This post is the beginning of my exploration of a complex gas-liquid reaction system... reactive adsorption of NOx gases to produce nitric acid. Lately, I have become more interested in gas-liquid reactors. Also, I am getting more familiar with ways to incorporate Prode Properties into Mathcad programs and I think this problem will be a good test of the Mathcad/Prode combination.
Background on Nitric Acid Production
Nitric acid is produced by a three step process, (1) catalytic oxidation of ammonia with air, (2) condensation, and (3) reactive adsorption. In this series, I will concentrate on the last step.
In the absorber, NO and NO2 are further reacted to increase the concentration of nitric acid to about 60 wt %. The liquid phase reactions approach equilibrium, so early models assumed full equilibrium and then used low stage efficiency to compensate. Researchers then found that the NO oxidation to N2O4 step occurs mainly in the gas phase and is rate limited.
Some previous models
Carberry developed a tray column design method that solved the column from the bottom to top, starting with the bottom liquid design specification. He assumed the following reaction was in equilibrium in the liquid on each stage.
3 NO2 + H2O <----> 2 HNO3 + NO
A rate controlled reaction was used in the gas between each stage:
2 NO + O2 ----> N2O4.
Hüpen et al. assumed a more complex reaction network. They assumed that all liquid reactions were not equilibrium limited but were limited by reaction and mass transfer rates. The NO oxidation reaction above in the gas was reaction rate limited. Other gas reactions were assumed to be in equilibrium.
My first model
My initial goal is to see how well the Mathcad/Prode combination performs with a combined reaction and non-ideal phase equilibrium problem. I also want to see if a counter-current absorber model can be built with this equilibrium model. Thus, I have chosen to update the Carberry model so that it will simulate, not design. Whereas Carberry used manual algebraic solutions of the equilibrium problem, I will automate the process using the simultaneous equation solver in Mathcad.
Due to the complexity of this problem, I am going to simplify the problem at first and then add features (i.e. eliminate assumptions) to working program versions. In this manner, I will know where to look if there is a problem when a new feature has been added. This approach should also reveal how sensitive the model and process are to the assumptions.
When I have completed this first model, I plan to compare it to results reported by Carberry and Hüpen et al.
It's all about the journey
Although I have some idea where I'm headed, some of this journey will be in new territory for me. I may even find a severe limitation of either Mathcad or Prode Properties [or me!] that will terminate the trip. If that happens, I (we) will have learned something possibly valuable. It might also lead to another journey!
Hüpen, Bernhard, and Eugeny Y. Kenig. “Rigorous Modelling of NOx Absorption in Tray and Packed Columns.” Chemical Engineering Science 60, no. 22 (2005): 6462–71. doi:10.1016/j.ces.2005.04.060.
Carberry, James J., Chemical and Catalytic Reaction Engineering, Dover (2001)