How to Remove Devices from a WinCC OA System - Options and Best Practice

[AngusETM]

Hi all,

I'd like to start a discussion on options and best practice for removing devices from a WinCC OA system. This could occur, for example, when a device is physically removed from the system the SCADA is controlling.

Of course deleting the panel instances from screens is easy and obvious, but what to do with the device's datapoint is a more difficult question. As far as I can tell there are broadly two options, which are:

1. Delete the Datapoint
Simply, deleting the datapoint.
Advantages: Quick and simple, removes any I/O conflicts for the import of new devices which re-use the I/O, ensures the device cannot appear in any queries used in scripting.
Disadvantages: The historical data of the device is no longer able to be returned in a dpQuery because the link between the datapoint name and the ID number is broken.
Risk: The datapoint ID could be manually re-used by somebody later on, which would result in both objects sharing the same key (datapoint ID) in the historical database.

2. Disable the datapoint
This would take the form of:
a. Disabling the individual datapoint elements.
b. Deleting or modifying the I/O configs so no conflict occurs when importing future datapoints which re-use the same I/O.
c. Writing all scripting queries to ignore removed devices via a datapoint element flag which is referenced in every query.
d. Ensuring all historical queries after the date the device was removed ignore the device, via a datapoint element of type 'date' listing the removal date.
Advantages: Allows historical queries for times when the device was active to still include the device.
Disadvantages: Much more complicated to remove a device, requires manually disabling multiple DP Elements per DP, as well as deleting the I/O configs, including "removed" and "removed date" elements in the datapoint type structure from the beginning, and writing all scripts to ignore removed datapoints.
Risks: Missing a step in de-activating the elements could cause some elements to still be active.


No doubt this problem has been encountered before. Are there other options or ways to deal with this (particularly in regards to historical queries)? How have you solved this problem in your systems, and what were any advantages or problems of the method(s) used?

[AngusETM]

I should add that for myself, so far I've simply deleted the datapoints. This has ensured that no scripting can reference them to cause problems, nor can the I/O have any conflicts. However, now historical reports for the devices can't be accessed. We're considering adding the datapoints back in to allow this for various reasons but that opens up a whole can of worms about modifying the scripting and so-on.