Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations cowski on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!
Plant Optimisation 1
- Thread starter LEKE316
- Start date
-
1
- #2
Get a course on SPC.
There may be some easy things that you can find just by listening to your plat operators. They know where they waste energy and ehre they have reruns. This takes getting to know them and building a trust. Work on their problems first. Once they see you are willing to work with them, they will help with the tough things you need to put in place after you learn Statistical Process Control (SPC). Read a Dr. Demming book and use the steps he recommends from Dr. Juan.
It's a lot of fun and you get to meet some new people.
There may be some easy things that you can find just by listening to your plat operators. They know where they waste energy and ehre they have reruns. This takes getting to know them and building a trust. Work on their problems first. Once they see you are willing to work with them, they will help with the tough things you need to put in place after you learn Statistical Process Control (SPC). Read a Dr. Demming book and use the steps he recommends from Dr. Juan.
It's a lot of fun and you get to meet some new people.
Check under chemical engineering programs, Aspen Products
You could try Tracerco, a company that provides many online optimisation techniques, the link attatched is specific to FCCU optimisation
Take a look at this book.
HTH, dogtop
HTH, dogtop
I agree with dcasto. Your operators are going to be the ones that know where all the bottlenecks are. However, most of the times they will describe the symptoms to you, and you as the engineer, have to find the root cause.
Most of the operators at my plant have "theories" of why things are, and sometimes they're right, but other times they're just close. Still, very helpful.
Most of the operators at my plant have "theories" of why things are, and sometimes they're right, but other times they're just close. Still, very helpful.
UmeshMathur
Chemical
Several steps are required. I would summarize them as follows:
(1) Expert analysis and fix-up of all basic regulatory controls - loop configuration logic, instruments, analyzers, loop tuning, valve performance and so forth.
(2) Model-predictive, multivariable constrained control, referred to as MPC, e.g., with DMC Plus, RMPCT, etc. These tools also perform approximate unit-level real-time optimization. Several thousand applications of this technology have been implemented at many hundreds of petrochemical and refinery units worldwide for the last 30 years. This technology is very beneficial and makes an immediate and noticeable difference in the way a major process unit works. It sits on top of the regulatory controls and generally writes all the regulatory controller set-points every minute. Therefore, that lower control layer must function flawlessly to maintain MPC benefits. An ethylene plant typically requires 15-25 such MPC applications, depending on the number of cracking furnaces. A typical refinery unit would require anywhere between 5 to 10 MPCs, sometimes more.
(3) Add rigorous, closed-loop, real-time optimization (CLRTO) as the next layer above MPC. This requires plant-wide chemical engineering models to run under a non-linear optimizer. Examples of such products are RT-OPT and NOVA. Examples would be one CLRTO for a complete ethylene plant, or a cat cracker in a refinery. CLRTO uses "open equation" solution methods and has been used for over 15 years in ethylene plants. This technology is better suited for petrochemical processes than refinery units, in my opinion, mainly because of uncertainties in feedstock composition in most refinery units. At present, again in my opinion, the traditional steady-state simulators are incapable of doing serious optimization work at a plant-wide level.
There are many hundreds of published papers describing the MPC and CLRTO technologies, some going back over 30 years. I would recommend you study a few of them carefully before calling the vendor companies. There is no panacea for maximizing profit, and it takes many years of extremely diligent work by a very large team of specialists in each plant before all this works properly.
Of course, the cost runs into the millions of dollars and you need some high caliber staff to maintain this stuff, but the paybacks are generally within two years for each major piece, many times a lot sooner. A huge bonus is much better staff training, in many areas, and an inexorable focus on control systems performance and maintenance.
(1) Expert analysis and fix-up of all basic regulatory controls - loop configuration logic, instruments, analyzers, loop tuning, valve performance and so forth.
(2) Model-predictive, multivariable constrained control, referred to as MPC, e.g., with DMC Plus, RMPCT, etc. These tools also perform approximate unit-level real-time optimization. Several thousand applications of this technology have been implemented at many hundreds of petrochemical and refinery units worldwide for the last 30 years. This technology is very beneficial and makes an immediate and noticeable difference in the way a major process unit works. It sits on top of the regulatory controls and generally writes all the regulatory controller set-points every minute. Therefore, that lower control layer must function flawlessly to maintain MPC benefits. An ethylene plant typically requires 15-25 such MPC applications, depending on the number of cracking furnaces. A typical refinery unit would require anywhere between 5 to 10 MPCs, sometimes more.
(3) Add rigorous, closed-loop, real-time optimization (CLRTO) as the next layer above MPC. This requires plant-wide chemical engineering models to run under a non-linear optimizer. Examples of such products are RT-OPT and NOVA. Examples would be one CLRTO for a complete ethylene plant, or a cat cracker in a refinery. CLRTO uses "open equation" solution methods and has been used for over 15 years in ethylene plants. This technology is better suited for petrochemical processes than refinery units, in my opinion, mainly because of uncertainties in feedstock composition in most refinery units. At present, again in my opinion, the traditional steady-state simulators are incapable of doing serious optimization work at a plant-wide level.
There are many hundreds of published papers describing the MPC and CLRTO technologies, some going back over 30 years. I would recommend you study a few of them carefully before calling the vendor companies. There is no panacea for maximizing profit, and it takes many years of extremely diligent work by a very large team of specialists in each plant before all this works properly.
Of course, the cost runs into the millions of dollars and you need some high caliber staff to maintain this stuff, but the paybacks are generally within two years for each major piece, many times a lot sooner. A huge bonus is much better staff training, in many areas, and an inexorable focus on control systems performance and maintenance.
My opinion on calling in the vendors is to try to do it yourself first or with a consultant that is on by the hour, not performance based (a piece of the action or iimplimenting his wizbang system). After everyone in your organization is completely out of ideas, then go for the wizbang stuff.
I've seen it so many times where all the gizmos were brought in and all they did was tune up instruments, take advantage of measurement errors that should poor performance that wasn't there to begin with, and then walk out with a pocket full of dollars leaving you with a bunch of stuff that doesn't do anything more.
Remember, a management consultant is someone that steals your watch and then charges you to tell you what time it is.
I've seen it so many times where all the gizmos were brought in and all they did was tune up instruments, take advantage of measurement errors that should poor performance that wasn't there to begin with, and then walk out with a pocket full of dollars leaving you with a bunch of stuff that doesn't do anything more.
Remember, a management consultant is someone that steals your watch and then charges you to tell you what time it is.
- Status
Similar threads
- Locked
- Question
- Locked
- Question
