Ceng 412 Spring, 1998

Testing Thermo Packages

Table 1 Errors Reported for Different Thermodynamic Packages

Package Number ofParametersVaried in Fit Max Error in Bubble Point RMS Error in Bubble Point Max Error in Vapor mol frac Acetone RMS Error in Vapor mol frac Acetone
Peng-Rob.
6
 0.5241 0.1510 0.00733 0.00295
PR-BM
6
 0.5247 0.1493 0.00713 0.00290
RK-SOAVE
6
 0.4404 0.1274 0.00667 0.00272
RK-Aspen
6
 0.5182 0.1443 0.00858 0.00314
RK-BM
6
 0.4409 0.1256 0.00647 0.00267
BWR-LS
6
 2.3552 0.6486 0.01384 0.00566
NRTL
6
 0.1313 0.0522 0.00839 0.00247
ENRTL-HF
9
 0.0838 0.0361 0.00934 0.00279
NRTL-HOC
6
 0.1288 0.0544 0.0128 0.00395
LK-PLOCK
3
 0.3484 0.1402 0.01384 0.00566
UNIQUACK
6
 0.1193 0.0549 0.00483 0.00277
VAN LAAR
6
 0.5927 0.2648 0.0161 0.00638
WILSON
6
 0.1234 0.0529 0.00824 0.00155
Margules from Matlab
3
 0.4010 0.1355 0.00965 0.00359
IDEAL from Matlab
0
 4.8360 2.4093 0.04580 0.02630

Table 2 Parameters set in the Packages

Package Type of Parameters Name of Parm. Set No. of Iterations SSQ at end
Peng-Rob. unary & binary/symmetric OMGPRS/1 PRKIJ/1 5 614.9
PR-BM unary & binary/symmetric OMGPRS/1 PRKIJ/1 5 590.7
RK-SOAVE unary & binary/symmetric OMRKSS/1 RKSKIJ/1 5 554.1
RK-Aspen unary & binary/symmetric RKAPOL RKAKAO 5 599.0
RK-BM unary & binary/symmetric OMGRKS/1 RKSKIJ 5 530.1
BWR-LS binary/symmetric & binary/symmetric BWRKT/1 BWRKV/1 8 8074.0
NRTL binary NRTL/2 8 270.7
ENRTL-HF Unary & binary RKTZRA/1 & NRTL/2 296.0
NRTL-HOC binary NRTL/2 8 453.9
LK-PLOCK binary/symmetric LKPKIJ-1 5 1231.6
UNIQUACK binary UNIQ-1 10 371.0
VAN LAAR binary VANL/2 7 2172.4
WILSON binary WILSO/2 7 280.8
Margules from Matlab binary/symmetric 277
IDEAL from Matlab None 0

 

A matrix of errors from each package divided by those from the ideal simulation in Matlab was created and is show in Table 3.

Table 3 Errors Relative to those from Ideal simulation
          Max T    RMS T   Max yAce   RMS yAce
PENG-ROB 0.1084   0.0627   0.1600     0.1122
PR-BM    0.1085   0.0620   0.1557     0.1103
RK-SOAVE 0.0911   0.0529   0.1456     0.1034
RK-ASPEN 0.1072   0.0599   0.1873     0.1194
RK-BM    0.0912   0.0521   0.1413     0.1015
BWR-LS   0.4870   0.2692   0.3022     0.2152
NRTL     0.0272   0.0217   0.1832     0.0939
ENRTL-HF 0.0173   0.0150   0.2039     0.1061
NRTL-HOC 0.0266   0.0226   0.2795     0.1502
LK-PLOCK 0.0720   0.0582   0.3022     0.2152
UNIQUAC  0.0247   0.0228   0.1055     0.1053
VAN LAAR 0.1226   0.1099   0.3515     0.2426
WILSON   0.0255   0.0220   0.1799     0.0589
Margules 0.0829   0.0562   0.2107     0.1365

Here are the temperature errors sorted by the RMS errors.

Figure 1: 1 Gives the Max Abs Errors & 2 Gives RMS Errors in T

From this figure (and the table of errors), we can put all the packages into four categories:

Category
Range of RMS Errors: % of Ideal
Packages
Range of Max Abs Errors: % of Ideal

Excellent

Between 1.5 and 2.5%

ENRTL-HF, NRTL, WILSON, NRTL-HOC, UNIQUAC

Between 2.2 and 2.7%

Good

Between 5 and 6.5%

RK-BM, RK-SOAVE, Margules, LK-PLOCK, RK-ASPEN, PR-BM, PENG-ROB

Between 7 and 11%

Fair

About 11%

Van Laar

About 12%

Poor

Nearly 27%

BWR-LS

About 49%

Figure 2: 1 Gives the Max Abs Errors & 2 Gives RMS Errors in y for Acetone

Are errors in this figure are relative to the ideal simulation errors.

The reductions in the errors in the vapor composition of Acetone were less dramatic than those achieved by the packages for temperatures. We might put the packages into three categories this time:

Category
Range of RMS Errors % of Ideal
Packages
Range of Max Abs Errors % of Ideal
Good between 5.5 and 12% WILSON, NRTL, RK-BM, RK-SOAVE, UNIQUAC, ENRTL-HF, PR-BM, PENG-ROB, RK-ASPEN between 10 and 20%

Fair

 between 13.5 and 15% Margules, NRTL-HOC between 21 and 28%
Poor between 21 and 24.5% BWR-LS, LK-PLOCK, VAN Laar between 30 and 35%