Geomodels (GM)

Description

Geomodels Research Centre regroups several universities and public research organizations having the purpose of applying and developing new geosciences technologies in order to accurately understand and characterise a) the systems and processes that determine the creation, location, and the quality of the geological resources and reservoirs with the focus on the hydrogeologic and energetic reserves, b) the ground geo-mechanical behaviour, and c) the phenomena related to the geological risks and their effects on the land.

The geological processes modelling will have the main objective of quantifying and forecast spatial and time phenomena. The research work focuses the aspects of the three-dimensional geological bodies, the numerical simulation of the processes taking place in these bodies, and risk analysis.

Researchers

• J. Carrera (UPC)
• P. Batlle (UPC)
• R. Gogu (UPC)

Context

• Inverse problem
• General hydrology

Projects

1. PIM
2. PROSIT
3. 3D hydrogeologic modeling of sedimentary media
4. Hydrogeological spatial database concept for Barcelona
5. Geological spatial database system as a first step in generating accurate 3D sedimentary geological models
6. Use of GML to code hydrogeological modeling data
7. Stability and settlement modeling in landfill

Description "PROSIT (PROCESS ORIENTED SIMULATION AND OPTIMIZATION TOOL)"- geomodels

What is it?:

Prosit means “Process Oriented Simulation and Optimization Tool”. It is a computer program, designed as a flexible and extensible framework, to solve groundware-related simulation and optimization problems. One of the aims is to make it easy to expand the program to incorporate new methods and algorithms, e.g. numerical methods, optimization algorithms and other linear system solvers.

There are several motivations for providing a new framework. Both the increase of understanding of the physics of the environment and the increased computational power of current computer systems allow for the modelling of more processes and for coupling more models. In addition, the advances in numerical mathematics provide a growing toolbox of spatial discretization methods, adaptive time stepping schemes, numerical solvers etcetera. These tools make us better equipped to deal with issues such as numerical oscillations and convergence problems.

In order to take the fullest advantage of available methodology and tools, we must not only be able to implement new mathematical models and modelling tools into software, but we must also be able to combine them with other, existing models and tools.

The Prosit program is designed according to the Object Oriented programming paradigm and thus makes use encapsulation to avoid unnecessary coupling of modules and of inheritance to allow code-reuse and easy expansion. The class definition has implemented so that concepts of the mathematical modelling process are represented by classes. A special feature of Prosit, from which it derives its name, is that the central direct problem class is not the one that represents the partial differential equation, but a class called “process”, which represents a natural process such as diffusion or advection and that contributes to such a partial differential equation.

Where are we?:

Right now, the finite element numerical method has been implemented for the modelling. It has been validated with the Henry problem, analogue to a simplified saltwater intrusion. For the optimization problem the Levenberg-Marquardt algorithm has been implemented. Classes have been built for storing and manipulating spatial data, time functions, generic nonlinear problems, temporal discretization etc.

Work in progress:

The expansion capability is now being put to test by implementing two expansions. First, work is being carried out to implement a multipoint-geostatistics (MPG) simulation algorithm into prosit. This will enable the addition of more geological knowledge into the simulation process.

Second, we are implementing the mixed element method. This is an eulerian numerical method for solving partial differential equations that has significant differences with the finite element method: it produces a continuous flow field, and the basis functions are associated to sides of the element and not to nodes.