CBM (Condition Based Maintenance)

[Curious George]

I'm a reliability engineer and we are starting look at ways we can use our historian data to develop metrics for condition based monitoring.  Is Seeq a good tool for generating such analyses?  Can someone suggest an example to help me get started?

[Guest Jon Peterson]

We believe Seeq is a very important component of Condition Based Maintenance programs as well as Reliability Centered Maintenance (RCM). Seeq has the analytic capabilities as well as reporting and collaboration capabilities. Just analyzing is not enough--the results need to be incorporated in the business.

A simple CBM example to help get things going is calculating run hours and on-off cycles; say on your pumps or compressors. Then, of course, calculating weekly and monthly run hours/on off cycles. Here's an example, with a few screenshots using the Seeq example data.

We'll work with a compressor power signal. In Seeq create a new Seeq Analysis and find this signal by navigating down the cooling tower example asset in Seeq. 

One you get your signal up, the next step is to create a Seeq "Condition" for when the compressor is on. This is a simple value search--we'll consider the compressor is on if power is greater than 5 kW. The image below has the Value Search pane open:

The condition is Compressor On, plotted are the capsules (green bars along the top) that represent when the compressor is on. 

Things like run hours and on-offs make sense when counted or summed periodically. For this example we will do this aggregation over weeks using the Seeq Periodic Condition Panel and select Weekly. Of course, you can do this over what ever period you want. And, you can follow these steps to do this for multiple periods, like Weekly and Monthly. 

Now we are going to use the Seeq Signal from Condition tool to count the on-offs per week and run hours per week. You can see the settings I chose in the images, but here's a quick explanation of each:

• Signal or Condition: this is the item to be aggregated. In this example, we choose the Compressor On condition
• Summary Statistic: we are using Count for the on-offs and Total duration for the run hours. Note that you can set the time units with Total Duration.
• Where to place the statistic. I like to use start because this represents the value for the entire week and it's common with historians to do "step-right" plotting of data.
• Linear vs. Step. Step generally makes the most sense--there is not much meaning in interpolating linear between two aggregated values. 

Of course, it is easy to experiment with these settings and choose what you think is best. I zoomed out to a month to get multiple values plotted. 

Note that I removed the weeks from this image. The last step in Seeq Workbench is to create a worksheet for the two "metrics" we created. The easiest way is to duplicate the worksheet twice, and remove everything but the metric. Here's one of the final worksheets:

I need to point out the newest bar on the right edge. Notice how it is slightly dim--that indicates the week is in-progress so we don't have a final number yet. 

The last step is to publish these into a Seeq Organizer Topic. Go back the the Seeq home page by clicking the 4-square in the upper right. Then select new Organizer Topic. Once in the topic, add the content from the two metric worksheets you created. I won't go into the details here--but please ask if you have any questions on using Seeq Organizer. Here's the final result:

As I mentioned above, this is a good starting point for CBM. A future release of Seeq allows for monitoring. So, it will be possible to put limits on these metrics and alert if they exceed some threshold. 

[Curious George]

Thanks.  This gives me something to work on.