WORKING WITH OIL WELLS PRODUCING GAS BELOW THE BUBBLE POINT IN TIGHT ROCK In Daedalus, you must remember that the Base math engines are only suitable for gas or still oil situations. If you are modeling oil wells that are, or are expected at some point to be, producing gas below the bubble point, then you must use one of the Multiphase engines. We suggest using the Multiphase-Parallel engine because, especially for long lateral wells with multiple frac clusters, it will always run faster. You will also want to adjust the Relative Permeability curves to manage the gas production below the bubble point. When using the Rel Perm feature, the Krg curve must begin at the far lower left corner of the graph if the gas is to be produced through the wellbore right away, i.e. as soon as the bubble point is passed and gas starts to be released out of the oil. If it is set over to the right in any amount, the model will delay gas production and allow the initially released gas to remain in the reservoir. The other big item that affects this type of case is the bubble point and how that value is calculated in our system. In Daedalus, the GOR is calculated based on the average reservoir pressure. In other words, the average pressure across the entire reservoir must reach the bubble point before gas will start to be produced based on the Rel Perm curves. However, in very tight rock, the bubble point is often reached much more quickly in the near-wellbore region, so gas begins coming out of solution and being produced quite soon; often sooner than the model is able to predict, since the average pressure is always going to be higher than the near-wellbore pressure. The bubble point and other correlated values in the model can be adjusted by modifying the oil gravity, gas gravity, the separator pressure and/or the discovery GOR. By making such adjustments, you can (artificially) raise the bubble point to accelerate the time at which the model begins releasing gas, thereby achieving a short term history match on both the pressure and the GOR. However, it must be remembered that these results were achieved by changing the fluid values and/or separator pressure to something that they are actually not, so this approach will not yield a valid solution over the long term. Because Daedalus does not accurately account for the early-time GOR due to its inability to isolate and apply the near-wellbore pressures to determine BPP the cross-over time and calculate the resulting gas production, you can expect it to under-estimate the cumulative gas forecast. Nevertheless, over the longer term, this early-time effect becomes negated and Daedalus will provide a good overall solution. To illustrate this effect, the following graphics were provided by Apache Corp. and demonstrate how Daedalus differs from CMG-IMEX in modeling this type of situation. The first image below shows how the model is set up in IMEX.
The next graphic shows how you would model the same situation in Daedalus. DAEDALUS:
The third image shows the comparative results. Note that Daedalus misses the early time GOR, but after about one year, it matches the IMEX forecast quite well. Because of this effect, Daedalus also tends to over-state the initial oil production by a bit, but then has a slightly steeper decline slope, so the long term production forecast equalizes out (in this case, the cumulative oil lines converge after about 5 years). The same pattern can be observed for other wells, as shown in the following example. IMEX:
DAEDALUS: These results show in more detail how Daedalus doesn t catch the early gas production (because of the average reservoir pressure vs. near wellbore pressure situation previously discussed), but the model does settle into a very similar forecast after the ARP passes the BPP and stabilizes at about 2 years into the model. Also, the long term cumulative production forecasts converge after about 7 years.
SUMMARY For lateral wells with multiple fractures completed in tight shales, Daedalus accounts for oil (and gas) recovery somewhat differently than IMEX, but the two models are almost identical in forecasting the longer term cumulative production (in these cases, about 5-7 years and beyond). Although the model can accurately determine the pressure at any point in the reservoir and at any time, it remains a limitation that it cannot isolate near-wellbore pressures and use them to calculate the GOR. The model has to use the average reservoir pressure for that purpose. Nevertheless, the model still provides a good long term EUR solution, as shown by the examples above. GRM is looking at potential ways to more accurately handle the short term near-wellbore pressure effects and all users will be notified at such time as the solution is released.