TECHNOLOGY

Earth model assists Permian asset valuation Patrick J. Curth Rick Mauro James R. Courtier Scot Evans Gary B. Smallwood Co. Laredo Inc. Houston Tulsa

Laredo Petroleum Inc.’s Permian-Garden City as- basin’s multi-stacked horizontal targets available set is an unconventional resource play with more for development. than four potential stacked zones covering a 1,700 square-mile fairway in five counties in the Midland Garden City basin, . Early in the play’s evolution, Tulsa- Laredo has developed extensive acreage on the east based Laredo recognized the need to develop a side of the Midland basin with interest in more DRILLING & proprietary database along with a process to char- PRODUCTION than 350 sections representing 178,000 gross and acterize each targeted reservoir. 148,000 net acres with most concentrated in Glass- This article describes a multi-domain model cock and Reagan counties. Producing intervals to that Laredo, assisted by Halliburton Co., used to date include the vertical Wolfberry interval and define the potential for the Permian-Garden City acreage. the horizontal Wolfcamp shale (Upper, Middle, Lower), the The model represents an integrated workflow combining Cline shale, and Canyon formations. Additional horizontal geoscience and engineering data with multivariate statistics. targets include the shallower Spraberry, the Strawn, and the The process began with acquisition of high-quality data deeper Atoka-Bend-Woodford (ABW) zones. including 3D seismic, microseismic, cores, well completion With an average combined thickness of more than 5,000 and production histories, and petrophysical information. vertical ft for all of the targeted zones, the Midland basin is The data were then analyzed, processed, and incorporated unique among US shale plays. into a predictive three-dimensional (3D) model. The result, Early success with horizontal wells in both the Wolfcamp demonstrated here, is a tool used in the planning of devel- and Cline intervals encouraged Laredo Petroleum to build a opment wells to optimize initial production rates and esti- large technical database from which to pursue early efforts mated ultimate reserves and yield a better understanding of to increase production and execute a full drilling program. the complexities of a multi-zone stacked resource. All the targeted zones have flowed oil to the surface from off- setting vertical wells, but horizontal drilling and hydraulic Midland basin fracturing have made these objectives economically viable. Conventional production began in the Permian basin in the Keeping in mind the goal of having an economic program 1920s and now covers more than 86,000 square miles in that took advantage of what each stacked horizontal target West Texas and southeast New Mexico. Fig. 1 presents the had to offer, Laredo Petroleum understood that the number geologic architecture of the basin, showing the subdivision of horizontal wells to be drilled, the capital commitment it into the Delaware and Midland basins by the Central basin would take, and the associated operational considerations platform. Laredo Petroleum’s acreage in the eastern half of made integrated development planning mandatory. the Midland basin is high lighted in red. Conventional production in the Permian comes from sev- Earth-model program eral horizons ranging in age from Permian down to Ordo- Laredo Petroleum invested in extensive data capture over vician. Beginning in 2008, Laredo Petroleum targeted the its entire Garden City asset that included geophysical (3D more basinal source rock and tight carbonate reservoirs of seismic, gravity and magnetic data, and microseismic the Wolfcamp and Cline formations, using horizontal drill- surveys), logs (conventional openhole and dipole), cores ing and hydraulic fracturing. Fig. 2 illustrates the Midland (whole and sidewall), and well testing data (single zone and production tests). A key element PERMIAN BASIN STRUCTURES FIG. 1 of the Laredo Petroleum–Halliburton partnership is integrating these data Levelland Lubbock into a 3D geologic earth model, then Northwest using it to support decisions about shelf well spacing, lateral length, and hy- draulic fracturing design. New The program has two phases: Mexico • Phase 1 focuses on the initial coarser scale assessment of the overall Snyder Garden City area. Lamesa • Phase 2 focuses on a detailed pi- lot area identified in the first phase. Texas Sweetwater The goal is ultimately to drill the Big Spring best wells, as soon as possible, and de- termine the most efficient ways to ac- celerate that drilling for maximum net Eastern Midland Central basin shelf present value. platform Midland EM: Phase 1 basin The first phase confirmed previous Delaware work by Laredo Petroleum, focusing basin San Angelo on the Wolfcamp and Cline forma- tions in the Midland basin through attribute modeling of well and petro- Area physical data and using the extensive shown Ozona 3D seismic data Laredo Petroleum had arch acquired over Garden City. US Shef eld channel Results from Phase 1 also provided insights into well spacing and priori- Val Verde basin tization of leases based on subsurface modeling and dynamic simulations Marathon-Ouachita including integration of microseismic, fold belt Laredo leasehold and core data, produc-

Note: Not to scale z1006OGJdcu-z01 tion and history matching with dy- namic simulation of producing wells, and geomechanical properties and fracture modeling. MIDLAND BASIN: MULTISTACKED HORIZONTAL TARGETS FIG. 2 One of Laredo Petroleum’s goals in Phase 1 was to obtain an overview of the rock property heterogeneity in each Wolfcamp and Cline zone over much of the asset. This process identi- fied potential “sweet spots” based on such static-model indicators as hydro- carbon pore volume (HCPV) and geo- Clearfork Spraberry mechanical properties. Dean But a lack of production history Upper over the acreage base made direct cor- Wolfcamp Canyon relations to potentially indicative rock Middle Penn Wolfcamp properties difficult. Laredo Petroleum Cline and Halliburton recognized the need Lower Stawn Wolfcamp ABW for a specific pilot program in which z1006OGJdcu-z02 Fusselman better production data were available TECHNOLOGY

WOLFCAMP: HYDROCARBON PORE VOLUME VARIATION FIG. 3 get requirements. All the data Laredo Petroleum had accumu- lated indicated their properties had large resource potential, but further refinement of the earth model was needed to achieve early sequencing of drilling the highest potential acreage first and establishing the basis for the lowest cost-per-barrel unconven-

z1006OGJdcu-z03 tional development of the Gar- den City asset.

INITIAL EARTH-MODEL PLANNING AREA FIG. 4 EM: Phase 2 The second phase focused on a specific planning area (Fig. 4), which has served as the primary pilot area for expand-

Martin ing the scope of the earth model and statistically tying the Howard Forsan Howard Lomax Mitchell results to actual well production. Howard Phase 2’s focus has also been to provide a higher reso- Glasscock lution understanding of such reservoir attributes as brittle- Lees ness, total organic carbon, and HCPV based on actual well data and newly acquired and reprocessed full-wide azimuth seismic data. We analyzed more than 80 seismic, petrophys- ical, and engineering attributes to gain insight into those that would highlight the most productive intervals within each formation. Fig. 5 shows the general workflow for the Garden City earth-model process. Howard Glasscock Sterling A key feature of Phase 2 was to add production data and history for dynamic and add detail to the static model listed above. The goal was to correlate the patterns discerned from the static data with actual produc- tivity results and to use multivariate statistics to help de- velop a predictive model. Shale resource plays are character- ized by a wide variety of petrophysical properties both well Glasscock and seismically derived. Robust multivariate statistics indi- Reagan cate which properties affect production, inclusion of many of which did not at first seem intuitive. Fig. 5 shows the general inputs. Phase 2 - initial focus area Dynamic reservoir simulations then provide further un- derstanding of reservoir performance and then these his- tory-matched dynamic models refine both horizontal and vertical spacing in a multi-stacked, target-rich environment. Stiles The heterogeneity of the producing rocks from all the pro- spective Wolfcamp and Cline intervals is better understood Irion

Laredo leasehold Reagan by use of the earth model and allows Laredo Petroleum’s de- 10 miles velopment program to be based on lower drilling and oper-

z1006OGJdcu-z04 ating costs. Laredo Petroleum used the fully integrated earth mod- el developed in Phase 2 both in identifying overall “sweet to validate results. Fig. 3 illustrates an example of the varia- spots” (both vertically and horizontally) and in picking tion in HCPV for one Wolfcamp interval. landing points and the horizontal laterals. As Results from Phase 1 highlighted how the earth model Fig. 6 illustrates, the model, now tuned to integrate static process could assist in well planning to build potential field properties and production results, is used to improve lateral development scenarios and estimate corresponding bud- placement within a given reservoir and the corresponding reservoir rock to generate sufficient production. sufficient to generate rock reservoir touch enough productive will fracturing hoping hydraulic and steering without well the of landing practice industry to more common opposed as areas reservoir productive most the with contact maximizes well the that ensuring steering, geo real-time guides then Themodel point. landing ideal EARTH-MODEL WORKFLOW EARTH-MODEL RESERVOIRCONTACT horizontal spacing Petrophysical data Intrawell vertical, Seismic attributes Seismic horizons Completion data Landing points Production data Formation tops Lateral length (stay-in zone) Frac design, Geosteering selection spacing Inputs 4 3 1 2 5 Layer-based modeling utvral nlssWell planning, eldoptimization Multivariable analysis 1 4 2 petrophysical modeling , - 3 tual production history. production history. tual to ac wells horizontal existing the matched and modelled workflow and tools Cypher engineering Halliburton’s fined. re to be designs stimulation and well the allows properties Improved fracture design is currently under way. under The currently is design Improved fracture dynamic and static of both understanding A detailed Reservoir-properties model 5 Optimized wellbore Higher IPs,EURs geometries Results TECHNOLOGY FIG. 5 FIG. 6 - -

z1006OGJ-z08 z1006OGJdcu-z05 TECHNOLOGY

PRODUCTION CORRELATION FIG. 7 of productivity compared with actual data from existing pre-earth model 200 wells. To date, the model has been compared with actual results in more 175 Deliver highest EUR, than 30 horizontal wells with an aver- highest value wells age correlation coefficient for the four Wolfcamp and Cline intervals of 0.85 150 (Fig. 7). A 90-day initial production volume from a specific lateral in an 125 existing horizontal well is used to re- duce variations in rate and flow back. Comparisons are then made between 100 the actual 90-day production and the model’s prediction for the same lateral 75 placement in the integrated model. Fig. 8 shows this comparison for Predicted production,* % Predicted production,* an 8,000-ft lateral in the Upper Wolf- 50 Avoid lower productive zones camp. The predicted higher produc- tivity zones are highlighted in the 25 brighter (yellow) colors and predicted lower values in the darker colors. In this case, the well is placed in what 0 0 25 50 75 100 125 150 175 200 would be considered a good zone and the volume predictions match well Actual production,* % with the actual data. 150706OGJdcu-z07 *90-day oil production. Given this validation, a logical ex- tension is to evaluate an existing low-

UPPER WOLFCAMP: HIGHER PRODUCTIVITY ZONE FIG. 8 performing well bore to see if the 50,000 model matches actual volumes. It can then be examined for more productive zones in the same vertical drilling lane as potential future in-fill candidates. As Fig. 9 shows, the test was an ear- ly pre-earth model horizontal well in the Lower Wolfcamp with actual pro- duction that is about half the average

90-day cumulative oil, bbl producer. This is consistent with what the model would have predicted, but 20,000 in the same section are several zones that potentially would be higher in productivity based on the earth mod- Comparison Actual results Model prediction el, improving both estimated ultimate

90-day initial production 46,302 45,985 150706OGJdcu-z08 reserves and rate of return. Given the stacked unconventional pays in the Midland basin, there is the potential for considerable production tools include a fracture simulator to assist understanding of and cash flow acceleration by developing the entire column the effect of complex fracture growth, integrated with a new simultaneously. This requires an approach, however, that compositional reservoir simulator capable of modeling and enables optimal lateral placement and design of the stimula- further defining the fracture patterns associated with suc- tion program so that fracturing in individual intervals does cessful resource plays. not interfere with those under and overlying. Laredo Petroleum has chosen to characterize the Perm- Comparison with 30 wells ian–Garden City asset with an earth model that will ulti- Evaluating the integrated model begins with its predictions mately reduce uncertainly in production rates and estimated Eprinted and posted with permission to Halliburton Company from Oil & Gas Journal July 6 © 2015 PennWell Corporation Stillwater. in environmental from sciences University, State from theof Wisconsin—Milwaukee, University and amasters Windham Putney, College, Vt., in sciences amasters geological from Petroleum.Lariat in geology holds abachelor Curth of arts exploration, vice-president, was 1997-2001, at He intion, sold was May when Latigo 2006. manager explora- as vice-president, and served as exploration Petroleum Latigo Inc. in 2000 Before joining he Laredo, joined 2006. October land, Petroleum at Laredo Inc., Tulsa, since senior exploration vice-president, andbeen Patrick J. ([email protected]) Curth has authors The project. 2of the on Phase collaboration and contribution its for Houston, Geophysical, Global acknowledge The authors Acknowledgment multiple disciplines. from input ceives re that workflow awell-defined and technologies, modeling of modern use program, acquisition astrong are model earth of asuccessful blocks Thebuilding properties. rock and lithology in complex by variations driven is ductivity pro in variation that understand operator an that requires plays shale of unconventional Thescience reserves. ultimate LOWER WOLFCAMP:PRODUCTIVITYZONE 90-day initialproduction Comparison Eprinted andpostedwithpermission toHalliburtonCompanyfromOil&GasJournal

Actual results 25,100 July 6 © 2015PennWell Corporation Model prediction 24,000 - - Professional Engineers. TexasUniversity of at Arlington. is He amember of the Society Lewisburg, Pa., and from the amaster of science in geology from Bucknell in geology holds University, abachelor of arts beginning25 years, the Monterey with shale in California. Evans ventional and naturally for fractured reservoirs Halliburton. involved has He been uncon- with and Corp. Landmark Graphics— ExxonMobil has acombined of experience with 32years managementproject line. product He service development for Halliburton’s consulting and of integration management asset and business EvansScot ([email protected]) is vice-president from Louisianaence inUniversity, State geology Baton Rouge. the of Rochester, University Rochester, NY, and amaster of sci- from Mauro in geology holds abachelor of arts Corp., consulting organizations, and Halliburton. ing. Mobil of experience with years has He 35 Halliburton’s director for unconventional consult- line,product having service as previously served Co.’sburton consulting and management project director of integrated management asset in Halli- Rick Mauro ([email protected]) is from of MissouriScienceand University Technology, Rolla. andneers holds abachelor of science in chemical engineering of Petroleum Smallwood is amember of the Engi Society 2000. worked the Permian and Delaware basins for ARCO 1990- andin the Texas Panhandle division for BP2000-03, America worked ite Wash Resources 2003-09, Samson experience with in April 2013. Before joining he Laredo, gained extensive Gran- modeling and developmentpresident, field reservoir planning, 20, 50, 000 000

90-day cumulative oil, bbl FIG. 9

150706OGJdcu-z09 Wash team in leader 2011 and vice- Granite and manager engineering reservoir nical advisor and became tech senior engineering as reservoir August 2009 com) in joined Laredo mallwood@laredopetro. B. Smallwood (gs- Gary of London. sity dynamics from Royal Holloway, Univer- master of science in basin evolution and andfrom a theof Leicester University bachelor of science in applied geology holds a seniora geophysicist. Courtier he spent 10 at ConocoPhillips years as manager for the Bakken. Before that, recently as regionalmost subsurface Corp., at Hess 8 years spent he Laredo, joining technology.ences Before exploration and geosci- 2014 as vice-president, injoined Laredo August [email protected]) (jcourt James R. Courtier - - TECHNOLOGY -