Comparing benefits and returns of APC/Optimisation: site wide approach or unit approach

I have been listening and reading up on the benefits and return of APC/Optimisation. Seemslike philosophically there are two approaches - site wide top down umbrella or unit by unit approach....anyone wanting to shed some lights?

  • You can check possibilities on the next link:

    www2.emersonprocess.com/.../index.aspx

    www2.emersonprocess.com/.../IndustrialEnergy.aspx

    Links above give an idea what can be "optimized" in refining industry. There are always 2 main challenges ( regarding production itself ) - optimization for production / throughput / energy efficiency ( process optimization - MPC / APC ) and keep production / process running ( process equipment issue ). There is also necessary to take into account SIS side / implemetation.

    There can be next separate topics which can improve plant operation costs like energy management, process safety, auxilliary processes ( steam / heat / energy generation ) and so on.

    From my point of view is the right way indentify opportunity first ( thru audits for customer plant present situation ) - means opportunity table with solutions which can improve situation thru any way. Table can be sorted with "the highest ROI solutions" on the top. Opportunity for saving costs isn´t always in process optimization, significant savings you can rich also thru reduction downtime / shut down or practices which can impove response / repair in case of shut down ( because of regular maintenance activities and so on ). Complex point of view for possibilities is better that focus on one application and sometimes there are interactions between applications which can improve situation cross the plant.    

  • In a refinery, plantwide optimization in real-time is not feasible.  That is a function for planning and scheduling.  One is dealing with longer term economics and very simple constraints vs short term economics and many contraints.  The minute that one tries to optimize multiple units (a plant) together one is faced with intermediate storage (building or drawing of stocks) and different yield requirements for each unit that needs to be reconciled as an overall entity.  It becomes a multi-period optimization problem which is not really feasible to do in a real-time fashion.  Furthermore, many refinery control rooms are segregated by areas - one operator has responsibility for an area.  How does one optimizer multiple units that are the responsibility of different operators - end result one operator who doesnt understand what is occurring will put the optimization in manual mode?  A refinery control room is not staffed by mere mortal operators not Process  Engineers with optimization knowledge and expertise.