Analise automatizada de testes de pressã

AUTOR(ES)
DATA DE PUBLICAÇÃO

1992

RESUMO

The objective of an automated well test analysis is to obtain íormation parameters by using nonlinear regression techniques applied to bottom hole pressure versus time data collected during the test period. In this work several nonlinear regression methods were analysed and their performances compared. Some of the methods considered are based on the L2-norm minimization, as the least squares and the singular value decomposition. Other methods minimize the Ll-norm, such as the least absolute values (LAV), modified least absolute values (MLAV), the combination (MLAV-LAV) and the Nelder-Mead s method. A new form of penalty-function was proposed to constrain the optimization in methods based on the solution of squared linear systems of equations. For methods based on the solution of over determined linear systems of equations, for which conventional penality-functions can not be applied, changes in the domains of the unknown parameters were proposed in order to restrict the minimization to feasible (physically and mathematically) regions. The modified least absolute values (MLAV) and the combined method (MLAVLA V) presented the best performance among alll methods considered in the study, concerning number of iterations, convergence from poor initial estimates or robustness with respect to infeasible regions. Two reservoir models were analysed: infinite homogeneous reservoir and infinite double-porosity reservoir. The results of the nonlinear para.meter estimation using either pressure data, pressure derivative or pressure and derivative data were compared. The use of both pressure and derivative as the model function yielded better performance than using I only pressure or pressure derivative data. Various examples with synthetic and field data were discussed

ASSUNTO(S)

minimos quadrados poços de petroleo analise de regressão

ACESSO AO ARTIGO

Documentos Relacionados