Tabular calcs with mixtures often return Dew T< Bubble T using PQ input pair

[mapipolo]

I've traced a number of issues back to this behavior of the tabular backend and have collected a bunch of cases to demonstrate it. The code is somewhat lengthy, so I will provide just an example and results.

I initialize arrays refrig and tSatVap with the first two columns below:

refrig	tSatVap	tSatLiqRefprop	tSatLiqBicubic	dT
R440A.mix	233.150017	231.6475746	246.3096705	14.6620959
R407D.mix	233.1500186	226.3933816	233.2254995	6.832117836
R449A.mix	233.1500201	227.4313526	233.4113896	5.980036991
R401A.mix	245.3722394	239.6272034	245.3913809	5.764177523
R404A.mix	226.1500205	225.3889674	229.6831741	4.29420675
R404A.mix	226.1500205	225.3889674	229.6831741	4.29420675
R404A.mix	231.4500081	230.7227298	234.3192037	3.596473883
R404A.mix	231.4833394	230.7562719	234.3405164	3.584244464
R404A.mix	232.155001	231.4320517	234.7466609	3.314609206
R404A.mix	233.1500159	232.4331244	235.2648581	2.831733721
R440A.mix	263.1500251	262.1155579	264.8568705	2.741312685
R507A.mix	233.1500237	233.1479196	235.6368096	2.488889966
R404A.mix	247.1500229	246.5116515	247.7053921	1.193740596
R404A.mix	247.038908	246.3999553	247.5892415	1.189286169
R404A.mix	246.1500249	245.5064014	246.6500922	1.143690836
R507A.mix	250.9341316	250.9245656	251.9898921	1.065326512
R404A.mix	243.1500264	242.4903483	243.366103	0.875754681
R507A.mix	235.6367987	235.6339113	236.4775308	0.843619451
R507A.mix	254.2611231	254.2498315	254.9242524	0.674420982
R507A.mix	240.9012428	240.8963585	241.5002823	0.603923744
R404A.mix	255.3500214	254.7530596	255.351379	0.598319337
R410A.mix	260.9278003	260.8323188	261.4154559	0.58313711
R507A.mix	254.9847256	254.9730511	255.5207537	0.547702564
R507A.mix	240.3722455	240.3675786	240.9012692	0.533690679
R507A.mix	240.2696508	240.2650255	240.7848147	0.519789221

Then we step through, using tSatVap and refrig as inputs, changing the "BICUBIC&REFPROP" backend string to "REFPROP" for the nontabular calculation:

for (int i = 0; i < tSatVap.Length; i++)
{
    AbstractState s = AbstractState.factory("BICUBIC&REFPROP", refrig[i]);
    s.update(input_pairs.QT_INPUTS, 1, tSatVap[i]);
    pSatLiq[i] = s.p();
    s.update(input_pairs.PQ_INPUTS, pSatLiq[i], 0);
    tSatLiqBicubic[i] = s.T();
}

Once this loop is complete, we have the tSatLiq***** outputs in the third and fourth columns above. As we see, the bubble point temperatures are lower than the dew point temperatures as calculated by Refprop directly (as we would expect), but higher than the dew point temperatures as calculated by the tabular backend (I've sorted the results by decreasing dT, the disparity between the Refprop and tabular result). I calculated with the TTSE backend as well, but the results were identical to those returned using the bicubic backend.

Using: January 11th nightly build of 5.1.2, C#.

[mapipolo]

NOTE: Until a few minutes ago, the data in the above table was incorrectly entered. It has been edited to correct the errors.

[ibell]

Could this just be a result of the interaction parameters? Do CoolProp & REFPROP agree when you don't use tabular backends?

What version of REFPROP do you use? Do you use a custom HMX.BNC file?

[ibell]

Sorry forget my last comment - must be something with the interpolation, not the interaction parameters. Would be useful to just look at error in prediction of p for given T for these given fluids. So we can see where the problems start. Could do you that? Calculate dewpoint pressure for given dewpoint temperatures for these fluids?

[mapipolo]

No problem. For bicubic, I ran this in a loop:

AbstractState s = AbstractState.factory("BICUBIC&REFPROP", refrigs[i]);
s.update(input_pairs.QT_INPUTS, 1, tSatVap[i]);
pSatVapBicubic[i] = s.p();
s.update(input_pairs.PQ_INPUTS, pSatVapBicubic[i], 0);
tSatLiqBicubic[i] = s.T();

And similarly for Refprop:

AbstractState s = AbstractState.factory("REFPROP", refrigs[i]);
s.update(input_pairs.QT_INPUTS, 1, tSatVap[i]);
pSatVapRefprop[i] = s.p();
s.update(input_pairs.PQ_INPUTS, pSatVapRefprop[i], 0);
tSatLiqRefprop[i] = s.T();

The results were as below (again, I've ignored TTSE because results were identical to bicubic):

refrig	tSatVap	pSatVapBicubic	pSatVapRefprop	tSatLiqBicubic	tSatLiqRefprop	dP	dT
R440A.mix	233.150017	47550.00382	47541.78582	246.3096705	231.6475746	8.217998456	14.6620959
R407D.mix	233.1500186	70198.33488	70232.44298	233.2254995	226.3933816	34.10809874	6.832117836
R449A.mix	233.1500201	101305.9639	101325.111	233.4113896	227.4313526	19.14702065	5.980036991
R401A.mix	245.3722394	98637.09132	98640.82158	245.3913809	239.6272034	3.730263846	5.764177523
R404A.mix	226.1500205	94181.29423	94061.93947	229.6831741	225.3889674	119.3547581	4.29420675
R404A.mix	226.1500205	94181.29423	94061.93947	229.6831741	225.3889674	119.3547581	4.29420675
R404A.mix	231.4500081	121154.2973	121119.5991	234.3192037	230.7227298	34.69819676	3.596473883
R404A.mix	231.4833394	121341.2697	121307.1227	234.3405164	230.7562719	34.14703283	3.584244464
R404A.mix	232.155001	125157.622	125133.8304	234.7466609	231.4320517	23.79163436	3.314609206
R404A.mix	233.1500159	130984.2256	130975.2615	235.2648581	232.4331244	8.964092735	2.831733721
R440A.mix	263.1500251	182936.9578	182939.2217	264.8568705	262.1155579	2.263882957	2.741312685
R507A.mix	233.1500237	138654.8973	138661.6102	235.6368096	233.1479196	6.712901083	2.488889966
R404A.mix	247.1500229	237832.0631	237891.8519	247.7053921	246.5116515	59.78878533	1.193740596
R404A.mix	247.038908	236780.0902	236840.4144	247.5892415	246.3999553	60.32418061	1.189286169
R404A.mix	246.1500249	228492.9593	228557.0512	246.6500922	245.5064014	64.09188081	1.143690836
R507A.mix	249.8166787	277111.1911	277143.8853	250.9341432	249.8076771	32.69418687	1.126466057
R507A.mix	250.9341316	289209.2682	289233.4566	251.9898921	250.9245656	24.18845521	1.065326512
R404A.mix	243.1500264	202163.4786	202232.6175	243.366103	242.4903483	69.13891423	0.875754681
R507A.mix	235.6367987	154824.7331	154859.2576	236.4775308	235.6339113	34.52446542	0.843619451
R507A.mix	240.9012428	193913.1158	193978.3	241.5002823	240.8963585	65.18420171	0.603923744
R404A.mix	255.3500214	325902.3562	325904.7201	255.351379	254.7530596	2.363878097	0.598319337
R410A.mix	260.9278003	529947.282	529947.2369	261.4154559	260.8323188	0.045167093	0.58313711
R507A.mix	254.9847256	336464.3549	336462.1653	255.5207536	254.9730511	2.189518309	0.547702565
R507A.mix	240.3722455	189670.5954	189734.5299	240.9012692	240.3675786	63.93448419	0.533690679
R507A.mix	240.2696508	188856.2116	188919.8556	240.7848147	240.2650255	63.64403836	0.519789221

dP = abs(pSatVapBicubic - pSatVapRefprop)
dT = abs(tSatLiqBicubic - tSatLiqRefprop)

[mapipolo]

If it's relevant, I'm using Refprop 9.12, the version that you provided at the beginning of last week.

[ibell]

I think I have mostly fixed these issues. Please confirm. It will never be perfect, but I think I made it much better.

[mapipolo]

It looks like your work was very effective. All discrepancies have been fixed to within an accuracy of 0.045°C. Thank you!