Effici Ef ficiency ency & Pe & Perfo rformance rmance - PowerPoint PPT Presentation

Vis Visual ual Ana Analytic Tech lytic Techniques niques fo for r Ope Operational rational Effici Ef ficiency ency & Pe & Perfo rformance rmance Imp Improv rovements ements Haskayne School of Business CORS 58 th Annual Conference, May 31 st , 2016

Presentation Outline • Thank you to CORS & Haskayne School of Business • About Verdazo Analytics Inc. (& a wee bit about me too) • Outline of presentation Part 1: Upstream Oil & Gas Industry Part 2: Operations Analytics Part 3: Analytics Journey Part 4: Analysis Challenges Enterprise class. User-friendly. Discovery Analytics.

About Verdazo Analytics Inc. Founded 10 years ago as VISAGE (rebranded to VERDAZO in 2016) • Recognized a need, particularly in operations, for data integration & visualization • Upstream Oil & Gas focus up to 2016, currently expanding to other industries • Currently active in >70 companies • E&P Companies (from start-ups to large North American producers) • Reserves Evaluators • Banks/Investment Groups • Market Research Organizations • Service Companies • Enterprise class. User-friendly. Discovery Analytics.

Part 1 Upstream Oil & Gas

Upstream Oil & Gas Huge margins when times are good… not so much now • Capital intensive ($3.5 million = average horizontal well Drill & Completion cost in • 2014) with some wells costing in excess of $20 million Completion technologies allow us to get more production more quickly • Reactive industry, particularly to commodity prices • Lots of uncertainty… not always well understood or adequately represented in plans • There’s lots of data • Still heavily reliant on Excel • Enterprise class. User-friendly. Discovery Analytics.

What challenges do Petroleum Producers face? Low commodity prices & dramatic price fluctuations • Wells are expensive to drill • Well count per Engineer is high (especially after lay-offs) • Strive for growth with less resources • Predictable cash flow • Too many spreadsheets • Enterprise class. User-friendly. Discovery Analytics.

Horizontal wells have changed the production landscape Enterprise class. User-friendly. Discovery Analytics.

Horizontal wells have changed the production landscape Enterprise class. User-friendly. Discovery Analytics.

We can’t predict prices, but we can protect against them Images from VERDAZO Blog: Forward Curves Are a Poor Predictor of Future Spot Prices Enterprise class. User-friendly. Discovery Analytics.

Deep staff cuts: a common approach, but a good one? Example company: spends 65% of G&A on employees (including benefits and bonuses) • G&A represents 20% of total operating costs • employees are 13% of total operating costs • 20% staff reduction = 2.5% reduction of total operating costs (not taking into account • the added costs of severance) the impacts to analysis capacity and capability are dramatic and could undermine • their ability to realize operational efficiencies targeting operational efficiencies could be more fruitful and could result in • sustainable improvements Enterprise class. User-friendly. Discovery Analytics.

Operational Efficiencies: How big is the prize? Province Revenue Potential AB $ 2,236,719,763 SK $ 382,188,022 BC $ 162,083,665 MB $ 40,715,664 Total $ 2,821,707,114 Enterprise class. User-friendly. Discovery Analytics.

Part 2 Operations Analytics

Types of Analytics Enterprise class. User-friendly. Discovery Analytics.

The process & roles for a successful analytics project Does this fit operations analytics? It does in well-bounded analytics projects, but… Source: Five Faces of Analytics presentation by Dark Horse Analytics Enterprise class. User-friendly. Discovery Analytics.

What’s unique about Operations Analytics? Significant variability in assets, production technologies, reservoir issues (e.g. CBM, • tight oil, liquids rich gas, water floods…) Conditions change over the life cycle of the well (with all wells at different stages) • Data currency is important (i.e. up-to-date data) • Team approach (management, engineers, field operators…) • Multiple engineering disciplines (drilling, completion, facility, reservoir, production) • Multiple departments (operations, engineering, production accounting, financial • accounting…) Enterprise class. User-friendly. Discovery Analytics.

What’s required for Operations Analytics? Tool selection is the starting point. The analytics tool needs to: support an iterative process of continuous learning, investigation and • collaboration enable a narrative … a set of visualizations that tell a story • be nimble to adapt to evolving needs • support “ Discovery Analytics ” workflows • Enterprise class. User-friendly. Discovery Analytics.

What is Discovery Analytics? “Discovery Analytics is a sequence of explorations, each predicated on the discovery and insight of the last exploration. It’s about a path of exploration that can change with each new discovery … it’s not something that can be anticipated. Some tools let you build an environment to explore data, but only within the bounds of how it was built and limited by the technical and domain expertise of its creator .“ Enterprise class. User-friendly. Discovery Analytics. 17

Key Analytic Needs Source: What do data analysts need most from their tools? Enterprise class. User-friendly. Discovery Analytics.

The importance of the narrative Don’t rely on one visualization type, or one performance measure… assemble multiple perspectives that comprise an informative narrative. An illustration of multiple visualization types could include: Rate vs Time 1) Cumulative Production vs Time 2) Rate vs Cumulative Production Each offers an important, 3) and unique, perspective Percentile (Cumulative Probability) 4) Percentile Trendlines 5) Probit Scale 6) The following examples are from VERDAZO presentation: Understanding Type Curve Complexities and Analytic Techniques Enterprise class. User-friendly. Discovery Analytics.

The importance of the narrative An example of three performance measures that tell a different story… Also consider: Payout • NPV • Completion cost • Operations implications • Etc. • Measure against what’s important to you! Image from VERDAZO Blog: What production performance measure should I use? Enterprise class. User-friendly. Discovery Analytics.

Narrative Example: 1) Rate vs Time Strength: good for early production comparative analysis. Weakness: not as good for longer term production comparative analysis. Enterprise class. User-friendly. Discovery Analytics.

Narrative Example: 2) Cumulative Production vs Time Weakness: not as good for early production comparative analysis. Strength: very good for longer term comparative analysis. Also useful for quick payout analysis. Enterprise class. User-friendly. Discovery Analytics.

Narrative Example: 3) Rate vs Cumulative Production Weakness: does not effectively communicate the time it takes to achieve a level of cumulative production. Strength: provides a visual trajectory towards Estimated Ultimate Recoverable (EUR). Enterprise class. User-friendly. Discovery Analytics.

Narrative Example: 4) Percentile (Cumulative Probability) Strength: communicating statistical variability of a dataset. Weakness: it only represents a single moment in time. Enterprise class. User-friendly. Discovery Analytics.

Narrative Example: 5) Percentile Trendlines Percentile Trendline = extrapolated percentile of a collection of wells for each period in time. Strength: provides a meaningful comparative context to assess performance. Image from VERDAZO Blog: So what is the problem with production type curves? Enterprise class. User-friendly. Discovery Analytics.

Narrative Example: 6) Probit Scale (Cumulative Probability) Strengths: 1) the shape can help determine if the results trend towards a lognormal or normal distribution 2) a “Probit Best Fit” regression can provide a variety of statistical insights including a measure of uncertainty (P10/P90 Ratio) Weakness: it only represents a single moment in time. Enterprise class. User-friendly. Discovery Analytics.

Enhance the narrative with normalization Comparative analysis using normalization is an effective means to put performance into a meaningful context. Types of data normalization include: Time normalization • Time alignment to a common starting point (e.g. first production, peak rate). Lets you compare behavior from that • common starting point. Dimensional Normalization • Establish a meaningful comparative context (e.g. production/100m completed length lets you compare wells of • different length and quantify production gains as wells get longer) Fractional Normalization • Used to characterize temporal behavior relative to a timed-benchmark (e.g. Production rate as a percent of peak used • to characterize decline behavior) See SPE Presentation Understanding Type Curve Complexities and Analytic Techniques for more details. Enterprise class. User-friendly. Discovery Analytics.

Part 3 The Analytics Journey