4 Replies Latest reply on Feb 16, 2017 5:17 PM by tmcmanus

    AF as an alternative to ISA-95 using RDBMS

    eradwan

      Hi There,

       

      In my company we have a 10 years old MES app that's based on the idea of aggregating daily and hourly readings from PI into an Oracle DB while building an ISA-95 model using tables like equipments, UOM, materials, etc. The main apps here are Emissions Monitoring, Production Accounting, Validation for Online Analyzers and Operations Logs. We're seriously thinking to revamp this model and use an AF based solution. However, I have some worries that I need your help to clarify. One of the main tables we have in Oracle is the'Readings' table which contains millions of record but have a useful feature which is to allow certain user to validate certain readings based on their knowledge of operations, the question is how to implement this in PI/AF, where to store the validated figures and have an audit trail for that. Also, I'd appreciate it if you can share any materials of references for developing such kind of the mentioned apps on top of AF and whether we can proceed without a RDBMS or not.

       

      Regards,

       

      Emad.

        • Re: AF as an alternative to ISA-95 using RDBMS
          ernstamort

          Hi Emad,

           

          Sorry I don't have any external references for you, but in general I really do like your approach. I wrote some blogs about MES-PI integration:

           

          Historian and MES Integration

          and

          PI and MES: Equipment Model

           

          And you are correct: As long as you stay within the ISA 95 and 88 frameworks, there isn't a problem for the migration. The logic for your additional user requirement to validate a metric will be in your client app and stored as an attribute\property. If it is AF or your Oracle database the underlying structures are just SQL tables that are being accessed\modified\updated by your client app.

           

          If these are sensor/real time data you could also consider migrating them into the PI archive data and store the acknowledgments either in a separate tag or as an attribute.

           

            Holger

          3 of 3 people found this helpful