Matrices in Petroleum Engineering

Embed Size (px)

Citation preview

  • 7/30/2019 Matrices in Petroleum Engineering

    1/3

    Matrices in Petroleum EngineeringUsing Tridiagonal Matrices to ModelReservoirs

    Juan F. Pea

    Oct. 10, 2003

    Juan Pena

  • 7/30/2019 Matrices in Petroleum Engineering

    2/3

    Abstract:

    Linear algebra is used in engineering to solve problems that

    involve multiple complex equations. In the petroleum

    industry, engineers must model reservoirs to find theconditions needed to maximize recovery of hydrocarbons.

    Reservoir simulation involves many complex partial

    differential equations, which are solved using a computer

    program that models the reservoir being studied. To thepetroleum engineer it is great that the computer program

    does all the simulating for him as long as all the correct

    parameters are put into the program. We will look at howreservoir simulators use linear algebra to model reservoirs.

    A hydrocarbon reservoir is a porous media through which hydrocarbons flow to a

    well or many wells. It is simple to measure a pipes flow capacity as a function of

    pressure; however, porous media flow is different because there are no certain paths

    through which the fluid will travel. Reservoir models look at fluid flow through the

    reservoirs by determining how saturations and pressures change as functions of time.

    Nonlinear partial differential equations make up the fluid flow equations and since they

    are not linear they have to be transformed into linear forms using the Taylors series. The

    Taylors series makes the derivatives finite differences, which are changes in x and

    changes in t. Since the reservoir model is divided into blocks or cells, there is an

    equation for each cell. In a model with N cells, we get N equations with N unknowns,

    which is a lot of equations and unknowns to be solved by hand. The equations can be

    written in a matrix form called a tridiagonal matrix, which has three diagonal elements

    and zeros in all the off-diagonal elements.

    Cell1 a1Po b1P1+c1P2 =d1

    2 a2P1 b2P2+c2P3 =d2

    3 a3P2 b3P3+c3P4 =d3N anPn-1 bnPn+CnPn+1 =dn

    Juan Pena

  • 7/30/2019 Matrices in Petroleum Engineering

    3/3

    AP=d

    [ a1 \ \ ] [ ] [ ][ \ b1 \ ] *[p1] = [d1]

    [ \ \ c1 ] [ ] [ ]

    Then by using the Thomas algorithm, a modified form of Gaussian elimination,

    we solve the equations for the unknown pressures P. Once the pressures are known the

    results are given as functions of time. The idea behind reservoir modeling is to examine

    how a reservoir will do at some future time. The engineer identifies which cells contain

    wells, whether they are producers or injector. By running the simulator, the engineer gets

    a table of results, which includes saturations, pressures, and amount of fluids left by the

    time they are occurring. This allows an engineer to determine when in the future is a

    good time to stop producing. Modeling is very important to the petroleum engineer

    because there is no other way of determining how a reservoir will perform under certain

    conditions. Since reservoirs last long times, it is impossible to wait to see what a

    reservoir will do, so the engineers just model the reservoirs and can determine many

    important conditions. Some of the conditions to a reservoir are: where to place the wells,

    how many wells to have, what spacing between wells, at what rates the wells should be

    produced, and how long the wells should be produced. Modeling simply makes a

    petroleum engineers job easier, and without linear algebra, modeling would be very

    difficult to perform.

    References:Crichlow, Henry B. Modern Reservoir Engineering-A Simulation Approach. Prentice

    Hall. New Jersey 1977

    Fanchi, John R. Principles of Applied Reservoir Simulation. Gulf Publishing Co.

    Houston, TX. 1997

    Juan Pena