Jump to content

Lindsey.Wilcox

Administrators
  • Content Count

    71
  • Joined

  • Last visited

  • Days Won

    14
  • Points

    31 [ Donate ]

Lindsey.Wilcox last won the day on November 16

Lindsey.Wilcox had the most liked content!

Community Reputation

29 Excellent

5 Followers

Personal Information

  • Company
    Seeq
  • Title
    Analytics Engineer

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. Background As a starting point, I have a signal that indicates the process error code (Error) and 2 signals that indicate the current container (Container 1 and Container 2). I would like to create a histogram that summarizes which containers have the most frequent errors. More specifically, I want to know: For the Container 1 signal, what is the distribution of container values when the error code is equal to 44 and 47 For the Container 2 signal, what is the distribution of the container values when the error code is equal to 45 I want the results summarized as a single histogram. The following steps describe how this can be achieved using conditions and capsule properties. Solution 1. Create a condition with a capsule for each value change in the Error signal. This can be accomplished in Formula using the following syntax $errorSignal.toCondition() Note: the .toCondition() operator assigns a Value property to each capsule that indicates the value of the Error signal during the capsule. This property can be used for aggregation in a Histogram. 2. Filter the condition created in Step 1 to only include capsules with errors of interest for the Container 1 signal. This can be accomplished in Formula, using the following syntax: //Create intermediate conditions for each error $error1=$errorCond.keep('value',isEqualTo('44')) $error2=$errorCond.keep('value',isEqualTo('47’)) //Combine the 2 intermediate conditions $error1.combineWith($error2) 3. Assign a ‘Container’ capsule property to each of the capsules in the Error Cond 1 condition that indicates the value of the Container 1 signal. This can be accomplished within Formula, using the following syntax: $errorCond1.removeLongerThan(1week).transform($capsule -> $capsule.setProperty('can', $container1.average($capsule))) 4. Repeat Steps 2&3 for Container 2 a. Filter the condition created in Step 1 to only include capsules with errors of interest for Container 2. This can be accomplished in Formula, using the following syntax: $errorCond.keep('value',isEqualTo('45')) b. Assign a ‘container’ capsule property to each of the capsules in the Error Cond 2 condition that indicates the value of the Container 2 signal. This can be accomplished within Formula, using the following syntax: $errorCond2.removeLongerThan(1week).transform($capsule -> $capsule.setProperty('container', $container2.average($capsule))) 5. Combine the error conditions with container property to create a single, composite condition. This can be performed in the Composite Condition tool. 6. Finally, create the histogram based upon this composite condition.
  2. Users frequently ask if it is possible to plot multiple y-axis variables against a single x-axis variable in scatterplot. While this functionality is not currently available, it is logged as CRAB-10474. However, as a work around, Organizer topic can be leveraged to display scatterplots side-by-side to compare the relationships of multiple y-axis variables against a single x-axis variable. An example of this is shown in the following image:
  3. Users frequently ask if it is possible to export a scorecard to Excel. While the functionality to export directly from Scorecard is not currently available, it is a commonly requested feature that is already logged as CRAB-15132. This post documents the current work around. Let's start with the following Scorecard: If we switch to trend view, we can see that the Scorecard metrics are displayed on the trend. The export button above the trend only exports signal and condition data, not metrics. However, we can use the Signal from Condition tool to create signals that display the same results as the Scorecard metrics. Once the results are calculated as signals, the Export button can be used to export the signal and condition data displayed on the trend.
  4. Hi @MarcS- I suggest submitting a support ticket so that one of our support engineers can investigate the issue. To do this, please submit a copy of your Seeq server logs: https://seeq.atlassian.net/wiki/spaces/KB/pages/114395156/Viewing+Logs+and+Sending+Log+Files+to+Seeq. Thanks, Lindsey
  5. Hi Jemma- I made a mistake in my previous post; it has been updated to show the correct Formula. Thanks for pointing that out 🙂 Lindsey
  6. Hi Jemma- You can achieve this by adding another property (i.e. 'sub-category) to the conditions in Formula: Please let me know if you have any additional questions. Thanks, Lindsey
  7. Hi Jemma - To do this, you will need to create a separate condition for each subcategory. There are several ways that you can do this. One simple way would be to use the Custom Condition tool to manually select capsules in the Pump condition for each sub-category condition: Once you have the sub-categories identified as separate conditions, you can combine all of the cause conditions just as we did in Step 2. Please let me know if you have any additional questions. Thanks, Lindsey
  8. Background In this use case, process engineers are investigating a process during which downtime frequently occurs. They have already identified the downtime periods as a condition in Seeq. Additionally, process engineers have also identified the 3 leading causes (pump, compressor, and level) of downtime and created conditions for when these parameters exceed an alarm limit, causing the process to shut down. To identify the leading cause of each downtime event, engineers would like to determine which of the 3 cause conditions are present 1 hour before the downtime event and which event happened first. The following method can be used. Method 1. Create a condition for the 1 hour prior to downtime Process engineers are interested in identifying which of the 3 causes are present in the 1 hour prior to process downtime. To do this, we first need to create a condition for the 1 hour prior to downtime. This can be accomplished in formula using the following syntax: $downtime.beforeStart(1h) 2. Combine all cause conditions and assign capsule properties To perform the analysis, we need to combine all of our cause conditions into a one condition, Leading Events. However, to know which capsules in the Leading Events condition correspond to which cause, capsule properties must be used. This can be achieved using the following Formula: //Assign a property to each of the cause conditions $compProp=$comp.setProperty('cause','compressor') $levelProp=$level.setProperty('cause','level') $pumpProp=$pump.setProperty('cause','pump') //Combine the cause conditions $compProp.combineWith($LevelProp).combineWith($pumpProp) You can view the Cause property of each capsule in the Leading Events condition by adding it to the capsules pane. 3. Filter the Leading Events condition to only contain capsules that overlap with the 1 hour before downtime capsules. This can be accomplished within the Composite Condition tool. 4. Create a signal that reports the value of the 'cause' capsule property for each capsule in the Leading Events before Downtime condition. This can be achieved in Formula using the following syntax (Note: '4h' indicates a maximum interpolation of 4 hours between data points and may be adjusted as needed). $eventsBeforeDowntime.toSignal('cause',startkey()).toStep(4h) 5. Identify the first value of the cause signal for each downtime event. When we calculate the first value in the Cause signal, the result should overlap with the downtime events, so that the cause can be associated with the downtime. To do this, first create a new condition using the Composite Condition tool to identify the union between the 1 hour before downtime condition and the Leading Events before Downtime condition. Now, use the Signal from Condition tool to identify the first value in the cause signal. Note that it is important to place the result of this calculation at the end of the bounding condition so that the results can be associated with each downtime event. 6. Create a scorecard that lists the leading cause of each downtime event
  9. This post summarizes the Performance Loss Monitoring use case covered in the Advanced Analytics 101 webinar in September 2020. In this webinar, the use case was explored by addressing 4 key analytics questions: Why? What data is available? What method(s) can I use with the available data? How do I want to visualize the results? Why? A manufacturing company needs to track performance losses. If engineers are able to identify and quantify those performance losses, the results can be used to justify process improvement projects or to do historical and global benchmarking. The current method to do this involves retroactively wrangling data in Excel. This exercise is very time consuming, so developing a method to automatically generated monthly reports has the potential to save up to 1 week of valuable Process Engineer' time per month. This is time they get back to work on improvement projects and other value added activities. What data is available? For this analysis, two data tags are needed: the target production rate and the actual production rate. What method(s) can I use? Step 1: Identify time periods of lost production To identify the time periods of production loss, I first need to calculate the difference between the target rate and the actual reactor rate. This can be accomplished in the Formula tool. Now, to identify the production losses, I can use the Value Search tool to identify whenever the value of this new signal is greater than 0. Step 2: Quantify the total production loss The Signal from Condition tool can be used to calculate the totalized production loss during each of the Production Loss Events capsules. How do I want to visualize the results? Ultimately, I’d like to create a weekly report that summarizes the production per day of a given week. So in this case, I’d like to create a histogram that aggregates the lost production for each day of the week.
  10. Hi Felix- To do this, you will need to replace the missing data gaps in your signals with 0. This can be done by following the steps in the following forum post: Please let me know if you have any additional questions. Thanks, Lindsey
  11. Hi @Adam Georgeson- I am happy to report that this issue has already been resolved in Seeq version 0.49. Also glad to hear you like the new homepage! Thanks, Lindsey
  12. Hi Theresa- Glad you solved the issue. In previous versions of Seeq, the number of capsules displayed in Chain View was limited to 30. However, this limitation was lifted in Version 22.0.48. Thanks, Lindsey
  13. Christophe- This functionality was added in R22.0.48. Glad you were able to upgrade your Seeq installation and take advantage of the new feature! Lindsey
  14. Hi Omar- I recommend changing the signal to be step interpolated. This can be achieved through the Formula tool using the following syntax: $signal.toStep(1d) The maximum duration of '1d' indicates that Seeq should interpolate between data points spaced by 1 day or less. When you select a maximum duration, choose a duration that is large enough to interpolate between the gaps in your data, but not so large that it unnecessarily impacts that performance. Please let me know if you have any additional questions. Thanks, Lindsey
  15. Hi Adam- Here is a screenshot of the new home screen from a mobile device: I will note that there is currently a known issue when viewing the home screen in Safari (such as on an iPhone). This bug causes the list of folders under the Shared and All categories to appear jumbled. This may make navigation difficult when using Seeq on an iPhone. This bug is already logged as CRAB-20100. Please let me know if you would like to be linked to this item so that you will be notified of our progress towards correcting the issue. Thanks, Lindsey
×
×
  • Create New...