1072   Statistical Data and Reporting

Created: 25 Mar 2013

Status: In Force (green)

Part: Part 7-1 (2011; Edition 2)

Links:

Page: 43

Clause: 6.4.4

Paragraph: last before 6.4.5

Category: Issue for edition 2 of this part

Issue

The clause defines:
"For periodic calculation, the “event” ClcExp set to TRUE can be used as an event to report the new value (the statistical value) by the report control block or it may be logged as historical statistical data for later retrieval."

The ClcExp needs to be a member of a DataSet to be used as an event. But in this case only the ClcExp data object would be reported according to RBE (Report by exception). So, the "the new value (the statistical value)" cannot be reported.

We cannot use the concept of Statistical Data to send a set of values on a cyclical base AND start the cycle at the full hour! The ClcStr of CDC SPC (Enables the calculation start at time operTm from the control model (if set) or immediately, cannot be used because of the RBE definition of reports.

rptEna of the report control block could only be used to start, when a client sents a "enable=true" ... which may be received 5 seconds before the full hour or 10 seconds later ...

Proposal

Define a possibility that a set of values of any number of data set members can be sent with a report.

The report usually is carrying a single member's value (RBE, dchg, qchg, or dupd), ALL members in case of integrity period (intpd) or general interrogation (gi), or a set of values of a set (less than ALL) of members in case of bufTm > 0. So, a client has to manage any set (1, many, all) of members' values.

It should at least be possible to send ALL member's values based on an event of any member's value change.

Discussion Created Status
25 Jun 13 In Force (green)
I propose to remove the text and instead to modify the description of figure 28 (4th parag. on page 48).

Change "The lower part of the figure shows the conceptual model of the historical statistical data. In this model, the calculated values (in this case the maximum values with calculation mode set to PERIOD) are stored in sequence in a log. …"

To something like

"The lower part of the figure shows the conceptual model of the historical statistical data. In this model, the values are calculated periodically (in this case the maximum values with calculation mode set to PERIOD) and are stored in sequence in a log. The expiration of the statistical calculation is indicated by the “event” ClcExp set to TRUE and at this moment the statistical values that have the trigger option dupd are logged even if the values have not changed. The same concept can be applied to the reporting of statistical data. …"

Modify fig. 28 accordingly.
31 May 13 Discussion (red)
At the end of a statistical Calculation (CalcExpired), every data update values will be reported.
To make sure that they are all reported in the same report - use a bufTim value for which you know that the applciation will have updated all the calculation output, and optionally a dedicated DataSet.

24 May 13 Discussion (red)
This issue was discussed years ago with the result, that it is problematic to define the synchronicity of values belonging to a certain event inside a report independent from the stack implementation. Instead of adding something to reports, the LN GLOG had been introduced, which should have a defined synchronicity for the data values it logs based on some trigger. Unfortunately the result is not reported.
I agree that the description in 7-1 needs to be enhanced. ClcExp can only be used as trigger to force a read of the statistical value, especially if this value has not changed. Alternatively, statistical values could have trigger option dupd, generating an event of them at the end of ths statistical calculation period even if the value does not change.
26 Mar 13 Discussion (red)
Of course, this would have an impact on IEC 61850-7-2 and -8-1, too. 25 Mar 13 Triage

 

Privacy | Contact | Disclaimer

Tissue DB v. 21.10.16.1