Why fine-grid flow simulations?
The modern geological models are normally built in fine-grid (20 m – 50 m grid size or less). Conventional full-field flow simulations are usually performed over an upscaled grid (100 m – 300 m) in order to save on computation time and average the calculated pressure/saturation to tame the uncertainties in well-by-well data. Upscaling is a reasonable procedure in single-layer homogeneous reservoirs. But if the reservoir has an uneven permeability profile or made up of multiple formation units the upscale procedure can seriously damage the actual picture of fluid propagation, often leading to confusion.
The picture below shows that an upscaled water propagation scenario can miss the moment when water is actually reaching the producer. It happens when the upscaled model blurs permeability profile and simulated flow extends to the layers which in reality do not flow.
Obviously, fine-grid flow simulations have no practical value until the flow scenario is calibrated to the downhole measurements of the actual reservoir flux profile.
Why matching the flux?
Reason #1 – Complicated reservoir flow profile
The actual reservoir flow rate is a function of formation permeability, formation pressure and wellbore skin-factor, and all the above may vary across the production/injection zone. Unlike permeability and pressure, there is no reliably measured data on skin variation along hole. Without flux calibration the simulation software will be moving fluid through all formation units following the permeability and pressure profile. In reality, not all formation units are taking water or producing fluid as scheduled by the permeability model.
The picture below shows a case when non-calibrated model (left) does not see the water coming through the lower part of formation (right) and will be struggling to explain the early water breakthrough in offset producer.
There are several reasons why water goes to the lower permeability formations. But as soon as the actual reservoir flow profile is measured through with downhole tools these data should be used to calibrate the 3D model, which will obviously lead to more reliable production forecasts.
Reason #2 – Behind-casing communication
In many practical cases the actual injection water channels above or/and below perforations making reservoir flux profile very different from the one specified in the input schedule of 3D model.
The picture below shows that most of injection water channels up behind casing to the lower permeability reservoir. The further study revealed this was caused by intense depletion of the upper reservoir in this area.
Very similar complicacies arise in producers: fluid (oil/gas/water) often propagates through non-perforated formation unit and once approaching borehole it channels up/down towards the perforations. The picture below shows a thief water production from a lower formation.
The above cases create a lot of problems in history matching and provoke simulation engineers to overplay with “localized” permeability and “non-neighbour connections” thus taking dynamic model even further away from reality.
How to measure the flux?
The actual reservoir flux profile can be assessed through the spectral noise log and wellbore temperature modelling of “flowing against shut-in” response.
Spectral Noise logging technique localizes the flow units behind the casing (even if tool is conveyed through the tubing). But it needs temperature modelling to evaluate the flow rate in each unit as noise amplitude is not always a fair measure of flow rate. The flow rate evaluation technique is based on numerical simulation of borehole temperature profile under stabilized flowing conditions (several hours or more) and stabilized shut-in conditions (usually 4-5 days).
The picture below shows that injection temperature is not clearly responsive to location of behind-casing injection streaks and shut-in temperature is smeared across perforations due to developed heat transfer and it needs a spectral noise log to see the actual location of flow units.
Anyway, the noise-temperature combination shows that injection is equally shared between two narrow formation streaks (rightmost panel).
The combination of these measurements makes up an essential part of MGFM.
Why PLT is not enough?
Conventional spinner-based PLT survey measures production/injection profile inside borehole and cannot account for redistribution of the flow behind the casing and can’t be used in 3D calibration procedure.
Case #1 – Behind-casing redistribution
The picture below shows that only one formation unit is actually dispatching the flow through all perforations:
Even if injection/production stays within the perforated interval, in many practical cases PLT is the source of confusion and misunderstanding of reservoir behaviour
Cement fracturing and wash-outs create a buffer zone where flow from the bottom formation A4 dispatches fluid across all perforations producing an illusion that all formations A2, A3 and A4 are contributing to production.
Case #2 – Channelling
The picture below shows a thief water production from a lower formation.
PLT profile shows oil-water production from perforations and does not realise that reservoir across perforation produces oil only and all water is channelling from the lower formation.
Residual saturation is a function of flux and relative permeability of rocks.
Once the flux is matched one can adjust relative permeability to match the saturation logs.
The picture below shows that injection water is halfway towards producer while saturation log shows an early water breakthrough in lower formation.
The flux measurements and calibrations confirm that inejction water is equally shared between two formation units. This leads to the stage when adjustment to relative permeabilities is required.
The picture below shows that relative water permeability has been modified and resulted in good match with saturation measurements.
It is important to note that MGFM uses saturation logs only to check for formation layers which are invaded by water and discards the actual value of saturation because the latter is often inaccurate.