March 1, 2016

Trace individual runs


Rich post-processing capabilities found in pSeven are one of its key features. It is worth noting that you can easily store, trace, and visualize results of several workflow runs. The usage scenario described here illustrates this concept.

Assume that a simulation expert creates a workflow that models a static mixer - a device for continuous mixing of two flows of liquids with predetermined temperatures (figure 1). Then it is up to a domain expert to explore design alternatives using this simulation. The goal is to find such geometry of the mixer (diameter of the first inlet nozzle and angle of the second inlet nozzle) and velocities of two inlet flows that provide the minimum temperature of the mixed liquid in the outlet nozzle. In addition, the problem is subject to the following constraint: the pressure drop in the outlet nozzle (difference in pressure values between one of the inlet nozzles and the outlet nozzle) should be no greater than a given value. 

Figure 1. Static mixer             

Every workflow allows user to adjust its parameters in Run (figure 2) without switching to Edit. So the aforementioned domain expert can use pSeven Runner to accomplish his task without locking a license for editing (read more about pSeven Runner here).

  Figure 2. Static mixer Run

Workflow input values can be specified on the Inputs pane on the top left (figure 2.1). Parameter values can be adjusted on the workflow configuration pane (figure 2.2). Often used parameter configurations can also be saved to presets (figure 2.3), so you can quickly switch between different configurations without re-typing the parameter values.

After all parameters are set, the workflow can be run to get results. To manage and organize these results, you can assign a name to every run (figure 2.4). By default pSeven uses a different name for every run, generated using the current date and time. However this behavior can be changed in order to obtain a lightweight database containing results of multiple workflow runs, as explained further.

In pSeven, workflow results are post-processed on the Analyze screen. Stored results can be viewed on the Project database pane on the left; this pane shows results grouped by their run names and an additional <last run> shortcut (figure 3.1). The shortcut is convenient to see the most recent results only, while default names containing a date and time stamp (figure 3.2) help to keep track of the results when they start to pile on.


Figure 3. Static mixer Analyze

 Note that results grouping is entirely based on the run name you specify on the Run screen. It means that when you launch a workflow with the same run name a few times, all results are stored to the same group in Project database.Inside such a group, results are sorted by the data source (port name), so you can clearly see how values of workflowvariables change when it is run with different inputs or parameters. It also makes easy to analyze results of different runs by adding data of interest (figure 3.3, highlighted) to a report database and using any of the wide variety of visualization techniques. In this example, input and monitoring values from 28 runs are grouped into corresponding data series (figure 3.4), which are used to build two 3D surface plots.  Here is a more detailed view on the run result representation (figure 4). Bringing result values (figure 4.1, green), input values (figure 4.2, blue), and some auxiliary data (figure 4.3, orange) into the same table (or any other representation form) might prove illustrative and helpful for any domain expert working with pSeven.



Figure 4. Sample Viewer

As someone who works with pSeven on a day-to-day basis, I find these results tracing functionality quite useful in my line of work, i.e. software testing. And I'm sure domain experts also see how convenient it is, to organize results for post-processing with pSeven tools.

Sergey Tyuterev, QA Engineer, DATADVANCE