices. The Bayesian approachcombines all information on timeinterpretation, depth observations,thicknesses, surface correlations,velocity and associateduncertainties to makegeostatistical predictions of thevelocity 2illustrates howgeostatistical approaches to depthconversion under uncertaintyproduces: a) a most likely model;b) an updated velocity model; andc) an uncertainty estimate of theprediction, given in one standarddeviation.

9 The prediction error (onestandard deviation) gives aquantification of how much thehorizons can shift and still beconsistent with all input implemented algorithmwithin the Emerson approachensures tight integration of thegeophysics time/velocity domainand depth domain, strengtheningmodel updates and uncertaintyworkflows. In addition, thealgorithm allows for the utilizationof depth information alonghorizontal wells for example,conditioning the horizons tospecific zones along the benefits will be adjusteddepth maps and a 3D grid thatmatches the new well data.

10 Thiscan be done quickly by adjustingthe grid to maps, or through a rerunof the whole gridding and propertymodeling workflows, depending onhow much impact the new well ispredicted to make. The automatedprocess also makes a most likelyvelocity model in sync withinterpretation in time, depthinformation and their Evergreen, EnsembleWorkflow AutomatedUncertainty AssessmentToo often in the past, there hasbeen a tendency for domain-specific working practices to takeprecedence over the commongoals of the modeling , uncertainties tend to besustained within each of thedomains.