Pressure drop & corefloods for polymer flooding

Thinking out loud (1/2).

Pressure drop & corefloods for polymer flooding. The first time I was confronted to the way polymer flooding was modeled, I was struck by the number of parameters needed and the complexity of the whole thing. It feels like everyone wanted to be sure to capture absolutely everything: from the molecular interactions to the rheological behavior. The first image that came to my mind was a DJ mixing table: a lot of buttons and no idea which to touch to obtain the desired music without the help of an “expert”.

When you think about it in simple terms: what does polymer do? It changes pressure drop.

That’s it. 

Any parameter mentioned in the process is eventually impacting the pressure drop value  in the reservoir. Degradation? Lower pressure drop. Retention? Lower pressure drop along the way. Shear-thinning? Lower pressure drop. Shear-thickening? Higher pressure drop. Higher permeability ? Lower pressure drop.

I mean, this is just what we measure when we run corefloods. We compare the dP of water  to the dP with polymer to get the resistance factor (more pressure drop).

What I never understood is why people were spending so much money on corefloods and using so little of it : “let’s history match what we have obtained and move to upscaling”. What?

But the pressure drop curve contains everything: from retention to molecular interactions and even the dynamics of fingering when water injection is resumed. You can also have rheological behavior by changing the rates. 

And if we assume that the core we are using is representative from the reservoir, it means that each part of the reservoir with these characteristics should display the same dP curve, right? Now, would there be a way to make some extrapolation by taking some extremes? Could we obtain trends by analyzing the wealth of data everyone has generated over the years?

For instance, where is retention in the dP curve? That’s the slope. If you had no retention (and IPV), you should go from RF = 1 to RFmax in a snap. If you have a long core and intermediate pressure taps, then you can “see” how polymer adsorbs along the core.

Now imagine you want to see the impact of brine salinity on polymer retention: could you use the slopes of the dP curves to get trends, then the equation, to finally calibrate with real values?

By the way, the fact that a slope also exists at the tail of the slug when flushing with water shows if further proof was needed that water doesn’t evenly displace the polymer slug.

This kind of approach would be very valuable in all corefloods, with or without oil since both cases will likely be encountered at some point in the reservoir. I know some people who have run more than 2000+ corefloods using Bentheimer rocks: imagine the gold mine that could be exploited with some machine learning… I added a few illustrations to show how much information is contained in this simple graph.

In the end, in our fancy “polymer flood” models, the only thing we do is propagating pressure drops values…

Let it sink.

You can also check our YouTube channel for additional videos and podcasts and navigate the Polymer Flooding Guide for more content or our Academy for training courses.