DCS-APC (Advanced Process Control) a better and superior option to MPC (Model Predictive Control) in olefins plants

In recent times, most olefins plants crack ethane that is more readily available than a few decades ago.  Olefins plants built during the 1970s to 1990s were often more complex.  The olefins plants were equipped with more complex furnace configuration than today.  The older olefins plant furnaces were designed for a lot of flexibility; they could process ethane, propane, butane, light oils, heavy oils, naphtha, HNGL (heavy natural gas liquid) and kerosene.

The furnaces had multiple coils, typically six coils.  It was possible for four coils to crack ethane and two coils to crack butane.  Or even in certain cases, four coils cracked propane and the other two naphtha or kerosene.  The furnace severity control became a complex calculation as there was significant optimization to do, in order to run the furnace in the best conceivable way.  The “best possible way” could change, depending on whether the plant needed to maximize ethylene yield, or propylene yield, or control the severity at some optimal value, which was also impacted by the steam to hydrocarbon ratio and numerous other factors. 

The optimal way to run the furnace required the use of an olefins plant furnace model.  ABB Lummus back in the 1980s and 1990s successfully provided an excellent furnace model for olefins plant optimization.  Also, TechnipFMC offered Spyro furnace model.  These furnace models were able to examine the feedstock options and then determine the severity conditions needed for optimal conversion, while considering the goals and objectives of the plant – whether to maximize furnace run-length, minimize/delay coking, maximize ethylene yield, propylene yield or any other requirements. 

When raw materials other than ethane are cracked in an olefins’ plant furnace, the different effluent gases generated are a lot more than with simple ethane cracking.  In a simple ethane-only olefins plant furnace, there is no propylene generation and subsequently, you do not need a propylene fractionator, also called the C3= Splitter. 

The modern olefins plant with ethane-only feed are a perfect candidate for a DCS-based APC.  DCS-based APC is implemented all inside the plant’s existing DCS using standard and custom function blocks and some DCS-based logic.  The DCS-based APC comprises of traditional APC also called traditional advanced control.  DCS-based has the following widely used acronyms – TACS (Traditional Advanced Control System), ACAP (Advanced Control Applications Package), CLCC (Closed-Loop Computer Control) and MBC (Model-Based Control).

DCS-based Advanced Control began gaining traction during the late 1970s and early 1980s after computer control began to be applied for advanced process control in chemical plants and oil refineries.  But then, the amazing and resounding success by Shell Oil Company’s Dr. Charles (Charlie) Cutler’s DMC (Dynamic Matrix Control) algorithm resulted in a complete reversal of the DCS-based APC applications as people began to think that DMC and MPC in general was an amazing Panacea for all industrial process control. 

While DMC and MPC in general are excellent for truly multivariable processes, only about 20% of all industrial processes are truly multivariable.  An FCC (fluidized catalytic cracker), hydrocracker, crude distillation, main fractionation – in an oil refinery are truly multivariable, as many MVs (Manipulated Variables) affect many CVs (Controlled Variables) and the control matrix density is over 50%.  Control Matrix density is defined as the number of MV-CV pairs in a control matrix that have a non-zero process gain divided by the MV x CV.  In most chemical processes, this matrix density is less than 30% and these processes behave in a serial manner, the process flows from one equipment to the next very much in series and there is not really much multi-variability.  When the control matrix density is less than 30%, DCS-based APC can be easily implemented without the need for a full-blown MPC.

Any modern DCS – Honeywell TDC300 and Experion, Yokogawa Centum VP, Centum Excel, Emerson DeltaV, Foxboro IA Intelligent Automation, ABB OCS, Bailey etc., and many new PLCs from Allen Bradley Rockwell Automation, Siemens PCS7 and Siemens Tia Portal, and others provide the necessary standard and custom function blocks needed to easily develop a powerful DCS-based APC without the need for an MPC.

If you use a MPC (Model Predictive Control) system from Honeywell (Profit RMPCT – Robust Model Predictive Controller Technology), Aspen DMC (Dynamic Matrix Control), Yokogawa-Shell SMPC and Yokogawa PACE, Emerson PredictPro, Foxboro Connoisseur and any others commercially known, the price tag is extremely high and the annual maintenance fees are also extremely high.  If you use a DCS-based APC approach, the initial cost will be 50% or 30% of a full-blown MPC project and the annual cost will be one-tenth of an MPC.

Why is it that very few companies are implementing MPC at an enormous cost and not adopting the cheaper and better (even superior) DCS-based APC?  For many processes, DCS-based APC can outperform MPC and DCS-APC can even produce higher monetary benefits compared to an MPC.  Very few people are aware of this.  Many MPC and DCS vendor companies do not want to advertise the DCS-based APC approach as the profit margins are much higher with an MPC sale.

New process control engineers entering the control room only know academic theory based on Laplace domain and Discrete Domain (Z Domain), they are familiar with Matlab and Simulink from MathWorks, but none of these tools and none of this knowledge are of any value in the practical control room environment.  Very few process control engineers have the skills or the tools to identify transfer function models using process data from Excel files.  In addition to this required knowledge, the control engineer needs to know how to design and implement the standard and custom DCS/PLC blocks correctly in order to produce the desired control action.  Many of the above skills are just not there in both new and experienced process control engineers.

If you are in the market for APC or MPC in an olefins plant, please contact PiControl Solutions LLC first.  PiControl can help you design a DCS-based APC for your Olefins plant.  The DCS-based APC can be applied to both the furnace area and also the cold distillation area.  Equipment covered by the DCS-based APC will include all olefins plant furnaces, demethanizer, deethanizer, C2 splitter, C2 refrigeration, C3 refrigeration and cold box.  Additional APC schemes can be applied to various other main and peripheral areas. With DCS-based APC, you do not need new computer servers.

You do not need new software.  You do not need training on MPC using a 1000-page manual.  You do not need OPC DCOM, OPC UA, OPC Tunneller, you do not need your engineers to spend a lot of time and effort on IT challenges, firewalls, antivirus blocking and a whole list of headaches caused by Windows server operating systems, firewalls and the rest.  All you will need is to make sure you have enough capacity in the DCS for new standard and custom blocks and the cost of this is very less compared to the MPC route.  Contact PiControl today – www.picontrolsolutions.com, email [email protected].