2. Results and discussion
The results of Table 5 show that the value of average error for
mathematical model is about 0.09% and for the design of experiments
model is about 1.35%. There is no significant difference in the
error of the two methods, and the error of the developed design of
experiments model is acceptable in engineering applications; and
also it should be noted that BoxeBehnken design method needs a
few number of experiments (field data); therefore it could be used
as a data-acquisition and data-analyzer model in gas injection
projects. In this study, For instance, sixty data were used for the
mathematical model but BoxeBehnken model only required fifteen
experimental data (3 data for each well) to develop a model.
To debate the merit of wells, response of each pair of wells to gas
injection-while the other wells are injected with average rates-has
been studied. The performance results of pair of wells are shown in
Fig. 3 (hold values are the average gas injection rates i.e. level 0, for
each well):