YPF is the largest oil and gas company in Argentina, with a 43% market share in oil and gas production, and 58% in gasoline. As the third largest company in South America, YPF employs 72,000 people directly and indirectly, and holds 92 production blocks and 48 exploration blocks in basins around the country.
YPF was aiming to reduce its costs associated with oil wells’ maintenance downtime and equipment breakdown. The analysis showed that the root cause of inefficiencies was the lack of a robust maintenance scheduling process.
The main problem was that the scheduling process was decentralized, with planners independently allocating hundreds of work orders each month. Multiple planners were assigning tasks for multiple teams without proper coordination between each other. This prevented them from creating optimal schedules, resulting in downtime losses and inefficient resource utilization.
YPF approached Simcastia, the simulation and optimization business unit within Continente Siete (an Argentinian business data science company), to develop a scheduling tool to streamline asset management at all YPF facilities.
The pilot project was carried out for Rincón de los Sauces, an oil field located in Neuquén, Argentina, with 700 wells (including water injection and oil wells), about 100 working crews, and more than 100 weekly maintenance orders.
To manage an oil field maintenance system that included many interacting parts, custom policies, constraints, and time-dependent events, using spreadsheets and analytical optimizers was not sufficient. Simcastia consultants developed a simulation-based optimization solution to manage this complexity. They chose AnyLogic simulation software for its unique flexibility, which allowed them to model specific resource behaviors and custom process rules.
The simulation model included sites with GIS-referenced locations, preventive and corrective work orders consisting of multiple tasks, and resources (staff and equipment) with skills and working hours. All of these elements were modeled as agents with unique properties and behavior patterns.
The costs calculated in the model included:
- Wells’ production loss costs, including unplanned and scheduled interruptions in operations.
- Resource-related costs (both regular and extra work hours).
- Travel-related costs.
The solution developed by Simcastia was based on an AnyLogic maintenance process simulation model and custom optimization algorithms. The simulation-based optimization used algorithms to allocate resources to work orders and complete these work orders the fastest possible way.
The interface of the software solution allowed the planners to adjust model parameters, such as associated costs, weather conditions impeding some processes, resource availability timetables, and prioritization rules. By changing model parameters, the planners could feed the model data relevant to the changing environment.
Combining simulation with optimization, the tool produced operational plans for 9, 12, and 30 days. It also provided various statistics displayed in dashboards, including operational plan details, schedules by resource and site, costs and tasks by type, resource utilization rates, extra hours worked, and distances covered.
Dashboards: resource allocation and plan statistics.
The resulting solution was integrated with the client’s databases and SAP, becoming the part of the company’s planning software infrastructure.
This project provided the Rincón de los Sauces oil field with a decision-support tool for maintenance scheduling, which helped improve the site’s operational efficiency and resulted in:
- Work order execution time increased by 11%.
- Preventive maintenance fulfillment increased to 95% in six months.
- Corrective maintenance backlog reduction of 56%.
- Unplanned downtime oil production losses reduction of 50%.
The direct economic impact of the project included annual yearly savings of $18M at Rincón de los Sauces. As a second stage, implementation for Mendoza assets (much bigger and more complex) have already started, and the roadmap aims for nationwide implementation by the end of 2018, with expected savings of $234M per year.
To learn more, watch the project presentation at AnyLogic Conference 2016 or download it.