Gibbs vs Equilibrium Reactors

[nrgeng]

Today I am comparing the Last Run Reports from the Gibbs Minimization Reactor, Reaction Type: Reaction Package, with the Equilibrium Reactor for the same methanol reactions discussed previously. The data do not compare favorably, in my opinion, while using identical operating data for each. I hope that you can help with my problem.

I obtained the equilibrium constant, Keq, data for the Equilibrium Reactor from a reliable proprietary source. The Last Run Data are the following:

......Unit.......Heat Duty.....Conversion
.................... kW.........CO........CO2
Gibbs (RP)......(-140)......(0.48).....(0.14)
Equilibrium.....(-578)......(0.98).....(0.67)

Any suggestions as to the deviations would likely add to my learning.

[jasper]

Also here, please send me the fsd file with the comparison. Without looking at the reactor configuration and results there is little I can see or say.

[nrgeng]

Your request to send the fsd file for the Equilibrium Reactor presents a dilemma for me in providing proprietary information.

If you had access to experimental data from a journal article or another reliable source of equilibrium data, then I would be willing to calculate the equilibrium relationship with temperature for each reaction. I would then set up a new simulation based on that experimental data for each reactor Unit type, and finally I would contact you with the results. The results should match my existing problem using the proprietary data. What do you think?

Do you have access to these data? If not, any other suggestions?

[jasper]

can you demonstrate the problem with factitious data?

[nrgeng]

I have experimental data for another reaction, hydration of ethylene, that has been published. I will set up a new simulation based on that experimental data for each reactor Unit type, then I will contact you with the results.

A set of two (Keq, T) data pairs for the reaction will be used to regress the data to determine Keq = f(T), then COCO will be used to obtain the reaction conversion.

Two data pairs for the reaction between 200 C and 500 C should work well by bracketing the operating point of 320 C which I will use in the simulation.

This will be a lot of trouble so I hope that it works! Any comments on my approach before I begin?

[jasper]

If it is a single reaction, the result of the isothermal Gibbs reactor can be easily checked; just plot the Gibbs free energy (including formation terms, so calculate from enthalpyF minus temperature times entropyF) vs the degree of reaction. The solution should be at a minimum. For the adiabatic reactor, entropyF should be at a maximum. The points can be generated by setting up the fixed conversion reactor with the same reaction, and manipulating the conversion. The required properties can be taken from the reactor's effluent.

Edit: the fixed conversion reactor in combination with a parametric study is probably the quickest way to generate the verification data.

[nrgeng]

Please refer to the attached table. From chemical reaction thermodynamics, the fact that the equilibrium conversion is the highest attainable, higher than the rate of conversion, causes concern because a low equilibrium conversion could render a project financially infeasible. The consistently low Gibbs Reactor conversions are suspect as are the higher conversions of the Equilibrium Reactor. Which data are to be used for financial decisions?

The data for Ethanol Synthesis are taken from the literature as experimentally determined [1,2]. This is superior, in my judgment, to theoretical determinations. I favor the results from the Equilibrium Reactor using Method 1 for Ethanol Synthesis.

If the results from the Equilibrium Reactor for Ethanol Synthesis are chosen, then should the results from the Equilibrium Reactor for Methanol Synthesis be chosen also as being from the most accurate Unit? The results of both Methods 1 and 3 for the Equilibrium Reactor for Ethanol Synthesis are close. This tends to sway my choice again toward the results for the Equilibrium Reactor for both Ethanol and Methanol Syntheses.

Any further analyses and/or comments are solicited.

.............................................................................................T.......Keq
1.....H. M. Stanley, et al., J. Soc. Chem. Ind., Vol. 53, p. 205 (1934)..........145 C...6.8E-02
2.....R. H. Bliss and B. F. Dodge, Ind. Eng. Chem., Vol. 29, p. 19 (1937).......320 C...1.9E-03

*************

Ethylene, Gibbs vs Equilibrium.PDF

Table of Results

(93.21 KiB) Downloaded 1087 times

*************
End of File

NB. Limited to three (3) attachments so omitted 'Ethanol, Method 3' file. See table for results.

[jasper]

After version 2.5.0.10, the CSTR may not have been giving you the right solution. Please validate your CSTR solution after updating.

Before deciding upon what to use for a financial analysis, it would be wise to understand the differences. Given reasonable thermo, I would expect the Gibbs and equilibrium results to be close to each other (if not the case, I suspect something is wrong with either the choice of thermo, or one of the two reactor configurations. Both should typically have a higher conversion than a reaction rate based model, if indeed the rate and equilibrium terms come from the same source. If the rate model and equilibrium model are consistent, one should obtain the equilibrium model results with the rate model results in an infinitely large reactor. So from a financial analysis perspective, the equilibrium solution will typically give you the upper limit of what you can attain. Typically a an initial study is performed on that basis, and while refining the study, reactor size and rate based limitations are introduced.

Experimental research are as good as the quality of the work done, theoretical results are as good as the thermodynamic model predictions. I would not necessarily favour one over the other. In fact I would be highly suspicious with a large mismatch.

Have you been able to verify the solution of the Gibbs reactor from a parametric study and the fixed conversion reactor? Typically this is easier with fewer degrees of freedom (e.g. 1 or 2 reactions).

[jasper]

What thermo did you use?

[HenkF]

Just to add to the discussion, and as I was curious about the 'correct' answer, I entered the components in my GIBBS reaction environment. (which is based on SOLGASMIX (as used with FACT SAGE) and known to be a very reliable Gibbs multi-phase routine) Regarding the outcome, this seems to be more in line with the GIBBS calculation routine as provided with COCO. Some temperature bias, due to a different datasources (I used the one that comes with HSC Chemistry). I'm more comfortable with high temperature inorganic processes, but please keep in mind that in predicting a reliable result, escpecially regarding GIBBS minimization, there is no easy hit and run solution. You actually need to add all possible products that may be formed, and need to know about their kinetic barriers. I added a temperature plot (at 50 bar)of the reactants.
Regards, Henk