Workflow:

 This project presented the latest development in pressure-rate deconvolution and its applicability in reservoir characterisation and reservoir simulation of the Corrib Dry Gas field, Offshore Ireland. The field, with an estimated GIIP of 1.20 TCF, came onstream in 2015 at 350 MMscfd. It is currently producing off-plateau from 6 wells. The field is operated by Vermilion Energy and JV partners are Nephin Energy and Equinor.

OPC working on behalf of Nephin Energy were focussing on updating the existing Corrib numerical simulation model. The objective was to incorporate the results of the extensive PTA work which included pressure-rate deconvolution into this simulation model and create a framework for history matching. The study presented in this document is focussed on data obtained from well P101z.

Pressure rate deconvolution is a method for converting pressure and rate data obtained from a well with a variable rate history nto a much simpler form of constant-rate drawdown-pressure response function. The advantage of pressure rate deconvolution over classical PTA methods is it adds thousands of hours to a Bourdet derivative which greatly improves identification of the flow regimes required for evaluation of relevant Well/Reservoir Parameters in particular the connected volume to the well. The results of pressure rate deconvolution using the Convolution Explorer software is shown in (Figure 1).

The pressure and derivative curves can then be analysed using the basic PTA equations originally developed for a well producing at constant rate. The analysis of the results from Pressure-rate deconvolution allow evaluation of formation permeability, reservoir heterogeneities, shape of reservoir compartment drained by the well, quality of the completion (Skin) and analysis of the pressure behaviour during boundary dominated flow which provides an estimate of connected pore volume drained by the well. Convolution Explorer software was used to redraw the derivative unit slope to obtain an acceptable match in the unit rate response function (Figure 2). 

Four high frequency build ups were added to the full field reservoir simulation model on the same time frequency as a standard well test e.g. 1s, 2s, 10s, 1 min etc. This high resolution of timesteps was not necessary for this analysis however it was interesting to see how tNavigator would perform. This increased the simulation time from around 2 hours to 6 hours.

The history match of the existing reservoir simulation results (Figure 3) showed an important mismatch in both the flowing and build up pressure data. By applying pressure rate deconvolution, we find out exactly what is wrong with the history match. The res sim model permeability-thickness (kh) is 73% too high, the total skin is +17.6 when it should only be +5.7 and the connected volume is 10% too high.

To correct the reservoir simulation model the above-mentioned parameters required adjustment for P101z. Altering the kh and connected volume was very simple and fast using tNavigator. However, to overcome the challenge of modelling the Total Skin as per the correct value from PTA it was necessary to add local grid refinement (LGR) around the well. The refinement on grid cells increased the running simulation time from 6 hours to 15 hours in tNavigator. This was surprisingly impressive solving 6.1 million cells with 6,570 timesteps in less than a day. As for comparison, the same simulation arrangements on different reservoir simulation software needed a few days to complete the simulation run.

Once the simulation model was updated with the correct parameters it was simulated again and the results are presented in Figure 4. This time the res sim deconvolved derivative overlays the correct deconvolved derivative of P101z. As a result, the data plot is also perfectly matched especially the 2018 PBU. 

Outcome:

The user was successfully able to replicate all of the well/reservoir properties and architecture as described by pressure transient analysis. The simulation run times were significantly reduced in comparison to previous attempts in alternative software’s, even when adding LGR increased the demand of computational power. This made the project feasible and practical, allowing the user to deliver successful results to their client within the tight timeframe. Now that the study has been completed on well P101z and considering the positive results, the same analysis will be extended to the entire Corrib dry gas field.