These online demos are not supported on a mobile device.
This website is not supported by Internet Explorer - please consider using the latest browser

Maintenance planning

Problem

Business Problem

A mature offshore asset is facing a heavy work load of preventative and corrective maintenance scope.

The asset has multiple stakeholders with different opinions about prioritization, who are all competing for the same execution resources.

The asset's planning process is predominantly manual. The resulting maintenance plan is suboptimal and unstable with a year-on-year increase in backlog.

Goal

Use our Game AI approach to create better plans: incorporate different views, use resources more efficiently, and reduce the backlog.

Scale 9,000 work scopes are on ten different locations, all varying in required skills, material, number of workers and duration. Competition for recources Too much work for the available resources (i.e. the work boat and maintenance crew) - how to best plan? Different opinions Opinions are often not quantified (“whoever shouts the loudest”) which gives tension and poor quality decision making. Changing conditions Inspection results, production upsets or leaks are leading to continuous change in the portfolio - and in the priority of the work. The result is an ineffecient use of resources and more work coming in the going out: back log builds up. Result

Maintenance planning

Process

Game AI

We translated the asset’s maintenance planning into a resource allocation game, and taught an AI agent to play this game.

Through self-play, the AI learns to master the game mechanics and to develop strategies that outperform human planning capability.

Using our AI as a planning assistant, we can now develop maintenance planning strategies much more efficiently than before, and translate stakeholder views immediately into alternative plans to improve decision making.

Learning
the basics
of the game
0 GAMES PLAYED

Maintenance planning

Dashboard

Opinions

Use the toggles below to reflect what opinions need to be taken into account. Use the dashboard to understand the impact of these opinions on key metrics of the plan.

Process safety

no

Reduce backlog

no

Safeguard production

no

81%

Safety critical scope completed

7659 orders

Backlog at year end

9800 bbl/day

Production at risk

10.828 orders

Workorders executed

Crew efficiency

Work remaining

Vessel timeline

Production safeguarded

Load the
Game AI