2 Replies Latest reply on Nov 4, 2010 8:47 AM by MichaelvdV@Atos

    In pursuit of the perfect plant vs in pursuit of the perfect pi system.

    wpurrer

      Because there is still a forum for pi - administrator missing, I'm asking the question here.

      As i know, Pat Kennedy wrote a book or a vision "pursuit of the perfect plant", I'm asking if the the same question  which is  asked in direction of the "customer or plant" is also asked into the "Osisoft or PI - System itself.

      What is the my baseline:

      As you know, i have a personal "pi" or PIMS (plant information mgmt system) vision in my company. (see below ... My Goal is...)
      At this point I'm really struggeling with my goal. We have in the past (last week)  a couple of problems:

      • missing data because of osi bugs or some various reasons..
      • a lot of "extra work" because of missing features... (Archive reprocessing...) ace and pi pe lack of features.


      And really don't see a solution from osisoft for it. (How to run PI effectivly ...)
      The EA didn't solve this problem (neither from a technical nor from a systematical approach).
      The prasentations (Architecture and Best Practices: Recommendations for PI Systems ) are only advertisment in the direction of collective and virtualisation and doesn't really cope  with the daily issue where i struggle.


      What happend in the past:

      On the technical side:

      • We created a AF & Notification based survialliance of the system (does the same and a little bit more like the PI Managed in terms it looks on the whole datastream)
      • We create a service which filter and forwards the pi - log messages from the server to our central mgmt. software.
      • We create a tool which automatically creates the tags out of the dcs configurations.
      • we create a replacement for ace and pe (sdk - based) - because to manage calculations  with this two is not solving problems, it just making new ones..
      • We monitor this calculations with AF and notifications.
      • a couple of plugins to improve the usability (the Matrix View is one of them)


      On the mgmt sid:

      • We installed a team of people which are responsible for the Interface Nodes (approx. 80 Nodes, my approach is that the dcs guy also has to take care of the pi interface and the tag configuration)
      • We installed a team of people which are working as key users.
      • We installed a team of people which are able to develop plugins, sdk-based calculations,.... mgm the server, call techsupport.
      • We create a installation template for the (pi server, terminal server,..) which are rolled out to 2 sites.
      • We create a user guide with about 60 pages and 5 times 2 hours lessons for the people who work with the pi.
      •    (Difference to osiguides is, that it takes consideration of our plugins, architecture,...)


      What will happen in the future:

      • Plugin for Excel that here also our "search - tool" can be used.
      • we have to find a solution how to work with reprocesssing the archives, backfilling
      • we have to improve the monitoring our "sdk - based calculations"
      • and we have to generate the AF - Monitoring of our interfaces automatically (more quality then manual..)
      • we are waiting for a couple of updates because of some major bugs...
      • we have to improve our error handling that the responsible user directly gets informed of an issue (for example pe - syntax errors, sdk calculation based errors.. directly forwad to the user who created it)
      • find ways to improve the pi - system through product mgmt. (participation...)
      • Solve some internal problems with the data synchronisation of our systems (we have a couple of problems, where we have the ufl interface replaced)
      • Find a way to get collective to work.


      This stuff i mentioned is only helping in getting the "perfect pi system" it doesn't bring any value in terms of the "perfect plant"

      Whats your opinion on how to get to a "perfect PI system" also in terms of my vision/goal.
      What are your daily problems you struggling with? (Lessons learned ? )

        • Re: In pursuit of the perfect plant vs in pursuit of the perfect pi system.
          Ahmad Fattahi

          You are posing a very interesting, important, and at the same time broad issue here. I liked it when you draw the comparison with a book .

           

          Part of it is definitely improving OSIsoft products for which we have always been striving. We have come a long way in terms of reliability and efficiency and there is still a lot more to work on.

           

          The rest would vary on a case by case basis. Depending on your organization's needs and constraints, specific solutions might apply to you. That's why we have our Center of Excellence to get involved and engaged in a specific configuration and recommend the best feasible solutions. For smaller problems, our regular Tech Support would provide the answer.

           

          I don't know if vCampus in its current form is the best venue to collect such feedback. Most of the community members are developers and the tools and forums are geared toward development. Would a customized User Community serve you better in terms of addressing your specific concerns via sharing experiences and lessons on certain problems?

            • Re: In pursuit of the perfect plant vs in pursuit of the perfect pi system.
              MichaelvdV@Atos

              I don't know if this is an exact reply on your post, but here are my 2 cents.

               

              I'm on the SI side of the PI system. Specifically, the developers side.

               

              In my opinion, what gives OSI's software great value is that it's extendible. You are presented with 'bricks' (the standard software, PI, AF, ACE, etc), and as a developer, you add the cement that build the house. Sometimes you have to cut the brick wall to create windows and doors.

               

              This has, in my opinion, worked out great. The big issue here is resources vs results. I think, that a team of experienced PI programmers can do just about anything (with certain limitations offcourse), to achieve 'the perfect PI system' in a specific situation. As resources are not limited, you will have to prioritize requirements, and always be 'one step behind'.

               

              With standardized software, you are hitting a wall (no pun) when features or bugs are presented. You can however try to circomvent the bug/lack of feature by developing your own solution. Again, here resources are maybe the issue.