Petroleum midstream provider
Pipeline-Scheduling Optimization
Forward Deployed Engineer — Palantir
~3x cut in human time, ~2x energy savings
- Digital Twin
- Physics simulation
- State estimation
- OR-Tools
- Optimization
Problem
A petroleum midstream client needed to schedule pipeline operations against real physical and commercial constraints — tank sizes, commodity value, power use — a combinatorial problem that was being worked through manually, consuming significant operator time and leaving energy savings on the table.
Approach
Built a full-system simulation of pipeline operations under the client's real constraints, rather than a combinatorial solver. The simulation approach was a deliberate tradeoff: a solver might find a marginally better schedule, but a simulation the operations team can inspect, adjust, and trust is more valuable — and more maintainable — for a system that has to keep running after the delivery team leaves.
Outcome
Cut the human time required for scheduling by roughly 3x and reduced energy use by roughly 2x, while keeping the scheduling logic transparent enough for the client's own team to operate and adjust.