Sunday, January 15, 2012

How to define New Data Objects in IEC 61850?

The need to define new data objects is likely to have various reasons. One reason is that experts do not know which logical nodes and data objects are already defined. Let’s assume there is really a need for a new data object – there is not any data object that may fit.

Example (The following LN SIML, I found on the Web):

The standard LN SIML (Insulation Medium Supervision) provides the data object H2ppm (Measurement of Hydrogen (H2 in parts Per Million)):


There is a need to model H2ppm related semantic, e.g., “Hydrogen ppm Rate of Change” or “Hydrogen ppm Rate of Change Goodness of Fit”

These two semantic models are not defined in the standard. What is the best way to model these two?

  1. Defining values in GGIO? – maybe not,
  2. Defining new data objects in SIML? – may be the best solution (could be standardized later), or
  3. Defining something like H2ppm1 (measured value), H2ppm2 (rate of change”, and H2ppm3 (roc Godness of Fit)? – That is definitely wrong!

Why are the following data objects in conflict with the standard modeling method?

Here is the definition for LN SIML (of the example I found) using multiple instances of H2ppm, defined in the Insulation Medium Supervision (Product Specification):


The standard IEC 61850-7-4 Edition 2 defines the LN as follows:

LN: SIML Name: Insulation Medium Supervision (Standard IEC 61850-7-4 Edition 2):


These data objects H2ppm1, H2ppm2, and H2ppm3 are not allowed – it is not allowed to instantiate data objects (with some exceptions, then the data object in the standard LN needs to be defined as MyDataObject1 – with a “1” at the end)!

For details on instantiating data objects see the following excerpt of IEC 61850-7-1 Edition 2 that defines the extension rule for data objects:

14.6 Specialisation of data by use of number extensions

Standardised data names in logical nodes provide a unique identification. If the same data (i.e. data with the same semantics) are needed several times as defined, additional data with number extensions shall be used. The rules for number extensions shall follow the naming conventions defined in IEC 61850-7-2 and be as follows:

  • the number extension usage shall only be defined by the owner of the data namespace. This shall be done by adding the number extension 1 to a data object name (e.g. data1),
  • data with no number extension shall not be extended by third parties,
  • data with the number extension 1 can be extended. Number extensions may be ordered or not (1,2,3,4, or, 1,2,19,25),
  • if only one instance of an extendable data is present in an LN, it shall have the number extension “1”.

14.8 Example for new Data

New Data “Colour of Transformer Oil”


The above figure shows also that a data Namespace Attribute “datNs” has to be specified for each new data object.

For the above listed additional semantic it would work with the following (standard conformant) extended data object definitions:

Example (wrong – semantic is in instances):
H2ppm1 (measured value)
H2ppm2 (rate of change)
H2ppm3 (roc Godness of Fit)

A standard conformant solution is (define new data object classes):
H2ppm (measured value)
H2ppmRoc (rate of change, extended data with datNs=Vendor so and so )
H2ppmRocGdns (roc Godness of Fit, extended data with datNs=Vendor so and so)

Please find further presentations on model extensions:
Click HERE for post1.
Click HERE for post2.


Iqbal said...

Dear Schwarz,

Thank you for such interesting article. My name is Iqbal, a Researcher from an Electric Utility company in Malaysia. If you don't mind, I have a few questions related to the topic:

1. As a user, are we allowed to configure the data namespace using a client and/or ied proprietary tool? In my opinion, such feature will give the user the flexibility to utilize the signals, say for analysis purposes (e.g. using engineering workstation).

2. We have also encountered IED which only uses GGIO1 to map the MMS signals. Different MMS signals are segregated using data object Ind1 until Ind128. According to your article, this implementation seems 'wrong'. Can you comment further on this?

Karlheinz Schwarz said...

Dear Iqbal,

ad 1. It all depends on the tool of the IED vendor. Some are more flexible than others. The SystemCorp ICD Designer allows to define any extended data objects and logical nodes.

ad 2. GGIO is - as I usually say - the most hated and most liked LN! GGIO should be used if there is no other LN that offeres the semantics you need. You may also define a new LN for these signals ... It is no a question of right or wrong. It is more about doing it in the sense of semantic modeling or just do it as people have done it for decades.