11 Replies Latest reply on Sep 15, 2011 12:39 AM by skwan

    AF Templates and Error Results

    s.grainger

      Hello,

       

      I am currently working on the technical side of an AF Deployment (still in its early stages) and we are starting to grind through template standards.  An issue has come up that I'm wondering if there is a way around.

       

      Templates are, from our client's point of view, one of AF's strongest selling points.  There will likely be 1,000's of instances of individual templates deployed through this server, the issue is that the source systems (PI Server, CMMS, etc.) will likely not serve all templated data points for all templated element attributes (points missing in instrumentation will likely be the biggest offender - and we do not want to create PI Points that do not map to instruments).  We expect that most points will map for most templates, but there will be misses.  The result when an unmapped attribute is displayed in a graphic (Element Relative Display) is a long error string - worse yet, these element attributes that have not been mapped can be searched by the client during ad hoc data analysis.

       

      This would not be ideal by any means in a production system, so I'm wondering if there is a way to:

       

      1) Change the long error string "Cannot retrieve PI Point ' ' for Attribute ' '."

       

      2) Enable/Disable attributes for individual elements so that they do not show up in AF2 Searches.  We would not want to change templates to accommodate missing data points, as this would result in several hundred templates (nullifying the advantage), and we would prefer not to delete element attributes, as physically, the attributes still exist (despite not existing in automation systems), and the templating can be used to describe naming standards if ever the points are automated.

       

      Default values can be used when points are unmapped, but this does not correct the issue of a user searching and finding the attribute and then not finding any valid data.  It would be far easier for clients to run the search and come up with only data points that are enabled or mapped.

       

      I would appreciate knowing about any similar issues that have been encountered or any best practices that have been developed.

       

      Thank you!

        • Re: AF Templates and Error Results
          Ahmad Fattahi

          Good points Sara. To clarify your point a bit more let me make some description. When some underlying PI Point doesn't exist I can clearly see your point. How about other cases where a point exists but has been bad for a while? By extension there will be formula attributes with similar underlying PI Points included. Would you like to handle them differently as well?

            • Re: AF Templates and Error Results
              s.grainger

              When talking specifically about PI Points mapped to AF, I don't see a problem with having error messages show up in graphics and other tools.  In this case, to me it says that the point is mapped, but right now it is offline or in error.

               

              This would be different than the mapping issue I have, as many points will have the equivalent of null mappings and being able to search and find them (and have them report an error) isn't helpful.

               

              Thank you for the reply!

                • Re: AF Templates and Error Results
                  cmanhard

                  You can remove the PI Point Data Reference from the individual attribue instances by setting the Data Reference to <None> in PI System Explorer.  Later, if you decide to add a point to this attribute, you can use "Reset to Template" to return it to the template's definition, or, just reassign the PI Point Data Reference and Point path.

                   

                  This doesn't clear up your search issue, however, the icon associated with the attribute will be immediately distinguishable in the search results.  

                    • Re: AF Templates and Error Results
                      mhalhead

                      Sara Grainger

                      Enable/Disable attributes for individual elements so that they do not show up in AF2 Searches.

                       

                      I've requested something similar on the PI Coresight beta. This would be very useful not just for attributes that have not be mapped but also for none user attributes (e.g. attributes which are used in point creation).

                      • Re: AF Templates and Error Results
                        Asle Frantzen

                        Chris Manhard

                        You can remove the PI Point Data Reference from the individual attribue instances by setting the Data Reference to <None> in PI System Explorer.

                         

                        Chris, I have a question related to that.

                         

                        I've had to do this myself recently, as I was adding a lot of tags coming from a Production Reporting System (PRS). My challenge was that not all assets had all of the PRS tags. I've removed the PI Point data reference, but then it shows the default value + UOM. I'm concerned that users might be confused and take this as a live value.

                         

                        I did try to make these stand out by writing "N/A" or "Unavailable", but then I get errors stating that the value type doesn't match. And if I try to change the value type (to string) a new error states that the UOM is not compatible with this value type...

                         

                        So I guess my question would be if there is any other way of disabling these PI Point dr's which have to be in the template, but sometimes don't exist at certain elements? My main concern is the end users.

                          • Re: AF Templates and Error Results
                            s.grainger

                            Asle, that was exactly what I was going to say - the "default value" parameter always has to be valid vis a vis the point type, which means that it always has to potential to be interpreted by a user as a live value.  This could be troublesome.  I agree that it would be far better to be able to essentially hash out an attribute of an element to prevent users from even encountering it.

                              • Re: AF Templates and Error Results
                                s.grainger

                                I hate to self-bump this thread, but I'm wondering if there is any enhancement on the horizon for this issue.  Changing the icon in attribute searches and having nonsensical data is not exactly the optimal solution.  Disabling an attribute would be ideal.

                                  • Re: AF Templates and Error Results
                                    cmanhard

                                    Sara, short term, the best solution I can think of is to create a data reference that specifically returns a value which you desire, but have it marked as bad (IsGood=false).  This will keep you from getting the default value or having UOM conversion applied, but will allow you to have your string ("N/A", Empty String, or No Data, etc) returned as the value.  

                                     

                                    I don't currently have a concrete answer for the longer term feature request - which is to have a property on an attribute instantiated from a template that allows you to disable/hide it - other than it is not on the very near term horizon (AF 2010 R3).  I will make sure that the request is discussed in AF planning.

                                      • Re: AF Templates and Error Results
                                        skwan

                                        Sara:

                                         

                                        A couple of questions for you if you don't mind.  How many PI Point DR's do you anticipate having and how many of these would not have any values?  I'm trying to get a sense of the scale of this issue and thus impacting how best to handle this issue.

                                         

                                        Thanks.

                                          • Re: AF Templates and Error Results
                                            s.grainger

                                            Hi Steve,

                                             

                                            I expect that we will have most PI Point data references populate, but given that we can legitimately have hundreds of attributes in an element, I would expect that misses would be on the order of tens to hundreds (we would probably want to test a solution with 100-200 misses in 1000 data references).  

                                             

                                            An option has been suggested to use an array of templates to cover off on this issue, but unfortunately the sheer number of templates we would have to create would make templating untenable.

                                             

                                            Please let me know if you have any other questions, I would be happy to talk out solutions.

                                             

                                            Thanks.