In previous posts we gave an overview of business analytics ("The Basics of Business Analytics"-- Part 1 and Part 2) and described how model-driven prescriptive analytics can maximize profit by streamlining inefficiencies within your logistics network ("Optimizing Your Logistics Network Through Model-Driven Prescriptive Analytics").

In this post, I want to focus on a prescriptive analytics implementation. Utilizing the River Logic platform, we were able to quickly (within a couple of weeks) develop this deceptively simple looking “crude gathering by truck” model that will optimize the trucking fleet required to service a very large number of crude leases.

(Screenshot of actual model in the River Logic EO development platform)

Advantages of the River Logic platform include:

• Able to handle both the purchase or lease of various sized trucks from an any number of sources and will tell you the optimal fleet needed given the leases that need to be serviced.
• Capable of handling an unlimited number of leases with varying flow rates, stock tank sizes and geography.
• Leases can quickly be added to understand the cost of servicing, as well as the profit potential under various scenarios prior to committing and improve your negotiating position.
• Accounts for travel time and can account for variations, depending on the time of day and learns over time as actual travel times are loaded into the model.
• Crude can be delivered to any number of receiving facilities (terminals, refineries, etc.) with full constraints on any type of rack configuration and tankage.
• Can manage multiple crude types and associated tankage considerations

Granted that the model is currently standalone and has not (yet) been deployed to the cloud (we use Azure). We do have full data import/export capability to Excel, but the interfaces directly to other client-specific system can be developed to remove data entry and transposition errors. And while we haven’t fully developed the full suite of visualization tools (using MS BI, Spotfire, Tableau, etc.) beyond what is needed for development/testing, this is easily done and customized to the exact client specifications.

The model can be extended to include infrastructure downstream of the terminal and include other modes of transportation (rail — both manifest and unit train, barge or ship and pipe). The model can also be redone in reverse — to model the distribution of various refined products from multiple refineries to multiple terminals among various modes of transportation (truck, rail, barge pipe) and extended beyond that to go from terminal to consumption destination (gas station, airport, marine terminal, etc.). In the ultimate implementation, we would model/optimize your entire value chain from wellhead to gas station.

This same concept can be applied to the natural gas pipelines (optimize compression to meet receipt delivery commitments), petrochemicals (shipping polyethylene pellets in bulk to bagging facilities, getting it bagged, palletized onto containers, moved to ports, transloaded between containers and shipped to destinations), oilfield services (crews, rigs, sand, water, etc.) and into other, non-logistical areas (optimizing refinery turnarounds, optimizing drilling programs given a set of lease terms, subsurface issues or processing/logistics constraints, etc.) — essentially any type of constrained system; the more complex the better.

Prescriptive analytics is revolutionizing business planning and the advanced analytics space through its unmatched ability to help enterprises, supply chain groups, energy businesses, and sales and operations planning executives practice decision-making that is profitable and mitigates risk.