The Bottleneck That Wasn't
A sulfur logistics model found the real constraint two stations upstream of where every operations team was pointing and redirected a capex decision worth hundreds of millions.
Read the case studyThe Question
Case study · Oil and gas · Fleet sizing
Client brief
Ferus, a Calgary-based energy services company, runs a compressed natural gas distribution system across the North Dakota Bakken. The gas comes from oil wells that would otherwise flare it. Compression happens near the wellhead, and the gas rides in Ferus trailers to drilling rigs and frac fleets that burn it in place of diesel. The trailers double as storage: a customer's supply sits on a Ferus trailer parked on site until it runs down. Supply points and customers move every few weeks as rigs finish one pad and start the next. Winter weather, equipment maintenance, and road closures all eat into delivery windows. Ferus has promised its customers a 100% level of service and now has to decide how many tractors and how many trailers to put on the road to keep that promise at the lowest cost.
§ 1
A fleet is capital. Too few trailers and a rig runs dry, switches back to diesel, and the service promise breaks. Too many and the money sits parked in a field the operator may leave within a year.
The timing raised the stakes. In 2013 North Dakota flared more than 30% of the gas it produced. In 2014 the state adopted capture targets that stepped up toward 90% by 2020, and operators needed somewhere for that gas to go. Ferus and GE built the answer as a joint venture called Last Mile: GE's CNG In A Box compression next to the wellhead, Ferus trailers and logistics on the road. Statoil ran the first pilot near Watford City, then expanded it in 2014 to six drilling rigs and a frac fleet.
Moving from pilot to commercial scale is the moment a fleet gets bought. Ferus wanted the number before signing the orders, and it wanted the number to hold on the worst day of February, with the lowest cost that still delivered 100% service.
§ 2
Nothing in the network stays put. A route plan assumes fixed customers; here both supply and demand migrate with drilling activity, so the geography of next quarter is unknown today.
Travel times swing with weather. Trailers tied up as on-site storage are trailers not hauling. A tractor in the shop is a tractor not dispatched. Averaging each of these in a spreadsheet looks reasonable and hides the day that matters: a storm doubles drive times while two rigs run low at once and a tractor is down for maintenance. Service fails on that day, and a fleet sized on averages never sees it coming.
§ 3
SimWell built a discrete-event simulation of the distribution system in Arena. Because future well locations cannot be known, the model places supply points and customers on a latitude and longitude grid laid over the road map of the service area. A travel calculation tool returns drive times between any two grid points. A Markov chain moves the system between weather states and stretches travel times accordingly. A dispatch algorithm written for the model assigns tractors and trailers to deliveries the way the operation would, including the trailers left on customer sites as storage.
With the operation running in the model, Ferus could change the fleet and watch the service level. Scenarios covered tractor count, trailer count, and the operating processes for dispatching and swapping trailers.
§ 4
Ferus sized its tractor fleet and its trailer fleet from the model, with weather, maintenance, and road closures in the run rather than averaged out of it. The same model let the team test its operating processes side by side and choose the most efficient one.
Ferus went into the Bakken winter with the fleet the service promise required, at the lowest cost that promise allowed, and nothing bought on a margin of comfort.
Results
Ferus sized its tractor fleet and its trailer fleet from the model and chose its operating process from the alternatives it tested, holding the 100% level of service it had promised while keeping fleet capital at the minimum that promise allowed.
The industry outcome the network served has held. North Dakota's gas capture rate ran at 94 to 95% through 2025 and into 2026, above the state's 91% target. In February 2024 Ferus Natural Gas Fuels merged with Cryopeak LNG Solutions to form Cryopeak Energy Solutions, the largest private, vertically integrated LNG producer and service provider in Canada.
SimWell helps operators make better decisions under uncertainty. We build simulation and optimization models that reflect operational reality, test scenarios before committing resources, and transfer capability so teams can run decisions independently.
More case studies
A sulfur logistics model found the real constraint two stations upstream of where every operations team was pointing and redirected a capex decision worth hundreds of millions.
Read the case studyIndependent pre-feasibility logistics research found two routes prior advisors had missed and informed a US$25M commitment on deadline.
Read the case studyA simulation of RTC's Métrobus lines in Quebec City tests 30+ anti-bunching strategies in combination and scores each on riders' excess wait time.
Read the case studyNisl amet euismod suscipit eros lectus aliquam nisi, hendrerit donec