Automated history matching of unsteady state core flooding data is a common method of relative permeability calculation. In a carbonated water injection process, many pore scale mechanisms exist while a commercial simulator cannot model all of them. when we write Darcy equation for an injection process (macroscopic scale) all mechanisms helping oil to produce are included in relative permeability term (almost except viscosity decrease). can one justify the application of a black oil simulator such as Eclipse E100 for modeling of CWI core flooding experiments? is there any publication in this regard?
Dear Hamid, This is a good idea,Generally, For reverse calculation of relative permeability that's better to use Sendra core simulator, this is a user-friendly and simple software. In case of carbonized water injetion, there are some chemical reactions so you cannot use E100. You should use E300 if you decided to use Eclipse.For further information refer to chapter 10 (CO2/H2S Solution in the Aqueous Phase) of eclipse technical manual.
@mehdisafari Dear mehdi, thank you so much, if you look at the Sendra manual, the behind code is a simple black oil code and the results are not that reliable based on my inspections. I believe coupling eclipse and MATLAB results in more reliable outcome. By asking this question, I am looking for the justification to use E100. also in case of chemical reactions, I have seen published researchers that have been used E100 in order to back calculate rel perm curves for LoSal water injection.
I know, In case of losal, there is not any significant change in porosity, but for CO2 solution, that's different. so that's better to use E300 instead of E100. Any way, when you do injection in a core sample, typically porosity doesn't change significantly.Also, I think to calculate the relative permeabilities you can use simopt insted of coupling Matlab and Eclipse.
I always use MATLAB optimization toolbox but not simopt. woild you please provide any source or reference to read about this method?
Could you please explain how to use Matlab for history matching?I know about the Eclipse-Matlab coupling.But in case of history matching, how do you implement that?
you must write a function which its inputs are your controlling parameters, this function writs the desired data automatically to the data file with respect to inputs then performs the simulation and reads the desired simulation results from PRT file. the output of the function must be the value of objective function (it could be RMSE between simulation and experimental {oil production and pressure difference} results). Then, you must optimize this function by MATLAB optimization toolbox.
Dear Hamid,
I know about the procedure, I wrote a smart interface between Eclipse and Matlab couple of years ago. But this is what I am saying, How do you change relative permeability?Base on a function or randomly?or for example you use a correlation and try to find optimum of that correlation constant?because we have several points in relative permeability tables. Do you have any tested case or Matlab m file related to the issue?I have several m-files, if you want, we can share them with each other.
Regards,
I always use relative permeability models such as Corey, and set the model parameters as tuning parameters. Then, I generate relative permeability values in different saturations and edit the data file based on generated values. I am willing to see your files and procedure and hope to share my files with you. please kindly send me an email. thank you so much.
- Badges
- Science topic
- Similar topics
- Engineering
- Petroleum Engineering