Production problem with an oil well

Production problem with oil well AI-1

Since production start-up, pressure at this well has reduced much more than expected. There is a bit of concern in the office regarding the long term performance and forecast.

The figure below shows the AI-1 downhole gauge pressure (psi) and oil rate (bbl/d) data versus time.
Oil production history plot
As the pressure keeps decreasing, the well may produce below bubble point, with gas breaking out of solution. The well flowing pressure will also be constrained by the surface facilities and production may fall. The subsurface manager is thinking about an increase in well damage, as this happened in another field nearby. From the geologist, the well could be located in an isolated compartment with a small connected volume and the average reservoir pressure could start to decrease (depletion). What do you think ?

We could look at some “opportunistic” production shut-ins on the log-log plot. Thankfully, we also have a baseline PBU test for reference and comparison, as per best practices.
log-log plot and derivative for the oil well AI-1
The derivative data are consistent, showing a unit-slope straight line and a hump indicative of wellbore storage and skin, a stabilization up to 10hrs, followed by a decline in the derivative trend and another stabilization at late times.
From this plot, the skin doesn’t seem to be too much of an issue. There is no major change in the ∆P plots and as a consequence, no significant increase in skin (damage) is noted between these opportunistic shut-ins.

What is the derivative plot telling us about the well and reservoir ?

This derivative shape could be explained by various reasons: limited perforations / partial penetration, reservoir crossflow (contrast in skin or mobility between layers), multiphase flow in the reservoir, wellbore phase redistribution, linear composite behaviour (increase in mobility and/or storativity away from the well).
Some of these cases are not consistent with the other sources of subsurface information. According to the logs, there is no multilayer description and the full 60 feet section of the reservoir was perforated.
Could the derivative plot highlight the presence of some sealing faults?
Well, we see a negative slope and a lower derivative stabilization at late times, so according to the books, we do not see any sealing fault, do we …?

 

Let’s have a look at Deconvolution. Provided the right operations guidelines and analysis process, we can recover a reliable correct deconvolution response in red:
deconvolution for oil well AI-1
When compared with the conventional derivative in blue, the deconvolution response doesn’t show any derivative downturn. There is only a single derivative stabilization followed by an increase in the derivative. This could highlight the presence of some boundaries, i.e. some sealing faults close to the well. According to the well analysis below, the faults could be located at about 400 feet from the well.
well test interpretation showing the derivative, superposition plot and production history plot
Note the simulated downturn on the conventional derivative. By extending the shut-in duration, we would finally see the boundary response with an increase in the conventional derivative in blue. This response is delayed in time or “distorted” due to the derivative calculation algorithm.
The transient response with the presence of these near wellbore boundaries is the main cause for the decline in pressure. According to the deconvolution response, the reservoir is not closed but still open. As a result, pseudo-steady state regime hasn’t started yet and there is no depletion. These results could then be used to review the production forecast.

A misleading conventional derivative

What we have here is called a “distortion” on the conventional derivative. Sometimes a declining trend in the conventional derivative could be due to the presence of boundaries. In this case, the derivative is misleading. Deconvolution is free of any distortion and provided it is correctly recovered, it becomes the driving tool for the analysis.
This short video shows another distortion on the conventional derivative. More info is available in the basic training courses.

The derivative plot is a very powerful tool to understand the flow geometries around a well, the degree of damage from drilling and completion, the reservoir heterogeneities and boundaries.
But sometimes, the conventional derivative may be a bit tricky to analyse and worse, misleading. There are some other potential problems with the derivative and below are two common issues.
Beware of wellbore phase redistribution and the location of the downhole pressure gauge
The analysis above was made possible thanks to the permanent downhole gauge that was located at about 180 feet TVD from the perforations.

Below is an example from another well with a shallower downhole gauge:
derivative overlay with wellbore phase redistribution
We can see that the derivatives from the different PBU tests are very consistent and this may increase our confidence in the data. However, these responses are not representative and are incorrect !
When running a memory gauge at a deeper position in the tubing, the response in red was acquired:
Derivative overlay with deeper downhole pressure gauge
The deeper pressure gauge doesn’t show the same derivative shape and therefore the same permeability, skin, flow geometries and boundaries. The only difference between the two set of data is the position of the gauge. In fact, the downhole gauge data are affected by wellbore phase redistribution. As the fluid weight below the gauge changes over the shut-in duration, the gauge data are no longer a direct measurement of the pressure at sandface. With shallow gauges, it will be more likely for the data to become useless. More info on wellbore phase redistribution is available in this post and training video.
Multilayer effects
A crossflow in the well or reservoir may mask a particular reservoir feature or create some errors, resulting in wrong skin and permeability values. In addition, the analysis will be highly non-unique. Some extra care and expertise are needed to avoid taking a bad decision during operations or obtaining some false subsurface information.

Recommendations

The PBU tests should always be compared on a single log-log plot, and this is called a “derivative overlay”. It will help to spot some issues and identify the flow regimes and interpretation models. Unfortunately a consistent derivative may not be a guarantee for a representative well and reservoir response. And this is why pressure gauges should be located as close as possible to the perforations.
After a quality check on the pressure and rate data, Deconvolution should be investigated and used as the driving tool for the analysis.