It is a well known best practice to plan and schedule asset maintenance. However, with the attrition in revenues most companies face in the uncertain economy, it is more critical than ever to optimize maintenance and minimize unplanned downtime.
Because of advancements in technology, such as the launch of SAS' upgraded Predictive Asset Maintenance application announced this week, companies are able to not only contain maintenance and operating costs, in some cases they can improve customer satisfaction.
To build an appropriate predictive model for a given asset, SAS first works with the company to build a reliability information store. This process begins by partnering with data historians, and tapping into asset data and maintenance records. The software will then process asset condition data, gathered from the equipment, typically through sensors, to surface emerging issues and provide "alerts". In one case, the SAS team explained, they were able to provide an Oil and Gas customer with an alert 10 weeks ahead of a potential failure.
The application not only provides reports and alerts, but helps the operator to perform root cause analysis on the part with performance issues. Through the Case Management feature, the service organization can track the efficacy of service. Furthermore, SAS claims it has functionality to aid operators in determining when it makes sense to perform planned maintenance when they are in the proximity of a failed part that they are working on based on an alert.
Developing accurate predictive models for complex failure mechanisms in varying equipment configurations is a complex task. It requires aggregating information from multiple sources, complex statistical models, and sophisticated data acquisition and processing, often dealing with nonstandard data formats and semantics. We believe that while this field is still evolving, industry will benefit from standardizing data models and statistical approaches that can be reused to accelerate time to value in mission critical applications.