• Replies 7 replies
  • Subscribers 26 subscribers
  • Views 99 views
  • Users 0 members are here
Options
Related

AF hierarchy best practice in Upstream O&G

César_CF_PLU

Good afternoon guys,

 

Just wondering if you are aware of some standards or best practices already defined regarding the most efficient way to structure an upstream asset hierarchy in AF. I mean, we could organize an AF tree by following a "choke model" (4 groups of reservoir, wells, facilities, delivery), or simply following a field's "natural" organisation (field - north/south region - cluster a/b/c - wells/pumps/vessels/etc - pressure/temperature/etc), or even alphabetically (then a naming convention would be a key issue to resolve first)...so just wondering if some arrangement type has already been identified as optimal for our business. Considering the amount of re-work that would be involved in modifying it, it seems no trivial decision...or is it?

 

Thanks and regards.

  • Hi,

    I'm firmly of the opinion that there isn't "an" optimal structure for organising things in AF (and this is for any industry).   The most optimal configuration is the one that solves your problems for you.

     

    That may mean you have more than one tree (or even database) in AF that arranges things differently in order to solve different problems.  For example you may arrange wells, etc. by field because you want to get insights into field related problems or you might organise things by geographic (or other grouping) because you want to get insights into energy consumption related problems.  You can minimise the work involved in maintaining that by using the "weak reference" model  to build trees - but that doesn't always work very well and you just have to accept a certain amount of duplication.

     

    I've seen many clients try and build a "one size fits all" structure and they end up spending so long chasing that "holy grail" that they don't solve any business problem or having sub-optimal solutions to different problems.

  • Hi César,

     

    Like Ian mentioned, your AF approach should be driven by a business issue or use case. What are the problems you are trying to solve? Start small, targeting specific issues, and go from there. Don't try to model the whole plant or enterprise just with the objective of modeling it.

     

    See the PI Square thread linked below, there are some good posts there about the questions you need to ask first when planning your approach.

     

    Re: PI AF

  • Thank you guys for your quick replies.

     

    Absolutely, no size fits all, won't argue with that. I was just wondering if history has taught us some PI AF lessons that could be generally applicable . Something in the lines of "if your objective is this, be mindful of structuring your AF tree at least like that..." or "while it won't fit any specific needs, this is a generally applicable upstream hierarchy...". Having/testing several trees is a valid approach, and probably the right one for smaller assets, but developing and maintaining them for complex assets could imply a huge amount of work...just guessing here.

     

    On the other hand, (real-time) production optimization is a pretty generic objective when in comes to digitizing an upstream asset, so it could be in that regard that I was wondering if something with more pros than cons had been identified already; or more generally speaking, if some lessons on how to actually build the tree (an upstream-specific naming convention?) had been gathered and documented somewhere.

     

    Thanks and regards.

  • in reply to César_CF_PLU

    César, maybe the AF Kits could be a start. There is one for Drilling & Completion and another for Well Downtime Tracking.

     

    Asset Based PI Example Kits

     

    There is also an Upstream User Group, I am not sure if you're a member already. This is the kind of information that is shared within those groups.

     

    Oil and Gas Upstream

  • Cesar

    Look at this presentation from BP at our lastest PI World event in Barcelona: https://www.osisoft.com/Presentations/BP---Using-Analytics-in-PI-AF-to-Improve-Operating-Performance/

    Download the presentation slides and check #5, it shows a sample of their AF hierarchy structure..

  • You could also take a look at ISO 14224 and the taxonomy that they recommend for oil and gas assets.

  • in reply to Ian Gore

    Hi Cesar:

     

    This is great question.  As pointed out earlier, there is no one-way to define the way to build the PI AF hierarchy with the required Analysis, Event Framing and Notifications.  The PI AF hierarchy depends on the Business Process that we need to solve.  There some standards that can be adapted such S95 and others.

     

     

    We have provided some ideas for the process industries in a draft that you get (A Journey towards a Digital Transformation (draft C1-4) in PI Square).  Here, we present a fictitious story of a person that get in charge to implement an operational excellence strategy for their enterprise.  One of the first things that they do was to analyze the current business processes in their plant.  He finds that the planning and operations there is variance huge gap between projected production targets and actuals.  Based on these findings, they designed a Digital Plant Template to solve this particular goal.  The message for you here is to identify one of key problem that you would like to solve.

     

    The Digital Plant Template uses a UNIT Template with the proper metadata (parameters), analytics and event frame generation to transform raw data into operational insights. This is idea is coming from the work that I started in my previous company for monitoring wells.  These are running on target, in troubles, idle. down or maintenance.   Wells were classified according to their life (new, normal, trouble, old).

    In this story, they used the S95 general hierarchy to model a process plant and we included the event frame generation to aggregate the data into meaningful information based on operational events.  The events framed data is then visualized using PI Vision and Microsoft Power Bi (desktop and AZURE).  We created a TechCon Hands-on Lab with a step-by-step procedure to get a plant simulator example to learn from it. (Search for Digital Plant Template and you can find in PI Square). Attached is one of the versions for you.

     

    The suggested strategy starts from the top production level for each unit. Then, it goes inside the unit to define the quality for the each unit and start building soft sensors.  The strategy is to find the operational insights that identify the production constraints and to build predictive models.

     

    You could look at the example and see if you can adapt to the upstream asset hierarchy that you have in mind. 

     

    Warm Regards,

    Osvaldo

    A Digital Plant Template for transforming raw data into Operational Insights.

    JourneyDigitalTransformation DRAFT C1-4 Binder1.pdf