Open mostafa-razavi opened 5 years ago
@mostafa-razavi
I think the only way to get around this is to use the more standard fitting approach. Most molecular simulation studies estimate Pc by fitting the Antoine equation, rather than the Lee-Kesler equation. Although Antoine might not extrapolate as well from 0.85, it is more general. How much time would it take you to refit your vapor pressures to:
log(Psat) = A - (B/(C+T))
Then you just evaluate this fit at Tc.
@mostafa-razavi
If I have all the vapor pressure data and Tc values, I can do this fit for you very quickly and rewrite this section.
@mostafa-razavi
Since you have a lot on your plate, I am going to re-analyze the Tsat and Psat data you reported in the supporting information using the Antoine equation approach. I could compute my own Tc values, but to be consistent with your figures, would it be easy to provide me with Tc data? If not, I will fit it myself since it should be close enough.
Since you have a lot on your plate, I am going to re-analyze the Tsat and Psat data you reported in the supporting information using the Antoine equation approach. I could compute my own Tc values, but to be consistent with your figures, would it be easy to provide me with Tc data? If not, I will fit it myself since it should be close enough.
Thats would be great. The data in the SI are not the latest data though, but they are pretty close. Can you build the scripts in such a way that it's easy to replace the coexistence data with updated values?
The first three numbers are Pc[MPa], Tc[K], and acentric factor. The last three numbers are the uncertainties.
Mie-C12 2.4843e+00 6.8044e+02 2.2670e-01 1.4221e-02 4.3153e-01 1.7228e-04
TIP4P/2005 1.8576e+01 6.7344e+02 3.0431e-01 2.7904e-01 1.5189e+00 2.6246e-04
TraPPE-C1 4.8123e+00 1.9108e+02 1.6162e-01 2.0800e-02 1.4635e-01 2.5335e-04
TraPPE-C2 5.2823e+00 3.0274e+02 2.0879e-01 1.1013e-02 1.1039e-01 1.1104e-04
TraPPE-C12 2.3040e+00 6.6940e+02 2.3257e-01 6.1907e-03 1.8567e-01 1.0743e-04
TraPPE-iC4 4.1364e+00 4.0977e+02 2.2869e-01 9.3184e-03 1.5516e-01 1.8250e-04
TraPPE-iC6 3.5203e+00 5.0278e+02 2.3863e-01 8.1456e-03 1.7159e-01 1.8263e-04
@mostafa-razavi
Thats would be great.
Perfect, I am on it!
The data in the SI are not the latest data though, but they are pretty close. Can you build the scripts in such a way that it's easy to replace the coexistence data with updated values?
Yeah we will be able to repeat this process routinely with new data. So you are referring to the Tsat, Psat values that are outdated?
The first three numbers are Pc[MPa], Tc[K], and acentric factor. The last three numbers are the uncertainties.
OK, I will check my Tc against yours. Did you include all 5 temperatures in your rectilinear diameters/scaling fit? Or just the highest 3 or 4 temperatures?
Yeah we will be able to repeat this process routinely with new data. So you are referring to the Tsat, Psat values that are outdated?
Tsat, Psat, and rhoV
OK, I will check my Tc against yours. Did you include all 5 temperatures in your rectilinear diameters/scaling fit? Or just the highest 3 or 4 temperatures?
I used highest 4 temperatures.
@mostafa-razavi
OK, got it
@mostafa-razavi
I found one problem in the SI that you might have already caught. The data for TraPPE ethane and TraPPE dodecane are exactly the same (Tables XIV and XV).
@mostafa-razavi
So you only want the criticals for the seven sytems you posted above? You don't want me to analyze the gromacs data as well? They are not included in Figure 11, so I don't think we need them, but I just wanted to double check.
@mostafa-razavi
The first three numbers are Pc[MPa], Tc[K], and acentric factor. The last three numbers are the uncertainties.
I think the third value is actually rhoc (gm/ml), not acentric factor, right?
So you only want the criticals for the seven sytems you posted above? You don't want me to analyze the gromacs data as well? They are not included in Figure 11, so I don't think we need them, but I just wanted to double check.
I think those seven will suffice.
I think you're right.
I think the third value is actually rhoc (gm/ml), not acentric factor, right?
Yes, those are rhoC [g/cm^3]
@mostafa-razavi
So I have all the criticals now, and it will only take a few seconds to reprocess with new data.
Just for a sneak peak, this is how our critical values compare for 6 of the 7 systems (I need the correct ethane values). Note that the red blobs are random samples using the ITIC uncertainties with bootstrap re-sampling. They look like huge uncertainties but that is because I have 500 samples plotted.
@mostafa-razavi
Before we finalize the results, can you clarify what the +- values in the SI are. Are these standard deviations or 95% confidence intervals? I assumed they were standard deviations, which would lead to larger uncertainties.
@mostafa-razavi
I will throw together a short paragraph tonight explaining how to get Pc from the Antoine equation
I will throw together a short paragraph tonight explaining how to get Pc from the Antoine equation
Great. Thanks.
Before we finalize the results, can you clarify what the +- values in the SI are. Are these standard deviations or 95% confidence intervals? I assumed they were standard deviations, which would lead to larger uncertainties.
Those are still standard deviations.
TraPPE-C2 results
Tsat | Tsat+/- | Psat | Psat+/- | rhoL | rhoV | rhoV+/- | Hvap | Hvap+/- |
---|---|---|---|---|---|---|---|---|
259.7700000 | 2.4562000 | 2.0436000 | 0.1272500 | 0.4286000 | 0.0388300 | 0.0029145 | 9.5665000 | 0.1246400 |
235.1200000 | 2.0689000 | 1.0375000 | 0.0651210 | 0.4714000 | 0.0189220 | 0.0011943 | 11.2430000 | 0.0393570 |
206.8000000 | 1.0402000 | 0.3930300 | 0.0176810 | 0.5143000 | 0.0074593 | 0.0003208 | 12.6240000 | 0.0137160 |
173.8000000 | 1.0855000 | 0.0841210 | 0.0050847 | 0.5571000 | 0.0017975 | 0.0001003 | 13.7980000 | 0.0092920 |
137.5400000 | 0.3474200 | 0.0059114 | 0.0002075 | 0.6000000 | 0.0001560 | 0.0000051 | 14.8520000 | 0.0044012 |
@mostafa-razavi
I have attached a tex file with a simple paragraph. You will need to include references to the rest of the manuscript.
@mostafa-razavi
So when will you be able to give me the updated ITIC values for me to analyze?
@mostafa-razavi
Here are the TraPPE C2 results:
@mostafa-razavi
Here are some parity plots comparing our critical constants. For the most part our Tc and rhoc agree very nicely (as would be expected). However, there is a clear shift in Pc where the Antoine equation predicts slightly lower values. This is most pronounced for water. I will look for literature values to see which is better. Even if Antoine is worse, I think it is the normal method and would satisfy the reviewer.
@mostafa-razavi
Hmmm... so this is supposed to be a very accurate study for computing criticals with TIP4P models. Notice that there TIP4P/2005 Pc value (14.6 MPa) is much less than ours (Mostafa: 18.8 MPa, Rich: 17.0 MPa). Then again, their Tc value (640 K) is way less than ours (Mostafa: 673.4 K, Rich: 675.9 K)
@mostafa-razavi
I am looking into the values on the NIST standard reference database. They appear to cite a manuscript from the same group that is one year older than this one...
@mostafa-razavi
The manuscript I cited used a slightly more expanded version of rectilinear and scaling:
So this could cause some discrepancy, but I would be surprised by the 30 K difference. I am going to analyze their data myself.
@mostafa-razavi
I analyzed the data from the article I reference above. I only included data for T > 510 K (roughly 0.75 Tc) I get more similar results to them:
My values: Tc = 646 +- 4 K Pc = 16.3 +- 0.6 MPa rhoc = 312 +- 5 kg/m3
Their values: Tc = 640 K Pc = 14.6 MPa rhoc = 310 kg/m3
However, if I include all of their data I get results that are essentially what we got from ITIC:
But that is because the rectilinear and scaling fit is terrible for TIP4P over the entire temperature range:
@mostafa-razavi
I think this explains the discrepancy in our critical constants. We really shouldn't use the scaling/rectilinear fit approach over such a wide temperature range. I could try using just the three highest ITIC points...
@mostafa-razavi
Here is top three temperatures (one point is still below the 510 K cutoff):
@mostafa-razavi
I think we need to directly mention that this approach, rectilinear diameters in particular, is not adequate for TIP4P.
Note that they also used Antoine:
@ramess101
New data: All-version2-single-molec.xlsx
The new paragraph written by @ramess101 discussing Pc calculation:
@mostafa-razavi
Make sure the link is fixed (Equation ?? should be Equation 27.).
Also, I think I want to rewrite the second paragraph a bit:
Some caution should be exercised when extrapolating Equations 25-27 as ITIC data are typically not available near the critical point. For example, we recommend excluding low temperature ITIC data (Tr < 0.6) when fitting Equations 25-27 as these equations are typically not reliable over the entire temperature range. The results in Section 7 demonstrate that... (keep the rest the same).
Can you send me the updated version to make sure it transferred properly.
Good?
@mostafa-razavi
Text looks good. But we should include this reference for the Antoine equation:
J. S. Rowlinson and F. L. Swinton, Liquids and Liquid Mixtures. Butterworths, London, 1982
@mostafa-razavi
Just provide me with the updated ITIC values for single molecule (with uncertainties) and I can get you the new critical constants
Text looks good. But we should include this reference for the Antoine equation:
J. S. Rowlinson and F. L. Swinton, Liquids and Liquid Mixtures. Butterworths, London, 1982
Done!
@ramess101 Attached is the final ITIC coexistence points.
@mostafa-razavi
OK, so are all of these tabs single molecule values? Some are from the literature, right? Some are from gmx, are those corrected too?
@mostafa-razavi
OK, it looks like they are different from what I had before (even the smaller compounds). So I am going to assume they are all single molecules.
OK, it looks like they are different from what I had before (even the smaller compounds). So I am going to assume they are all single molecules.
Only C12 and isohexane are single molecule, the rest are Uintra.
@mostafa-razavi
OK. The values were different than the old SI, but now I get almost exact agreement with your previous values for the Uintra systems (alright maybe not exact, but slightly better).
@mostafa-razavi
How would you like the updated critical constants? Currently I have a separate file for each compound with columns:
Pc uPc (MPa) Tc uTc (K) rhoc urhoc (gm/ml)
@ramess101 I can put them together. Do we need to report the numbers in addition to Figure 11?
@mostafa-razavi
We should include them in SI before publication. But don't worry about that before resubmitting if you don't have time.
@mostafa-razavi
I am working on the response to reviewer 2 comment 2.5. Should have it done in 30 minutes.
I am working on the response to reviewer 2 comment 2.5. Should have it done in 30 minutes.
Great!
@mostafa-razavi
Here are the data files. Make sure to post the updated Figure 11 so we can verify that they were transferred properly. Also, ignore the literature TIP4P file, I included that for comparison. I still think we will want to include just literature simulation values instead of REFPROP criticals.
@ramess101 Anything to change in this paragraph in Example Simulations section?
Looks good.
On Friday, December 21, 2018, mostafa-razavi notifications@github.com wrote:
@ramess101 https://github.com/ramess101 Anything to change in this paragraph in Example Simulations section?
[image: image] https://user-images.githubusercontent.com/16358113/50368530-b34d6c00-0557-11e9-8935-c7951570f641.png
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/mostafa-razavi/ITIC-paper/issues/23#issuecomment-449530965, or mute the thread https://github.com/notifications/unsubscribe-auth/AWUvhJqSJwUAiP957k7PohDl_qlEnohQks5u7X8cgaJpZM4ZZFtu .
I personally want to keep the critical properties in the plots. They are not exact but reasonable estimates, even though our highest Tr is 0.85. If we can get out of this by adding a sentence or two, like the reviewer is suggesting, how would you respond to the reviewer's above comment?