Timothy Dai: Theoretical Study of Mixing and Fluid Miscibility in Microchannels


Timothy Dai, Negar Nazari, Luiz Eduardo Bittencourt Sampaio, Anthony Kovscek


The increasing demand for carbon capture and storage as a means to address escalating atmospheric carbon dioxide levels prompts new studies to understand miscible and immiscible fluid interactions in the subsurface. As anthropogenic CO2 emissions are stored in saline underground formations or repurposed for hydraulic fracturing and enhanced oil recovery applications, supercritical CO2 and other subsurface fluids interact, convect, and diffuse through microscopic, connected pores, rock fissures, and fractures. A better understanding of fluid mixing and miscibility behaviors on the microscale thereby assists in predicting and controlling underground fluid injection as well as improves the rate of success and efficiency of such processes.

This project investigates fluid interactions and scrutinizes the development of microscopic mixing as fluid flows through microchannels. Computational fluid dynamics are used to study such phenomena and explore possible relationships between the rate of injection and the development of miscibility. Additional parameter testing include viscosity, the presence of obstacles, and the pattern of obstacles. The results of this study are used to construct a quantitative model to explain the miscibility behavior between different fluids.

For the purpose of this study, a micro-capillary loop (500 µm wide by 30 µm deep by 1.18 m long) is reproduced and used as input to a fluid dynamics simulator, OpenFOAM. Simulations are designed incorporating two liquids of different colors, injected through Y-shaped inlets allowing them to flow in parallel through the entire length of the channel as they visibly mix due to diffusion. The velocities are chosen so that a small Reynolds number (< 1), comparable to those of porous media, is maintained. Each case is simulated several times with progressively tighter levels of mesh refinement and tolerance to ensure accuracy. The phase value alpha (0-1, where 0.5 indicates a completely mixed area) is recorded at incremental distances from the inlet. Exponential curves are fitted through the simulation results and are used to predict the relationship between each test result and the rate of mixing.

10 Comments on “Timothy Dai: Theoretical Study of Mixing and Fluid Miscibility in Microchannels

  1. Tim – You did a very nice job of guiding me through your poster. You made it all seem so simple, yet I’m sure it wasn’t that easy. Can you tell me more about how you think you (or someone else) can use the results? What does it mean for injecting CO2? Is it better to have the fluids mix quickly?

    What do the island represent in reality? Is it ike the complicated pore space of a sandstone?
    Congratulations. I hope you get to do the experiment in the lab some day.
    – Jenny

    • Hi Jenny, definitely! This project doesn’t seem to have immediate, practical benefits but understanding the way fluids mix on the microscale is important to better control the underground injection of gases in enhanced oil recovery or underground carbon storage— emphasis on control, because in some cases, we want fluids to mix quickly and in other cases we want them to mix slower. And I think your real-life equivalent of the islands is spot-on.

      • Hi timdai,

        Great work! It seems similar concept can be used in several fields in many ways. I was just curious about, what would be the possibile effect on net mixing time if you vary the wettability of the channel.

        • That’s a really interesting question, Sunil! The concept of wettability is still quite new to me, but I can surmise that increasing the wettability of the channel would decrease mixing rate, because increased wettability of the walls of the channel would promote more solid-liquid interaction while detracting from liquid-liquid interaction. That is my current hypothesis, and it can be tested using the simulator/lab experiments!

  2. Hi Timothy,
    Congratulations for your work. Could you talk a little bit more about the process of refining the mesh and defining the tolerance for your simulation? Were the final results very sensitive to those parameters?

    • Hi Livia, sure! For each case, we started with 500-1000 cells and a tolerance on the order of E-09, and for each refinement, we increased the cells by a factor of ~4 and decreased the tolerance by a factor of 10 until the change in the final outcome was negligible (<1/2% change). And to answer your second question, not really. After about 1-2 refinements, it would be very hard to distinguish solutions visually, but historical papers have continued to refine until a <1/2% change, so we stuck with that threshold for this project too.

  3. Hi Tim,

    Great poster! I really liked the design, it looks so crispy and professional, well done. Your video presentation was also really well done, you did a great job walking us through your work, and presenting it in a easy to understand way.

    I wonder if your results showed what you expected, or if there were any surprising findings? You also mentioned that part of the future work for this project is testing it in a lab setting, will the ultimate goal be testing it in the field? You mentioned a few places that this work can be applied, such as in the energy or health sectors, where (slash within what sector) is the ultimate goal for this work, and what types of testing might need to be done to get it ready for that?

    Thanks again! Great job!

    • Hi Bianca, it’s definitely a possibility! Some past papers I’ve encountered did similar field studies, but that requires a lot of extra equipment and less control in the parameters we can hold constant/change. The ultimate goal for this work, I would say, is the energy sector, and further testing with more realistic parameter values would be a reliable next step to get ready for applying this project to the energy sector.

  4. Hey Timothy,

    This is some really cool work! Do you or the lab that you worked with plan to follow up with the real-world experimental setup that you outlined at the end? Also, I understand that the coiled microchannel was a practical decision, but is it accurate to the real-life situations that your modeling is relevant to? Along the same line, do you think that having the fluids travel town a continual curve, rather than a more random structure, has any impact on miscibility?
    Again, great job!

    – Mitchell

    • Hi Mitchell, yes! Following up this project with experimental results is definitely the next step, and having theoretical + experimental results combined is traditionally what I’ve seen in past papers on similar topics. As for the coiled microchannel, we actually ran some tests with an entirely horizontal channel and compared it with our coiled design. The results from the two trials were non-negligibly different, but we stuck with the coiled design because, as you said, it could be more easily reproduced in a lab setting. For a similar reason, I would suspect a random structure to definitely have an impact on miscibility, but I think having trials with/without islands gave us a little peak into what would happen with more randomness/less smoothness.

Leave a Reply

Your email address will not be published. Required fields are marked *