A probabilistic approach to modeling struvite precipitation with uncertain equilibrium parameters

- Published constants related to struvite solubility can vary by an order of magnitude.
- Significant uncertainty also exists in wastewater concentrations and parameters.
- Struvite precipitation model uncertainty was evaluated in a Monte Carlo framework.
- Model projections and associated uncertainty agreed with treatment plant data.
- Proposed method can be used to evaluate uncertainty in any design situation.

The precipitation and accumulation of struvite (MgNH4PO4·6H2O) within anaerobic processes has been a costly problem for wastewater facilities. To anticipate and mitigate struvite buildup, solubility models have been developed that employ equilibria software for the evaluation of equilibrium equations. Unfortunately, these programs run under the assumption that chemical equilibrium constants are single, universally accepted values when, in reality, a wide range of values have been published for these constants.In this study, a struvite solubility model was developed in which the equilibrium constants were treated as empirically distributed variables within a Monte Carlo simulation to understand the effect of uncertainty on precipitation potential over a range of pH (6-8.5), temperature (0-60 °C), and ionic strength (0-1 M). Using field conditions measured at a struvite-afflicted treatment plant as model input parameters, the resulting uncertainty in the struvite supersaturation ratio was found to be highly consequential, with the 90 percent confidence interval spanning well over an order of magnitude. Additionally, a sensitivity analysis was performed on the model, identifying the third orthophosphate equilibrium constant and the struvite solubility product as the most significant source of uncertainty.