Hello there!
We're currently using Plaxis2D with its PaxFlow and Thermal Modules to simulate the behaviour of liquid soil as a filling material for buried power lines. In our case we got field data from a ton of sensors recording the actual soil-behaviour at a research project where liquid soil has been used as a fill material for the trench of a Gas insulated DC powerline. There is two issues that I'm currently having, that result in pretty significant discrepancies between the recorded data and the simulation:
1. My first question: Is there a way to modify the formula that is being used to calculate the soils' heat conductivity? (Page 208 in reference manual "thermal tabsheet")
Plaxis calculates the effective heat conductivity of the soil based on the assumption that it behaves like a porous solid material instead of an agglomeration of spherical soil particles. Therefore the resulting heat-conductivities easily end up in very optimistic regions of +4W/(mK) with regular parameters, plus does not take non-linear effect of saturation into account (especially coarse sand has a highly non linear increase of conductivity in the low saturated-range). This non-linearity results due to the rapid increasing of the contact surface between the spherical soil particles, as soon as the water content rises (since the adhesive waterfilms inevitably form "water-bridges" between the particles).
My solution so far was to radically change the input parameters being conductivity of the solid, water and vapour to fit the curve to the actual saturation-based conductivity. This resulted in a heat conductivity of 7 W/(mK) for the water being 10 times higher that originally supposed to be, while the solid conductivity ends up at around 0.8W/(mK). With these modifications we get realistic values of 0.5W/(mK) for dry sand and 1.5W/(mK) for Sr=0,3.
2. My second question, which results due to the focus on thermal and groundwater flow (flow-only-calculation type was chosen):
How is the temperature of infiltrating water calculated or is it just set to the temperature at the "entry-point" (for instance ground level for precipitation).
Since we recorded the precipitation in the field experiment, i implemented the precipitation as a table into the simulation. However, i could not find information on how the temperature of the infiltrating water is calculated or whether it's calculated and not just set to the ThermalBCs Temperature at the ground level (in the case of precipitation). I'm asking because our soil has a high permeability and the recorded sensordata states high heat-transport via infiltrating rainwater.
Thanks alot in advance! I hope someone can help us even though it's kind of an unusual "problem".
King regards
Louis
Dear Louis,
Thank you for providing all the details of your project. Before giving you our feedback, I would like to mention that this interesting use case you have given requires some extra communication that can be easily done via email and our service request.
Can you please submit a follow-up service request for us? Then we can pick it up and continue the discussion on this topic. To do so, please follow the link: https://apps.bentley.com/srmanager/ProductSupport
To provide information to your questions, please see below:
1. You are correct that currently, it is not possible to adjust the formula for the thermal conductivity. This is something we have recognized ourselves and we are busy with considering some added functionality to improve this behaviour as it may differ from project to project. Please follow-up with us in order to get more information about it.
2. In PLAXIS we do not separate water and soil temperatures, and infiltrating water is assumed to have ground surface temperature. Thus, appropriate temperature BCs should be assigned to the ground surface to properly define this. Please follow-up with us in order to provide further assistance if needed.
Answer Verified By: Louis Zrenner
Even though I'm kind of late: Thank you for your help, the results turned out to match the recorded data very well (given the fact that we - due to lack of time and geographical distance to field experiment - couldn't determine the input parameters for waterflow and heatflow to a highly-accurate degree). However there is an improvement and enhancement that I'd like to include:
Taking the temperatur coefficient of the conductor resistance into account
The heat dissipation of the conductor depends on the ohmic loss, which again depends on the current and temperature of the conductor. A higher temperature of the very-same conductor (with equal current and material) results in an increased power-loss (ideally equal to the dissipated heat energy). In our model we just assigned the most unfavourable heat dissipation as a constant heat flux density (refers to the heat dissipation at 80°C, max. recorded temperature). However I'd like to take time-dependent temperatures into account. One way to do this would be to create weekly or monthly-based stages. That way you could adjust the dissipated heat stage-by-stage by calculating it based on the "end of previous stage" conductor-temperature. However it's is more of an approximation and rather inconvenient due to the manual adjusting and lack of automatic calculation. Is there a more convenient way to deal with this issue, for instance by using variables?
Thank you for your feedback.
I am assuming you refer to a heat flux q (applied on boundaries like an outflow flux) to simulate heat loss from the conductor, and the request is to have a flux that is temperature-dependent q=q(T) (instead of q=const).
If that is the case, then, indeed, the only workaround is the one you describe, i.e. approximation by dividing the calculation into many phases and applying the flux dependent on T at the end of the previous phase.
Your request is, of course, valid and we will consider it in a larger scope of improvement/implementation for our thermal feature as would like to offer more flexibility and control for the user. I will add your comments to our research team.