Welcome to Capsules! And Welcome to the Innovation Experience in Seeq.

Written by Jon Peterson on May 26, 2016

Schneider Electric’s Automation Conference in New Orleans was a great week for food (of course) and Mardi Gras experiences, we didn’t leave anyone on Bourbon Street, and the event provided many positive interactions with Process Automation employees and customers. This conference featured the backbone of Schneider Electric’s automation offerings, transformative products in the history of the industry like Foxboro, Modicon and Triconex, highlighting connected products that offer better intelligence, and the apps, analytics and services driving IIoT.

One customer we spoke with has a 28-year-old DCS and historian infrastructure, a system that even preceded AimStar, the legacy Foxboro historian. He wondered if Seeq could enable insights on their data and, well, we’re looking into it. A very good event with great seminars and perspectives from Sandy Vasser of Exxon, motivational speaker Mike Lipkin, Gary Freburger of Schneider Electric, and others.

One thing the event reminded us is that we need to provide a basic overview of “capsules.” Capsules are a Seeq innovation that customers haven’t seen before and since we have been using them for years in customer projects we sometimes forget that. So here is a primer on capsules: a data object that is at the core of the Seeq investigation and discovery experience.

What is a capsule? A capsule is user-defined slice of time with a unique identifier. And most importantly you, the user, get to define the capsule according to what you want to investigate or analyze. Capsules include a start time, an end time, a name or handle, references to the expression that defined it, and perhaps other properties that you add to it.  They can be singular, as in “here is a capsule with data from an asset during its start-up phase on June 11th.” Or they can be in a collection as in “here is a capsule series with data from an asset’s start-up phase for the past 100 days.” Once defined, capsules may be overlaid on top of each other for visual comparison, or lined up side by side, analyzed with over 20 Boolean operators, or used as variables in calculations. In this diagram below, a capsule series of 3 capsules is being defined because they share a particular shape, which has been identified using Seeq’s Pattern Search Feature, and then aligned for easy comparison in “Stitch View” and “Capsule Time."


With that as a general introduction, let’s dig a little deeper. A site, asset, or process goes through many states or conditions over time and at any given time more than one state or condition can exist. Some examples of these states or conditions are:

  • Day of the week
  • Shifts
  • Periods of hot temperature (temperature > Y)
  • Periods of low pressure (pressure < Y)
  • Startup, production, shutdown phases

These are examples of context - what was happening during the sensor reading - sensor data in a historian is almost meaningless without context. Capsules provide a way to add context to sensor data.

As an example, let’s use weather data and add context that defines Sunday as the time period of interest, so that the capsule group contains a collection of capsules that have the data for Sundays. First, we use a Seeq function (like the Excel function library) to create a capsule series. This function doesn’t apply to a specific time series since that argument isn’t included; instead it simply provides a capsule group for the time period defined. You could also use this create capsules that define hours or minutes or shifts (8 hours, starting at xxx AM or PM) or production blocks, etc.

Here, we use the formula: days ( day.Sunday )

Here’s the result as it looks in Seeq Workbench, showing the equation and the resulting capsule series (the bars along the top of the trend viewer):

Formula: days ( day.Sunday )

In the details pane (lower right), all the data for the displayed capsules and statics such as duration, max, and min for the two displayed series show up automatically. So a capsule can be used to filter a time series, like a window onto just the time periods of interest.

Next we can change views to look at data for just Sundays sequentially. This is Chain View with the Sunday-defining capsules lined up side by side for easy comparison.


And then Capsule Time overlays the data from Sundays. The time-axis is time into Sunday rather than calendar time, which makes looking for drift or outliers in sensor data easy.


It’s pretty clear from this view that as temperature goes up, humidity goes down.

Finally, as an example of the calculations that are possible with capsules, here are several aggregations over the capsules. We are using a feature called “Series from Capsules” to create a brand new time series based on data from the capsules. The 4 new lines represent:

  • Maximum Temperature for each Sunday
  • Minimum Temperature for each Sunday
  • Maximum Humidity for each Sunday
  • Minimum Humidity for each Sunday

Capsule Series

These new series, created from the “Time Series from Capsules” feature, are an important tool for calculations and historical analysis since they can be generated from capsule series spanning multiple types of context as well as Boolean logic or calculations across those capsule series. And any new time series, created from all of these complex factors, can be visualized and used for calculations and analysis just like any other time series.

Welcome to capsules! And welcome to the innovation experience in Seeq.