Astra Tech Consulting

Plains Midstream Corporation (PMC) is one of Western Canada’s leading oil and gas midstream producers, operating large-scale facilities that manage the collection, transportation, and processing of petroleum products. These operations are complex — reliant on intricate networks of equipment, varying energy costs, and shifting market demands.

Despite the abundance of operational data available across their systems — including work orders, invoices, equipment specifications, maintenance reports, and process maps — much of this information existed in silos. Operators relied heavily on historical patterns and professional intuition to plan workloads and forecast operational costs. While their experience was invaluable, the absence of real-time predictive intelligence often resulted in inefficiencies, missed cost-saving opportunities, and suboptimal scheduling during high energy cost periods.

PMC needed a smarter way to plan — one that could bring together internal data, external market signals, and operational context into a single, actionable intelligence layer. They envisioned a solution that could guide their facility operators and managers toward the most economically and operationally efficient scheduling decisions possible, without disrupting established workflows.

Working closely with PMC’s operations and IT teams, we designed and delivered a data-driven work planning tool that harnessed machine learning to turn years of operational data into predictive insights.

At the heart of the system was a machine learning model trained on PMC’s own operational history — work logs, process documentation, maintenance records, energy consumption data, and even equipment-level specifications detailing energy-to-output relationships. We enriched these internal datasets with external information such as regional power demand and pricing data from Alberta Power, giving the model a broader economic context for its predictions.

Our engineering approach was both pragmatic and forward-thinking. The model was trained using well-established Python-based frameworks of the time, then deployed behind a lightweight Flask API that could be seamlessly integrated with other systems. On the front end, a modern React-based interface provided facility managers with a clear, intuitive view of recommended work schedules, cost forecasts, and what-if scenarios.

The result was a unified tool that combined data science, automation, and human decision-making — offering predictive guidance without removing human oversight. Operators could plan workloads, run simulations, and make informed trade-offs between cost, capacity, and timing, all from an interface that fit naturally within their existing planning routines.

The introduction of the Work Planning Tool represented a step change in how PMC approached operational planning. Facility managers who once depended on manual data analysis could now access intelligent recommendations that dynamically adjusted to changing conditions — whether driven by energy prices, maintenance schedules, or equipment availability.

In the months following implementation, PMC observed a meaningful reduction in operational energy costs as workloads began to align more closely with favorable power pricing windows. Cost forecasting also became more accurate, enabling leadership to make better-informed financial and strategic decisions.

Beyond the measurable gains, the cultural impact was equally significant. The tool became an everyday companion for operators — not a black box replacement for expertise, but a data-driven assistant that enhanced their judgment. Over time, it fostered a more data-conscious operational culture, where decisions were backed by evidence and supported by intelligent systems.

At Astra Tech Consulting, we believe that the most impactful technology is built at the intersection of data, engineering, and human experience. This project exemplified our approach — translating complex data ecosystems into tools that are as intuitive as they are powerful.

Our work with Plains Midstream Corporation was not about building a one-off predictive model. It was about integrating intelligence into the daily rhythm of work — ensuring that operators could make smarter, faster, and more confident decisions. By focusing on usability as much as accuracy, we created a solution that fit seamlessly into PMC’s operational context while unlocking new layers of efficiency and insight.

The ability to extract meaning from “dark data” — operational records, equipment specifications, and the unstructured information buried in reports — is one of Astra Tech Consulting's defining strengths. By combining our experience in data integration, machine learning, and enterprise-grade software development, we help clients transform this overlooked information into tangible value.

This project demonstrated the early potential of AI-assisted operational planning in heavy industry. At a time when machine learning was just beginning to move beyond experimental use cases, Plains Midstream Corporation took a pioneering step by adopting a model-driven approach to cost forecasting and work optimization.

The success of this initiative highlighted a broader trend that continues to accelerate today — the fusion of operational data, automation, and predictive analytics into intelligent systems that adapt to the dynamic realities of production environments. What began as a tool for scheduling optimization became a foundation for a new way of thinking about industrial efficiency: one where data continuously informs every decision, from the field to the executive boardroom.

As technology continues to evolve, Astra Tech Consulting remains focused on helping organizations build this kind of operational intelligence — where data, machine learning, and human expertise work in harmony to create measurable impact.

Want to learn how machine learning and operational data intelligence can reduce costs, uncover hidden efficiencies, and transform your organization’s decision-making?

Let’s build something intelligent together.