Experts, take a look at why this proii simulation doesn’t converge??

Using the various parameters taken from the actual tower, I intended to run the simulation to see the results, but it stopped immediately upon execution; it seems that convergence did not occur. Could some experts please explain the reason for this?

SPEC ID=COL1SPEC2, STREAM=S2,FRACTION, COMP=1,WET, VALUE=0.995 cannot be achieved in the simulation! It is achievable in actual production for the following reasons: I eliminated the requirement regarding the purity of the product at the top of the tower; by adjusting the reflux ratio, it is possible to achieve a certain level of purity, but no matter how high the reflux ratio is, the purity of the product at the top of the tower remains at most 98.32% (mol), with the remainder being essentially water! The problem lies here: when using SRK for calculations, the aqueous phase at the top of the tower cannot be completely removed from stream S4; as a result, there is still some water present in the product stream S2 at the top of the tower. Therefore, no matter how these parameters are adjusted (recovery ratio, number of theoretical plates, feed position…), it doesn’t help. In actual factory production, the water can be made very pure, and this is why convergence does not occur once the purity at the top of the tower is specified.

I have also been studying the issue of *water precipitation recently, and I have realized the problem mentioned by the teacher above. In situations like this, how should we handle the simulation results? Should we just ignore them?

I checked it for you and adjusted some parameters. I didn’t pay attention to the practical accuracy of your example; I only focused on ensuring convergence, for reference only! The main problems are as follows: 1. Material imbalance! ! In the example given, the net content of TOLUENE in the feed is 12.506 kg-mol (1.3%), whereas in your example, the amount of S2 obtained (based on TOLUENE as a reference) is 14.5 kg-mol, which is incorrect. A mistake that many people often make easily. Many people in this section have similar problems with their examples. 2. In actual production operations, this tower (as well as many similar distillation towers), due to the properties of the materials involved, results in one material being contained within the other. In S2 of the example, it is impossible to have no H2O present, and it is also not possible to meet the distillation standards you have set; the highest achievable purity is 98.3%. Therefore, in simulation calculations, it is necessary to keep in mind the realities of the actual situation at all times. However, the methods of dehydration in reality cannot be simulated in this mode of the software. Additionally, besides this thermodynamic method, there are some other methods that can also be used. $ Generated by PRO/II Keyword Generation System $ Generated on: Fri May 23 16:11:11 2008 TITLE DIMENSION METRIC, PRES=MPAG, ENERGY=KJ, STDTEMP=0, STDPRES=0 SEQUENCE SIMSCI CALCULATION RVPBASIS=APIN, TVP=37.778 COMPONENT DATA LIBID 1,TOLUENE/2,PXYLENE/3,MXYLENE/4,OXYLENE/5,14EZ/6,H2O, & BANK=PROCESS,SIMSCI ASSAY CONVERSION=API87, TBPEP=99, CURVEFIT=IMPROVED, & KVRECONCILE=TAILS THERMODYNAMIC DATA METHOD SYSTEM=SRK, DIFFUSIVITY=PRED, SET=SRK01 METHOD SYSTEM=BWRS, SET=BWRS01, DEFAULT STREAM DATA PROPERTY STREAM=S1, TEMPERATURE=144, PRESSURE=0.06, PHASE=M, & RATE(M)=961.999, COMPOSITION(WT)=1,1.3/2,98.6/3,0.04/4,0.02/ & 5,0.01/6,0.03 UNIT OPERATIONS COLUMN UID=T1 PARAMETER TRAY=51,IO=30 FEED S1,18 PRODUCT OVHD(M)=S2,12.5, WATER(M)=S4,1, BTMS(M)=S3, & SUPERSEDE=ON CONDENSER TYPE=BUBB DUTY 1,1,,CONDENSER DUTY 2,51,,REBOILER PSPEC PTOP=0.01, DPCOLUMN=0.008 PRINT PROPTABLE=PART ESTIMATE MODEL=CONVENTIONAL, RRATIO=5 SPEC ID=COL1SPEC2, STREAM=S2, RATE(KGM/H), COMP=1,WET, DIVIDE, & STREAM=S2, RATE(KGM/H),TOTAL,WET, VALUE=0.983 SPEC ID=COL1SPEC1, STREAM=S3, RATE(KGM/H), COMP=2,WET, DIVIDE, & STREAM=S3, RATE(KGM/H),TOTAL,WET, VALUE=0.999 VARY DNAME=CONDENSER,REBOILER REBOILER TYPE=KETTLE END Last edited by weiqj on 2008-5-23 16:49 ]

The concept of material imbalance does not exist! 14.5 kg-mol is just an initial value; there is no problem as long as it is not set according to design specifications. Of course, the closer the initial value is to the correct solution, the better for convergence – that’s all! “\"Many people often make this mistake\" ——:L The second point is correct; the calculations do indeed work that way, but only the person performing the operations knows the actual situation. If, as the original poster suggests, the actual value can reach 0.995, then it means that the equilibrium coefficient used for simulating oil-water separation differs from the actual value, and the LLV data needs to be improved.

You’re right after all; I’ll give you extra points. But it still needs to be pointed out that. Even if the feed plate value is changed from 18 to 28, convergence is still possible; 48 works as well too. So by handing over many values that should ideally be accurate to a computing software, it feels like playing a CS game – you can’t see the enemies, but as long as you shoot through doors or walls, they will still get eliminated, haha........ It can be imagined that if a designer is checking a problematic calculation result, he need not doubt these correct values, allowing him to find the issue more easily. That was my intention.

The answers provided by the two people upstairs are both excellent; this simulation example can be solved using them. But I would like to ask more in depth: what should we do when dealing with systems that still contain water? Do we have to rely on such vague methods to ignore the presence of water? Are there any other ways to effectively address this issue?

More accurate basic data is obtained through experiments to refine the model.

Did the PRO_II software make a calculation error? No. Do we need to correct the database data in it? It’s hard to believe that the tests we conduct will be more accurate than those provided by PRO_II. Of course, we can extract the relevant data from actual petrochemical plants and make modifications, but this is nothing more than adding a bypass to a control loop. I can tell the original poster that I made no modifications to the PRO_II database, and I achieved the 99.5% metric for the tower top S2 that the poster was hoping for. What I still question is how we connect with reality.