Skip to main content

 

📌 Background

In many chemical plants, reactors contain catalyst beds that gradually lose activity and need to be replaced on a routine basis. Because catalyst life can vary significantly, it is difficult to predict exactly when a failure will occur.

A premature catalyst failure would have a major impact on production, supply chain, and maintenance. Therefore, it is critical to plan preventative maintenance well in advance – at least a few weeks prior to failure – to avoid costly unplanned downtime and production losses.

 

🎯 Challenge

With TrendMiner, a workflow can be set up to monitor and predict catalyst performance:

  1. Searching on Process Values
    The first step is to use value-based searches to identify when process values reach certain limits. These searches can be saved and later used as the basis for monitors. In this way, early indicators, warning levels, and critical conditions can be defined, all providing valuable insights into catalyst performance. Event analytics can of course also be used for further analysis.

  2. Monitoring for the Search Conditions
    Based on the saved searches, monitors are configured to track real-time process behavior. Engineers can see immediately whether the reactor is operating within its optimal zone or if deviations are emerging. Current value tiles and monitor tiles can be added to the dashboard to highlight these states visually.

  3. Create a Performance Dashboard
    All key KPIs and monitor information are brought together in a central performance dashboard. This provides at-a-glance insights into the current health of the catalyst.

    Example of a performance dashboard

     

✅ Value

The combination of value-based search, monitoring, and dashboards creates significant benefits:

  • Early detection of catalyst failures – well before critical limits are exceeded.

  • Proactive maintenance planning – more lead time for maintenance and supply chain coordination.

  • Reduced unplanned downtime – by enabling timely catalyst replacement.

  • Efficient visualization – clear dashboards that make it easy to assess the asset’s condition at a glance.

Natasha, 

I like this. I am in Oil & Gas and we have many units using catalysts with a similar need to monitor to ensure the expected production run is met.

For instance, we have some feedstocks that include impurities, such as metals that impact the catalyst activity. We are measuring metals in feed in the lab and typically get results daily.

Ideally, we would have a start/end date context item for the projected run with a linear cumulative not-to-exceed metals allocation from 0 at start of run (SOR) to max allocation at end of run (EOR) and we would calculate the totalized value of metals on catalyst from SOR to today and compare with the allocation on a TrendHub plot that would show a projection in the future. A context item would be created when the totalized metals value exceeds the allocation. 

We have stumbled on a few things in trying to implement a solution for this in Trendminer. The integral aggregate function can cover only one day and would not extend from SOR to now; and TrendHub cannot plot data in the future. 

At TrendLab, we learned that we can possibly overcome these problems by using ML-HUB to perform the totalizing and allocation calculations and to feed a plot object with the ability to plot data in the future that we would drop in DashHub as a tile. That is next on our list of things to do


@yracette Thanks for sharing that feedback. Happy to hear that TrendLab inspired you to take a next step!