14 replies to this topic

[processengineer1989]

Hello everybody.

 

I'm a fresh process engineer, and I'm trying to optimize a gas treating plant and find new parameters for a better condensate and gas recovery.

 

My question is how should I proceed?  I'm willing to use HySys and Excel as tools but I definitely have the intention to change the parameters randomly until I have a good result.  The problem is the plant is so complicated and I assume there's a better method to proceed.

 

Please help!

 

Thank you so much.

 

S.B

Attached Files

•  Phase A.PNG   235.46KB   6 downloads

Edited by processengineer1989, 25 December 2017 - 09:13 AM.

[serra]

perhaps the minimizers available in Excel are not the best tools for that application,
working with Excel I prefer a library (Prode Properties) as discussed in many threads here,
anyway, you should define your goal (for example minimum costs, maximum product etc.)
some parameters to vary and run a minimizer (or the inverse which is the same) to identify the best solution.
With prode I have used both Mathcad and Matlab which include more powerful tools than Excel,
also consider the tools already available in your simulator.

[processengineer1989]

Goal: Maximise condensate recovery, that's the main one!!

 

after that comes: minimise fuel gas comsumption!

[serra]

running an external minimizer including, as objective function, a whole plant (with columns etc.) can result very time expensive,
even the so called derivative free minimizers can call hundreds or thousands times the objective function,
which can result impractical for a complex plant (you can set bounds on each parameter and reduce ranges but still the task is very complex),
simulators include tools for sensibility analysis on different parameters, you may wish to use these tools (which call directly the simulator code) instead of external tools.

As said, to work with tools such as Excel, Mathcad or Matlab I prefer a library which is much more efficient (see similar threads for detailed discussions),
the limit of this approach is that optimization is limited to subsets of complex plants,
for example a column plus a very limited number of equips, also without recycles the task is hard,
to give you an example, in many cases solving a single column with Prode takes a second or less, still the optimization can require 30 minutes or more

for complex plants, probably the tools (sensivity analysis etc.) already included in your simulator are the best solution.

[Bobby Strain]

You should take the tasks one at a time. First, build your model in HYSYS. Use operating data to verify the model. Then you can begin to optimize.

 

Bobby

[Saml]

If you have HySys you can build your model in sequential modular mode. Have a base case, and then switch to an equation oriented (EO) simulation.

 

This is not an easy step, so I would suggest you start small, like a single piece of equipment with a recycle, where EO would be an overkill, to understande what you are doing.

 

Also, don't try to bite too big, it takes months to build such a model.

[processengineer1989]

Thank you so much Bobby. That's what I am intending to do.

 

But After that, should I use Hysys Optimizer?  or is there another way to proceed?

 

Thanks again

 

Sarah

[processengineer1989]

Saml:

 

Well, No i'm not willing to bite tot big. I am working on finding a way to proceed.

 

Is Hysys Optimizer good?  And should I proceed equipment by equipment?

 

thank you

 

Sarah

[Bobby Strain]

Sarah,

      Build the HYSYS model with the main equipment and post a screen shot. It's difficult to tell what your process is since there are so many configurations. Then we can give you some guidance.

 

Bobby

[processengineer1989]

Right away M. Bobby.

 

I'll try to finish it as soon as possible and I'll post it in here.

 

Have a nice day

 

Sarah

[processengineer1989]

I have uploaded a screenshot of the process simulation from Hysys.

 

Sorry for the delay.

 

S. B

[Bobby Strain]

Sarah,

     The upload didn't get posted. Try again, carefully following the upload instructions.

 

Bobby

[processengineer1989]

Sarah,

     The upload didn't get posted. Try again, carefully following the upload instructions.

 

Bobby

Uploading is successfully done.

[Bobby Strain]

Sarah,

      It's a bit easier with the screen shot. First you will have to set limiting conditions on gas and recovered liquids streams. If you can vary the expander discharge pressure, then this is one optimizing parameter. The only other variable is the flow split of the feed gas to the two exchangers. You can vary the split until you get the maximum liquid from the first separator.

     To get meaningful results you need plant operating data to determine UA for each of the two exchangers. And you need to determine the expander and compressor efficiencies. So, this is not really an application for sophisticated optimization schemes. Just parametric studie. If you can use Excel VBA, you can automate the exercise. And if you don't know how to use VBA or VB.NET, you should learn.

 

Good luck.

Bobby

[processengineer1989]

Sarah,

      It's a bit easier with the screen shot. First you will have to set limiting conditions on gas and recovered liquids streams. If you can vary the expander discharge pressure, then this is one optimizing parameter. The only other variable is the flow split of the feed gas to the two exchangers. You can vary the split until you get the maximum liquid from the first separator.

     To get meaningful results you need plant operating data to determine UA for each of the two exchangers. And you need to determine the expander and compressor efficiencies. So, this is not really an application for sophisticated optimization schemes. Just parametric studie. If you can use Excel VBA, you can automate the exercise. And if you don't know how to use VBA or VB.NET, you should learn.

 

Good luck.

Bobby

Thanks alot.

I'll try to do that and I hope I won't find difficulties learning VBA. If not, I'll be back and ask questions.

 

Thank you so much again.

 

Sarah