PhreeqcUsers Discussion Forum

Processes => Inverse modelling => Topic started by: NannaG on March 04, 2019, 02:22:03 PM

Title: Inverse Modelling - uncertainty errors
Post by: NannaG on March 04, 2019, 02:22:03 PM
Hey All

Can anybody help me with my phreeqC script. I am trying to do an inverse modelling with two solutions, but I seem to get uncertainty errors if my uncertainty is lower than 0.9.

If I set the uncertainty to 1, phreeqC is able to run the inverse modelling. But I would like the uncertainty to be lower. 
The following is the errors i get. 

ERROR: Not possible to balance solution 1 with input uncertainties.
ERROR: Not possible to balance solution 2 with input uncertainties.

Thanks a lot.
Title: Re: Inverse Modelling - uncertainty errors
Post by: dlparkhurst on March 04, 2019, 05:21:00 PM
By default, the inverse modeling calculation includes mole balance equations for each element contained in the list of -phases. If you do not include a phase with Cl, for example, there will be no mole balance for Cl. PHREEQC will try to charge balance solution 1 and 2 with the elements that have mole balance equations. In your case Cl, S, and Mg will be missing and a very high uncertainty will be needed to charge balance without these elements.

However, your second solution has a large charge balance error, about 20 percent. Therefore, the uncertainty for solution 2 must be at least 0.2 to allow the concentrations to be adjusted to charge balance.

Below, I include extra phases and large enough uncertainties to achieve a model. You should not consider it a plausible model without more consideration. Think of the phases halite, gypsum, and sepiolite as sources or sinks of Cl, SO4, and Mg. You must consider what are reasonable sources and sinks of these elements in your system. A little bit of O2 is needed to make the redox relations work out. I also ignored the water produced or consumed in the reactions to avoid unreasonably large mole transfers. I am not sure quite why this is necessary, but perhaps it is related to the large charge imbalance in solution 2.