Initial Conditions Description (G)

Overview

Initial conditions are the mechanism by which the user specifies the state of the system at the start of the simulation. The required number of inputs for each grid cell depends on the operational mode, but as a minimum always equals the number of unknowns per grid cell, unless equation closure can be used to eliminate a parameter. For example, phase saturations must always sum to one. Therefore, saturations are only required for the number of phases minus one. The number of initial condition parameters that can be specified is determined by the number of degrees of freedom from the Gibbs phase rule. Some operational modes have special algorithms for specifying initial conditions, (e.g., STOMP-CO2 and the hydrostatic initial condition), but all operational modes accept specifications via I,J,K indexing and X-, Y-, and Z-Direction gradients. Initial condition parameters are set at the volumetric centroid of the grid cell. Initial fluxes on the grid-cell surfaces can not be set directly, only state variables at the grid-cell centroids can be specified. Normal execution mode simulations require initial conditions to be specified for the entire domain. Restart execution mode simulations do not require the initial conditions card, as the state of the system is recorded in the restart file. For restart execution mode simulations, however, the Initial Conditions Card can be used to overwrite parameters stored in the restart file. The default initial temperature is 20˚C. The default initial gas pressure is 101325. Pa. The default solute and species concentration is zero. Solute and species initial concentrations need not be specified if they are zero.

Application Notes

Initial conditions are specified by indicating a parameter name, a domain, and directional gradients. The domain is specified by a range of I, J, and K indices (e.g., I start, I end, J start, J end, K start, K end). For domains that span more than one grid cell, the specified parameter is assigned first to the grid-cell with the lowest I, J, K index, at the volumetric centroid of that grid cell. Initial condition values at the remaining grid-cells within the domain are computed from the specified value and the directional gradients. Directional gradients have units of 1/length. Units for the numerator for the directional gradient are those of the specified parameter. For example, if the initial condition parameter was temperature, with units of ˚C, and the z-direction gradient had units of 1/km, then the temperature gradient in the z-direction would have units of ˚C/km. Gradients are applied in three directions. The default gradient is zero, indicating no variation in the parameter value with distance. X-direction gradients are applied from the global x-coordinate value at the centroid of the grid cell with the lowest I, J, K index to the global x-coordinate value at the centroid of a grid cell within the domain. The same applies for y- and z-direction gradients. Gradients can be specified with a sign. Negative gradients have parameters that decrease with increasing coordinate directions and positive gradients have increasing parameters with increasing coordinate directions.

For most operational modes, the Initial Conditions Card is designed to provide the user with options for specifying the state of the system. For example, in STOMP-W the aqueous saturation and gas saturation must sum to one, and the aqueous saturation is related to the difference in aqueous and gas pressure. These relationships allow the user to specify two of three parameters: 1) aqueous pressure, 2) gas pressure, or 3) aqueous saturation, with the third parameter being computed by STOMP using the defined relationship between aqueous saturation and capillary head, via the Saturation Function Card. Whereas this approach gives the user flexibility in assigning initial conditions, care must be taken to ensure the specification is the user's intent. A good example of this would be if it was desired to create hydrostatic conditions with a water table located somewhere in the vertical (the z-direction) center of the computational domain. It might first appear as if this could be achieved by using the default gas pressure of 101325 Pa everywhere, specifying an aqueous saturation of 1.0 below the water table and then applying a saturation gradient in the z-direction above the water table. This specification, however, would not yield hydrostatic conditions, as the aqueous pressure below the water table would have been set to 101325 Pa for those grid cells with an initial aqueous saturation of 1.0.