The use of experimental designs for corrosive oilfield systems

A Design of Experiment approach was used to investigate the effect of hydrogen sulfide, carbon dioxide and brine composition on the corrosion rate of carbon steel. Three of the most common experimental design approaches (Full Factorial, Taguchi L{sub 4}, and Alternate Fractional) were used to evaluate the results. This work concluded that: CO{sub 2} and brine both have significant main and two-factor effects on corrosion rate, H{sub 2}S concentration has a moderate effect on corrosion rate, and higher total dissolved solids (TDS) brine compositions appear to force gases out of solution, thereby decreasing the corrosion rate of carbon steel. The Full Factorial Design correctly identified all independent variables and the significant interactions between CO{sub 2}/H{sub 2}S and CO{sub 2}/Brine on corrosion rate. The two fractional factorial experimental methods resulted in incorrect conclusions. The Taguchi L{sub 4} method gave misleading results as it did not identify H{sub 2}S as having a positive effect on corrosion rate, and only identified the strong interactions in the experimental matrix. The Alternative Fractional design also yielded incorrect interpretations with regard to the effect of brine on corrosion. This study has shown that reduced experimental designs (e.g., half fractional) may be inappropriate for distinguishing the synergisticmore » interactions likely to form in chemically reactive systems. Therefore, based upon the size of the data set collected in this work, the authors recommend that full factorial designs be used for corrosion evaluations. When the number of experimental variables make it impractical to perform a full factorial design, the aliasing relationships should be carefully evaluated.« less