Evaluating simple chromatographic formulae for Sor and EOR efficiency with a numerical SWCT model

Some of the residual oil after primary and secondary production can be extracted by enhanced oil recovery (EOR) methods. Before a full-scale EOR campaign is initiated, it is important to first perform an EOR pilot to estimate the residual oil saturation (Sor), before and after the EOR pilot. The reduction in S_or yields the expected efficiency of the full-scale EOR operation. A common method to estimate S_or is the Single Well Chemical Tracer (SWCT) test. Although there exist numerical models of SWCT tests, simple chromatographic formulae are still widely used to estimate S_or. Here, we study how well peak concentration time and mean residence time formulae perform in a chemical EOR injection scenario. Before the EOR injection S_or equals 22 % and afterwards 11 %, i.e., the real EOR efficiency is 22-11 = 11 %. Firstly, we generate synthetic tracer curves from an axial symmetric numerical SWCT model with parameters based on averages of input data from 39 published SWCT tests. The numerical model includes turbulent fluid flow and temperature in the wellbore, and fluid flow, temperature and chemical reactions in the oil-bearing formation. We use the initial reservoir temperature and the reservoir injection temperature from a simple analytical model in the formulae for the oil-water distribution constant of ethyl acetate to estimate S_or since these will be easily available for the interpreter. Using the peaks of the tracer concentration curves yields S_or estimates that are 0–4 % too low, whereas the mean residence time formula results in much larger errors of 4–12 %. The reduction in the pre-EOR S_or estimates due to pre-flushing is between 0 and 1 % and for the post-EOR S_or estimate between 2 and 3 %. The expected EOR efficiency, i.e., the difference between the pre- and post-EOR S_or values using tracer peaks is 10–11 % with the analytical temperature formula (10-12 % with the initial reservoir temperature). With the mean residence time formula, we obtain 4–6 % (5–6 % with the initial reservoir temperature). Since the true reduction in S_or is 11 %, we note that the peak time formula yields an excellent result, whereas the mean residence time formula yields poorer results. These results are when dissolution of calcite prevents pH to fall much below the region with lowest rates of hydrolysis of ethyl acetate. When there is only a low pH buffer capacity in the target formation, the difference between the tracer peaks and the mean residence time formulae is only about 1 %. The major limitation of the model is the assumption of axial symmetry. If complex geological settings, fractures and faults, horizontal wells etc. are to be investigated, a full 3D model is required.