Correlations among variables of the freeze decontamination process have not been developed to precisely predict experimental observations of purified solid ice out of a given aqueous solution on a macroscale. A lack of such information has been one of the main reasons for the very small practical application of the process so far. A simplified model has been presented in an effort to take into account the inhomogeneity of the solid phase. Pure ice spikes scattered on a cold surface have been supposed to grow opposite to the direction of heat dissipation, leaving unfrozen liquid channels among them along their length. The ice phase has thus been supposed to consist of pure ice spikes and a quantity of liquid entrapped among them. The liquid at the freezing front and in bulk is supposed to be completely mixed. Volume changes on freezing have also been taken into account. The model has been compared with experimental observations on a specially fabricated apparatus. Temperatures of a cold freezing surface have been continuously lowered while following a -0.0017 K/s gradient during experiments. An ultrasonic mixing technique has been applied to the liquid during freezing to eliminate impurity (cesium nitrate) concentration gradients in the liquid. A comparison of the model with experimental observations has been made with the help of a decontamination factor (DF) [concentration of impurity in the initial liquid/concentration of impurity in the frozen mass (after it has been melted)] versus V/V-0 (volume of unfrozen liquid left at any point during freezing/volume of initial liquid taken) plots. Although a scatter in the experimental results has been observed, there was broad agreement with the model trend. The average value of DF has been as high as 46. The investigation has highlighted the importance of the factors that determine the fraction of pure ice spikes in the growing solid at any point during freezing.