A regional agribusiness

Problem

Managers of a domestic agribusiness, with over 150,000 acres under cultivation, sought to assess and improve operations but the uniqueness of its private transportation network had prevented sufficient benchmarking and industry comparisons.  A multi-million dollar capital investment was under consideration to replace fleet components that were several decades old.  The company retained Princeton Consultants to implement best practices and metrics and to design a software tool to make ongoing real-time recommendations toward optimal efficiency.

The company harvests cane sugar and transports it to its mills via an internal industrial railroad and a connecting Class III railroad used by the community.  These discrete railroads have different equipment requirements, types of freight permitted, labor laws and safety regulations.   As with many agribusinesses, seasonality and same-day spoilage were key factors.  Optimizing asset utilization had to account for these and other complexities.

Approach

Princeton Consultants employed its optimization project methodology of a small team looking holistically at the business process and working incrementally and rapidly toward custom deliverables.  On this project, the three-man team completed a working software model in less than three months.  This speed was facilitated in part by the firm’s subject matter expertise in railroad operations.

Initially, the team members were embedded on-site.  Working with client executives, they assessed the agriculture, operations and IT departments.  They observed and interviewed yardmasters and harvesters to learn how they made scheduling decisions about competing resources under time pressure.  The consultants toured the facilities and fields, rode the trains during harvest, and walked the rail yards with stopwatches.  The same team members evaluated potential problem solving approaches with expert optimization modelers at Princeton Consultants and then designed iterations of a simulation that addressed fleet sizing, selection and scheduling, as well as the relative value of operational capital investment such as increasing train speed.  The final model showed substantial results that were easily validated.

Complexities:

  • Noisy Data: variances in location data and other key data points
  • Stochastic: variances in harvesting time, transit time, crew availability
  • Human in the Loop: Human scheduler reviews Optimizing Simulation results
  • Static/Dynamic: Changes are made and edited as the day's harvest and transit change
  • Real Time: "On the phone" need to redirect equipment and crew finish work on-time/ early/ late, and weather, breakdowns, and other conditions occur.

Results

Princeton Consultants’ optimization model demonstrated that the company could improve operating efficiency by approximately 12% with the same personnel and with 11% fewer railcars.  The capital investment required for new locomotives and railcars was significantly reduced.  Going forward, there will be large productivity gains through greatly reduced spoilage of cane sugar.

Time to Payback (TTP), a much more rigorous standard than Return on Investment (ROI), was almost immediate: the entire project was paid for in less than 2.5 months.   The optimization software will continue to serve as a trusted advisor for company executives, a superior deliverable compared to a pure strategy assessment that would have offered static recommendations with a limited lifespan.