Verification Of Pilot-scale Iron Release Models

A model for the prediction of color release from a pilot distribution system was created in 2003 by Imran. This model allows prediction of the release of color from aged cast iron and galvanized steel pipes as a function of water quality and hydraulic residence time. Color was used as a surrogate measurement for iron, which exhibited a strong linear correlation. An anomaly of this model was an absence of a term to account for pH, due to the influent water being well stabilized. A new study was completed to evaluate the effectiveness of corrosion inhibitors against traditional adjustment. Two control lines were supplied with nearly same water qualities, one at pH close to pHs and one at pH well above pHs. The resulting data showed that effluent iron values were typically greater in the line with lower pH. The non-linear color model by Imran shows good agreement when the LSI was largely positive, but underpredicted the color release from the lower LSI line. A modification to the Larson Ratio proposed by Imran was able to give a reasonable agreement to the data at lower LSI values. LSI showed no definite relation to iron release, although a visual trend of higher LSI mitigating iron release can be seen. An iron flux model was also developed on the same pilot system by Mutoti. This model was based on a steady state mass balance of iron in a pipe. The constants for the model were empirically derived from experiments at different hydraulic conditions with a constant water quality. Experiments were assumed to reach steady state at 3 pipe volumes due to the near constant effluent turbidity achieved at this point. The model proposes that the iron flux under laminar flow conditions is constant, while the iron flux is linearly related to the Reynolds Number under turbulent conditions. This model incorporates the color release models developed by Imran to calculate flux values from different water qualities. A limited number of experiments were performed in the current study using desalinated and ground water sources at Reynolds Numbers ranging from 50 to 200. The results of these limited experiments showed that the iron flux for cast iron pipe was approximately one-half of the predicted values from Mutoti. This discrepancy may be caused by the more extensive flushing of the pipes performed on the current experiments which allowed attainment of a true steady state. Model changes were proposed to distinguish between near stagnant flow and the upper laminar region, with the upper laminar region showing a slight linear increase. Predictions using the galvanized flux model were not accurate due to an inferior color release model that was developed for galvanized pipes. The model exhibits a high dependence on sulfate concentrations, but concentrations of sulfates in the current experiments were low. This led to low predicted flux values when the actual data showed otherwise. A new galvanized model was developed from a combination of data from the original and current experiments. The predicted flux values using the new model showed great improvement over the old model, but the new model database was limited and the resulting model was not able to be independently tested