ashraf sadraddini

Laboratory and field experiments have shown that dispersivity is one of the key parameters in contaminant transport in porous media and varies with elapsed time. This time-dependence can be shown using a time-variable dispersivity function. The advantage of this function as opposed to constant dispersivity is that it has at least two coefficients that increase the accuracy of the dispersivity prediction. In this study, longitudinal dispersivity values were obtained for the conservative NaCl solute transport in a laboratory porous medium saturated with tap water. The results showed that the longitudinal dispersivity initially increased with time (pre-asymptotic stage) and eventually reached a constant value (asymptotic stage). Four functions were used to investigate the time variations of dispersivity: linear, power, exponential and logarithmic. In general, because of the linear increase of dispersivity during a long time of transport, the linear function with R2=0.97 showed better time variations than the other three functions; the logarithmic function, having an asymptotic nature, predicted the asymptotic stage successfully (R2=0.95). The ratio of the longitudinal dispersivity to the medium length was not constant during the transport process and varied from 0.01 to 0.05 cm with elapsed time.

In multiphase flow systems of two immiscible fluids like air and water in a porous medium, e.g. an unsaturated soil, the flow properties depend on the amount and the spatial distribution of the phases within the pore spaces. At the pore scale, the phase distribution is controlled by the capillary forces depending on the pore size, surface tension, and wettability. For that reason, the relationship between the capillary pressure and liquid saturation (Pc–Sw relationship) is of high importance for the simulation of water flow in unsaturated soils. In this regard, dynamic pore network models can play a valuable role in predicting capillary pressure-saturation relationship. In the present study, numerical results from a dynamic pore network model which represents the void spaces of an unsaturated soil by a lattice of pores connected by throats are used to determine macroscopic relationships between capillary pressure and fluid saturations. Then using the resulted relationships from the pore network modeling and solving the partial differential Richard's equation using finite difference scheme, wetting patterns of drip irrigation have been simulated. Also, soil wetting patterns resulted from drip irrigation were investigated in laboratory experiments in a sandy soil with three dripper discharges rates of 2, 4 and 6 L.h-1 and different time intervals of 20, 40,… and 360 min. The performance of pore network model was evaluated by comparing the simulated wetting patterns with those of observed and simulated by HUDRUS software package. The results of current research showed that the pore network model for each of above-mentioned discharge rates had higher precision in comparison to HYDRUS model estimates of wetted pattern’s depth. With the exception of discharge rate of 2 L.h1, observed trend has been similar for estimation of wetted pattern’s radius and pore network model showed less error in comparison to HYDRUS model in estimation of wetted pattern’s radius.

Prediction of water table depth and outflow rate of drainage system are the most important issues in the literature of the drainage engineering. Three models of Kraijenhoff Van de Leur – Maasland، de Zeeuw - Hellinga and Glover – Dumm are frequently used for prediction of water level and discharge rate under unsteady state flow condition. For comparing the ability of these models an experiment was conducted using a laboratorial physical model. Data were collected for constant and variable head conditions in accordance to the boundary conditions of each model. The results showed good agreement with the observation data. For water level rising condition the Kraijenhoff Van de Leur – Maasland and de Zeeuw – Hellinga models، underestimated the outflow rate and overestimated the water table level as compared to the observed data، also Kraijenhoff Van de Leur – Maasland model''s prediction was better than that of de Zeeuw – Hellinga model. For the falling water table state the de Zeeuw – Hellinga model prediction was closer to the observation. The prediction accuracies of all three models decreased and converged while reaching the end of the experiment.

One of the main problems is the influence of surface water irrigation and non-uniform soil moisture profile below the soil surface which reduces the efficiency In this simulation، it was to influence both the water and irrigation. For this purpose، the surface flow equations and the equation of Nantes and the water in the porous medium equation to be done Rychadz Then، using the influence of surface irrigation was sirmod Simulation the flow of irrigation water and Results showed the model to solve the equation hydrus Rychadz error is less The accuracy of numerical simulations based on Richards equation for solving the given number of observations and comparison with moisture in the soil profiles developed in hydrus wet volume was evaluated RMSE hydrus results compared with observational values are)RMSE =0 / 11) and R2=. /92 in different depth Was achieved. The model also estimates hydrus appropriate volume of water to furrows adjacent furrows in the soil moisture distribution can be well simulated.