8 replies to this topic

[smg@2021]

Hello!!

 

I am planning to build a simulation model for Crude Unit exchanger train in PRO/II. The intent of the model is to optimise the existing preheat temperature and reduce the load on Crude Unit Heater.

 

I don’t have all the exchangers process datasheets but have all exchangers operating inlet and outlet temperatures and flowrates across them.

 

The question is how to initiate the simulation model building? From where do I start?

For example, should I first make a PFD in PRO/II and enter each exchangers outlet temperatures to find out the heat duty and then enter heat duty in each exchanger to find out outlet temperatures with changed operating conditions?? Or should I use UA instead of heat duty?

 

Any help would be much appreciated! 

[Pilesar]

From what you describe, using UA sounds like the best start. Ideally, you will want to eventually model the exchangers rigorously. Crude preheat trains are common simulations. Have you tried contacting the software support group to see if they have a Go-By to send you? It may be in one of their ready-made training modules.

[SilverShaded]

You should use UA for the majority of exchangers.  The problem you will find is that the data (assuming operating data) will be inconsistent with the measured flowarates and very probably some temperature measurements will be wrong.  You should first reconcile the data to try and average out the discrepencies and identify any faulty temperature readings, otherwise you will likely find that rundown temperatures are not close to reality or the model is unrealistic in some other way.

Reconcile the data before you get anywhere near a simulation, do it in excel.

For building the model you will need to integrate the pumparound duties from the column into the prehehat train and the return temperature to the column of the pumparounds should be estimated from the prehehat train (unless there is temperature control on the pumparound return).  Otherwise the model will not be heat balanced and useless for investigating energy recovery.

If any exchangers are two phase (e.g. Column oveheads intregrated with the prehehat train) this should be a rigorous exchanger model as UA is not adequate.  Sometimes exchangers before preflash drums, or before the furnace, can also have two phase flow on the crude side.

This is not a trivial excercise to get right, model build should be around 3 months to get an adequately tuned and reconciled model, if not, i wouldn't trust the results.

 

Also if its operating data your using and your creating the crude by back blending the products (which you should be), light naphtha stream D86's can't be believed.  If you have to also match a a debutaniser which may be heat intregated with the CDU then you may need to adjust the Light naphtha composition to get a good match on temperatures.

 

Edited by SilverShaded, 24 September 2021 - 03:23 AM.

[smg@2021]

You should use UA for the majority of exchangers.  The problem you will find is that the data (assuming operating data) will be inconsistent with the measured flowarates and very probably some temperature measurements will be wrong.  You should first reconcile the data to try and average out the discrepencies and identify any faulty temperature readings, otherwise you will likely find that rundown temperatures are not close to reality or the model is unrealistic in some other way.

Reconcile the data before you get anywhere near a simulation, do it in excel.

For building the model you will need to integrate the pumparound duties from the column into the prehehat train and the return temperature to the column of the pumparounds should be estimated from the prehehat train (unless there is temperature control on the pumparound return).  Otherwise the model will not be heat balanced and useless for investigating energy recovery.

If any exchangers are two phase (e.g. Column oveheads intregrated with the prehehat train) this should be a rigorous exchanger model as UA is not adequate.  Sometimes exchangers before preflash drums, or before the furnace, can also have two phase flow on the crude side.

This is not a trivial excercise to get right, model build should be around 3 months to get an adequately tuned and reconciled model, if not, i wouldn't trust the results.

 

Also if its operating data your using and your creating the crude by back blending the products (which you should be), light naphtha stream D86's can't be believed.  If you have to also match a a debutaniser which may be heat intregated with the CDU then you may need to adjust the Light naphtha composition to get a good match on temperatures.

 

 

Thank a lot for replying!!

 

Firstly, I doubt that the assumption that UA will remain same of each exchangers is correct since UA will change with time due to fouling of exchangers. Isn’t it?

Secondly, from what you are suggesting for integration of pump around duties, do I need to simulate Crude column also to achieve desired heat balance?

Lastly, our plant operates on frequent changeover from HS to LS crude within 3 days the crude change over takes place from HS crude to LS crude, it would be difficult task to incorporate the same in the model. Would you mind suggesting something?

 

And yes we have the naphtha stabilizer column with its reboiling integrated with the crude column pumparounds so I guess I would be required to adjust Light Naphtha D86 composition.

 

Really appreciate the help!!!

[smg@2021]

From what you describe, using UA sounds like the best start. Ideally, you will want to eventually model the exchangers rigorously. Crude preheat trains are common simulations. Have you tried contacting the software support group to see if they have a Go-By to send you? It may be in one of their ready-made training modules.

 

No, I didn’t know that software support group could provide a ready-made model. But it would be challenging and exciting to simulate it all by myself.

Thanks for the help anyways!

[SilverShaded]

You should use UA for the majority of exchangers. The problem you will find is that the data (assuming operating data) will be inconsistent with the measured flowarates and very probably some temperature measurements will be wrong. You should first reconcile the data to try and average out the discrepencies and identify any faulty temperature readings, otherwise you will likely find that rundown temperatures are not close to reality or the model is unrealistic in some other way.

Reconcile the data before you get anywhere near a simulation, do it in excel.

For building the model you will need to integrate the pumparound duties from the column into the prehehat train and the return temperature to the column of the pumparounds should be estimated from the prehehat train (unless there is temperature control on the pumparound return). Otherwise the model will not be heat balanced and useless for investigating energy recovery.

If any exchangers are two phase (e.g. Column oveheads intregrated with the prehehat train) this should be a rigorous exchanger model as UA is not adequate. Sometimes exchangers before preflash drums, or before the furnace, can also have two phase flow on the crude side.

This is not a trivial excercise to get right, model build should be around 3 months to get an adequately tuned and reconciled model, if not, i wouldn't trust the results.

Also if its operating data your using and your creating the crude by back blending the products (which you should be), light naphtha stream D86's can't be believed. If you have to also match a a debutaniser which may be heat intregated with the CDU then you may need to adjust the Light naphtha composition to get a good match on temperatures.

Thank a lot for replying!!

Firstly, I doubt that the assumption that UA will remain same of each exchangers is correct since UA will change with time due to fouling of exchangers. Isn’t it?
Secondly, from what you are suggesting for integration of pump around duties, do I need to simulate Crude column also to achieve desired heat balance?
Lastly, our plant operates on frequent changeover from HS to LS crude within 3 days the crude change over takes place from HS crude to LS crude, it would be difficult task to incorporate the same in the model. Would you mind suggesting something?

And yes we have the naphtha stabilizer column with its reboiling integrated with the crude column pumparounds so I guess I would be required to adjust Light Naphtha D86 composition.

Really appreciate the help!!!

Yes UA will change over time as the exchanger fouls. So will fouling factor so even a rigorous exchanger model would need to be re-tuned.

Fixing the temperature or duty is the worst of all worlds and would make an absolutely uselss model.

Yes you do need to model the columns and heat integrate them in the model with the preheat train. Pumparound duty will change as you modify the preheat train, which impact yields.

If all your doing is trying work out which exchangers should be cleaned (ie its NOT actually an energy study) then you can ‘get away’ with just the preheat train model.

It would help of you said what the actual purpose of the model is, e.g. just play around with the existing exchangers or try to revamp the preheat train in some way.

Generally for CDUs you can not disentangle the yield/energy interdependance without a full model. Yields are often worth an order of magnitude more than any energy effect, the exception being if you can increase capacity by modifying preheat train operation, even then it depends how you increase capacity, which depends on whether any other constraints exist in the system.

Edited by SilverShaded, 25 September 2021 - 12:54 PM.

[Pilesar]

If all you are doing is a cleaning case study to optimize when to clean each preheat exchanger, then consider using Hextran if it is available to you.

[smg@2021]

 

 

You should use UA for the majority of exchangers. The problem you will find is that the data (assuming operating data) will be inconsistent with the measured flowarates and very probably some temperature measurements will be wrong. You should first reconcile the data to try and average out the discrepencies and identify any faulty temperature readings, otherwise you will likely find that rundown temperatures are not close to reality or the model is unrealistic in some other way.

Reconcile the data before you get anywhere near a simulation, do it in excel.

For building the model you will need to integrate the pumparound duties from the column into the prehehat train and the return temperature to the column of the pumparounds should be estimated from the prehehat train (unless there is temperature control on the pumparound return). Otherwise the model will not be heat balanced and useless for investigating energy recovery.

If any exchangers are two phase (e.g. Column oveheads intregrated with the prehehat train) this should be a rigorous exchanger model as UA is not adequate. Sometimes exchangers before preflash drums, or before the furnace, can also have two phase flow on the crude side.

This is not a trivial excercise to get right, model build should be around 3 months to get an adequately tuned and reconciled model, if not, i wouldn't trust the results.

Also if its operating data your using and your creating the crude by back blending the products (which you should be), light naphtha stream D86's can't be believed. If you have to also match a a debutaniser which may be heat intregated with the CDU then you may need to adjust the Light naphtha composition to get a good match on temperatures.
 

Thank a lot for replying!!

Firstly, I doubt that the assumption that UA will remain same of each exchangers is correct since UA will change with time due to fouling of exchangers. Isn’t it?
Secondly, from what you are suggesting for integration of pump around duties, do I need to simulate Crude column also to achieve desired heat balance?
Lastly, our plant operates on frequent changeover from HS to LS crude within 3 days the crude change over takes place from HS crude to LS crude, it would be difficult task to incorporate the same in the model. Would you mind suggesting something?

And yes we have the naphtha stabilizer column with its reboiling integrated with the crude column pumparounds so I guess I would be required to adjust Light Naphtha D86 composition.

Really appreciate the help!!!

Yes UA will change over time as the exchanger fouls. So will fouling factor so even a rigorous exchanger model would need to be re-tuned.

Fixing the temperature or duty is the worst of all worlds and would make an absolutely uselss model.

Yes you do need to model the columns and heat integrate them in the model with the preheat train. Pumparound duty will change as you modify the preheat train, which impact yields.

If all your doing is trying work out which exchangers should be cleaned (ie its NOT actually an energy study) then you can ‘get away’ with just the preheat train model.

It would help of you said what the actual purpose of the model is, e.g. just play around with the existing exchangers or try to revamp the preheat train in some way.

Generally for CDUs you can not disentangle the yield/energy interdependance without a full model. Yields are often worth an order of magnitude more than any energy effect, the exception being if you can increase capacity by modifying preheat train operation, even then it depends how you increase capacity, which depends on whether any other constraints exist in the system.

 

 

The purpose of the model is

1.  Finding out which exchanger should be cleaned.

2.  Find out appropriate split ratios (we have 3 passes in the each train) in the train to get the maximum achievable preheat at given point of time.

 

That is the only purpose of the model.

[SilverShaded]

 

 

 

You should use UA for the majority of exchangers. The problem you will find is that the data (assuming operating data) will be inconsistent with the measured flowarates and very probably some temperature measurements will be wrong. You should first reconcile the data to try and average out the discrepencies and identify any faulty temperature readings, otherwise you will likely find that rundown temperatures are not close to reality or the model is unrealistic in some other way.

Reconcile the data before you get anywhere near a simulation, do it in excel.

For building the model you will need to integrate the pumparound duties from the column into the prehehat train and the return temperature to the column of the pumparounds should be estimated from the prehehat train (unless there is temperature control on the pumparound return). Otherwise the model will not be heat balanced and useless for investigating energy recovery.

If any exchangers are two phase (e.g. Column oveheads intregrated with the prehehat train) this should be a rigorous exchanger model as UA is not adequate. Sometimes exchangers before preflash drums, or before the furnace, can also have two phase flow on the crude side.

This is not a trivial excercise to get right, model build should be around 3 months to get an adequately tuned and reconciled model, if not, i wouldn't trust the results.

Also if its operating data your using and your creating the crude by back blending the products (which you should be), light naphtha stream D86's can't be believed. If you have to also match a a debutaniser which may be heat intregated with the CDU then you may need to adjust the Light naphtha composition to get a good match on temperatures.
 

Thank a lot for replying!!

Firstly, I doubt that the assumption that UA will remain same of each exchangers is correct since UA will change with time due to fouling of exchangers. Isn’t it?
Secondly, from what you are suggesting for integration of pump around duties, do I need to simulate Crude column also to achieve desired heat balance?
Lastly, our plant operates on frequent changeover from HS to LS crude within 3 days the crude change over takes place from HS crude to LS crude, it would be difficult task to incorporate the same in the model. Would you mind suggesting something?

And yes we have the naphtha stabilizer column with its reboiling integrated with the crude column pumparounds so I guess I would be required to adjust Light Naphtha D86 composition.

Really appreciate the help!!!

Yes UA will change over time as the exchanger fouls. So will fouling factor so even a rigorous exchanger model would need to be re-tuned.

Fixing the temperature or duty is the worst of all worlds and would make an absolutely uselss model.

Yes you do need to model the columns and heat integrate them in the model with the preheat train. Pumparound duty will change as you modify the preheat train, which impact yields.

If all your doing is trying work out which exchangers should be cleaned (ie its NOT actually an energy study) then you can ‘get away’ with just the preheat train model.

It would help of you said what the actual purpose of the model is, e.g. just play around with the existing exchangers or try to revamp the preheat train in some way.

Generally for CDUs you can not disentangle the yield/energy interdependance without a full model. Yields are often worth an order of magnitude more than any energy effect, the exception being if you can increase capacity by modifying preheat train operation, even then it depends how you increase capacity, which depends on whether any other constraints exist in the system.

 

 

The purpose of the model is

1.  Finding out which exchanger should be cleaned.

2.  Find out appropriate split ratios (we have 3 passes in the each train) in the train to get the maximum achievable preheat at given point of time.

 

That is the only purpose of the model.

 

Then what you need is something that automatically reconciles the data and re-calculates the UA or fouling factor.  Which can then also recommend which exchanger to clean based on benefit vs cost of cleaning and can also optimise the splits.  Really an online system. 

You could actually do this in excel, download data, reconcile the data, calculate cP's using Lee Kesler liquid enthalpy correlation, calculate the UA and use excel solver to optimise the splits.  No need for a simulator, unless you want rigorous exchangers to calculate fouling factor.

Edited by SilverShaded, 27 September 2021 - 06:20 AM.