While a two dimensional model such as we used earlier is typical of those used in most applications, the use of such a model implies that the geological materials do not vary significantly with depth and that there is not a significant vertical gradient in the hydraulic head. We also saw the need to resolve vertical details of contaminant transport which cannot be done with a two dimensional model. When we need more vertical details it is necessary to add layers to the model.
Once again, as suggested earlier, think of each layer as a layer of a layer cake. Up until now we have had just one layer for our layer cake. Multiple layers can be used to represent the vertical components of the hydraulic head or to more faithfully represent a vertically varying or layered geology.
As mentioned earlier there are two types of layers in MAGNET: computational layers and geological layers. Computational layers are used to sub-divide a single geological unit into more than one “grid layer”. Geological layers are used to differentiate units with substantially different geologic characteristics (e.g., aquifer vs. aquitard).
One way to think about the difference between computational layers and geological layers is to imagine a Rubix cube with the following characteristics. Imagine the Rubix cube is made from six square layers stacked one upon the other, each containing 36 small cubes. In total we have 6×6×6 = 216 little cubes.
Now further imagine that the first and second layers are red, the middle two layers are blue and the bottom two layers are yellow. In MAGNET the three sets of colored layers, each having a thickness of two layers of the same color, would represent geological materials, perhaps the red could be sand, the blue clay and the yellow gravel. In this scenario each geologic layer would be represented by two layers of 36 small cubes.
Now consider the six layers, independent of their color. These six layers are the computational layers and exist only to allow for a more accurate representation of flow behavior in the vertical direction. The more layers, the thinner each cube is and this results in more accuracy in the simulated results. However the more computational layers the more work that is required to solve the system of equations. For example, in the one layer case we would have 36 equations to solve and in the six layer case 216 equations.