Asset Health Insights

Optimizing Asset Health Management

Latest developments in IT and data science make a data-driven approach to asset health management more accessible. This enables asset-heavy organizations to optimize their asset health management and corresponding asset management strategies. ‘Asset Health Insights’ is VIQTOR DAVIS’ conceptual framework to deliver today’s asset health management solutions.

The need for data-driven asset health management and the opportunity it presents

Around 90%(1) of asset-heavy organizations (manufacturers, transport operators, etc.) see the management of their (industrial) equipment, vehicles and other types of physical assets as not very efficient. Major issues such as unplanned downtime and costly maintenance processes exist across different business lines such as production, after sales services, and product development departments.

Source: “Digital Industrial Revolution with Predictive Maintenance”, Siemens

At the same time, latest developments like IoT, Cloud & AI/ML enable a new, data-driven approach to optimize asset management. IoT enables (real-time) streams of data on the operational status and health of assets: streams of sensors measurements or streams of operational data such as throughput. Cloud technology gives access to scalable computational power in order to process large amounts and varieties of data. And generally available AI/ML techniques allow turning this data into actionable insights, e.g., Computer Vision to automatically detect asset defaults, or Anomaly Detection to identify failures early.

Benefits of data-driven asset health management

Data-driven asset health insights enable organizations to operate more effectively. This results in:

  • Increased asset uptime and therefore a higher capacity for core business processes.
  • Reduced total cost through increased asset’s lifespan, more uptime, and by not performing maintenance tasks when they are not needed.

Besides these core advantages, data-driven solutions can also have additional benefits such as:

  • Minimized failures through early detection, with possible consequences to improved safety or the avoidance of cascading damage to a wider system.
  • Empowered workforce through the creation of and access to highly-structured, historical asset condition data.
  • Optimized planning of maintenance and replacement by clustering activities and automatically taking into account any operational constraints
  • Improved quality. Keeping assets health results in better quality of service and improved adherence to manufacturing products specifications.

Key Challenges

Effectively applying a data-driven approach to asset health management has a number of non-trivial challenges.

First of all, often little to no failure data is available. This is a direct consequence of the importance of preventing asset failure; current asset management strategies often allow little failures to occur. To overcome this, we need to use understand the asset’s degeneration and its failures, which is the second challenge. This requires an active collaboration between (process) engineers and the data scientists building a solution.

Secondly, there is no one-size-fits-all approach for this. Different concrete scenario’s can require very different solutions: are we using predictive asset deterioration to predict current and future asset health or failures? Do we need an anomaly detection model to detect anomalous behaviour and signal early warnings? Or do we need root-cause-analyses after failures occur? Also, each of these approaches can again be very different across industries.

And finally, the adoption of predictive models and technical infrastructure are key points that require full attention when building asset health insights solutions.

Asset Health Insights: three conceptual components

An Asset Health Insights solutions consist of three conceptual components. This delivers actionable asset-health insights derived from a solid data foundation that allows for the optimization of asset management strategies.

1. OPTIMIZE ASSET STRATEGIES
The goal is to improve concrete asset management strategies by leveraging data-driven insights on an asset’s health. Improving current procedures for maintenance, inspection and/or replacement? This is not only about optimal timing of these activities. It is also about providing better information to the related planning processes.

2. ESTABLISH A SOLID DATA FOUNDATION
It is essential to convert the right physical information to digital data: physical information that tells enough about the asset’s state so that insights can be derived. Ideally this results in an asset 360-view: a data foundation with best practice standards on data integration, migration and quality that combines all available data on an asset’s current and past state.

3. GENERATE ACTIONABLE INSIGHTS
Use advanced diagnostics to assess an asset’s current overall health & likeliness of failure. Improve prognostics by predicting the future state of an asset. This can encompass failure time prediction and utilization scenario analysis (e.g. for financial planning). Build upon the established data foundation and use highly structured data in root-cause analysis when a failure does occur.

Any asset health insight solution consists of these three components. The starting point of designing such a solution often is a failure mode and effects analysis (FMEA or FMECA) of the most important assets. This followed by a confrontation of these failure modes with possible actions, desired insights and sources of relevant data. The goal is to make a plan for each ingredient, data, insights and actions/strategies, and check its feasibility.

Cases

Railway infrastructure: predicting long-term asset degeneration

The Dutch government task organization that takes care of the national railway infrastructure needs to prognose the remaining lifetime of various railway infrastructure components in order to do long term planning and budgeting.

Together with VIQTOR DAVIS they developed short-term predictive models (max 7 years ahead) and long-term predictive models (7-30 years ahead). These models were developed for various asset-degeneration indicators based on a plethora of data sources from throughout the organization.

The combination of the developed models gives an overall asset view with a good indication of the remaining lifespan of a railway infrastructure component. This allows informed decisions on maintenance and replacement requirements for the long term, while also offering added value for short term maintenance and replacement planning.

Large bottler of beverages: detecting misalignment of bottling line and predicting bearing failure

A large bottler with production facilities in many different countries wanted a platform to maximally gain value from the data streams produced by their machines. Together with a partner VIQTOR DAVIS unlocked the data streams and integrated the data to a cloud platform on which several dashboards giving insight into the state of the machines are built.

For specific machine failure modes, we have created detection and alerting modules. On a filling line we analyze vibrations from which we can detect misalignment of the machine. Other vibrations at different frequencies are used to detect future bearing failures: we have established control limits that are not violated in years, but when violated bearing failure will occur in the next few weeks.

Our solution is part of a larger program of digitalizing production facilities. Our detection and alerting method and our dashboards give the customer better insight in the state of their assets, saving cost by begin able to schedule maintenance and replacements instead of having to do them during production runs.

Want to know more?

VIQTOR DAVIS offers a unique combination: We are your partner in data strategy, governance, management, science and analytics. We provide professional services, full-service solutions and knowledge transfer. We call this Data Craftsmanship. Build your Asset Health Insights solution? Get in touch.

Visit the on-demand Emerce webinar, in which ProRail, Microsoft and VIQTOR DAVIS talk about ML-Ops.

* indicates required
Check

Bedankt voor uw bericht. We nemen zo snel mogelijk contact met u op.

Check

Succesvol geabonneerd op de VIQTOR DAVIS nieuwsbrief