Tracer diffusion
ping chen
Hammond, Glenn E
Thank you for the email. Please send a representative input file to better describe the conceptual model (layout of materials, initial and boundary conditions, etc.). Also, what are the plotted variables and the units on the axes of the plots below?
Glenn
From: pflotra...@googlegroups.com <pflotra...@googlegroups.com>
On Behalf Of ping chen
Sent: Friday, September 30, 2022 8:24 AM
To: pflotran-users <pflotra...@googlegroups.com>
Subject: [pflotran-users: 6813] Tracer diffusion
Check twice before you click! This email originated from outside PNNL.
Hello all
i am trfying to make reactive transport modeling couple diffusion and chemical reactions. firstly i tried diffusion with only one tracer in electrolyte. my diffusion cell is a clay column sandwiched by two filters, so i include two filter in the input file. MATERIAL filter1 and MATERIAL soil1 are assigned to region filter1, region filter2 and region illite. then the fit result of flux become strange, it increase to a high level at short time then back to steady state, rather than non-linearly increase to a steady state. the following is a screenshot of fit flux without and with filter block. could you help me find out the reason. thanks a lot.
--
You received this message because you are subscribed to the Google Groups "pflotran-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to
pflotran-user...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/pflotran-users/a9bbceaf-9840-4392-8874-539ccca48ec7n%40googlegroups.com.
ping chen
You received this message because you are subscribed to a topic in the Google Groups "pflotran-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/pflotran-users/f4bNtG47s1c/unsubscribe.
To unsubscribe from this group and all its topics, send an email to pflotran-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pflotran-users/PH0PR09MB78360A392E24118D8B0E2EA09A569%40PH0PR09MB7836.namprd09.prod.outlook.com.
Hammond, Glenn E
Ping,
When I looked at your observation file (*-obs-0.pft), I saw Tracer concentrations well above the boundary value of ~2.7d-7. For instance, ~1.d-4. This made no sense and I had to think for a bit. I notice that you had modified the tolerances for the reactive transport Newton solver. When I commented out that entire block, the concentrations were more reasonable (well below ~2.7d-7). I suspect that the tolerances are causing a “false convergence” and bad updates. With your tolerances, the code converges on itol_res (infinity tolerance on the residual), which is likely incorrect…I don’t know; I never use that tolerance for reactive transport (or for anything, for that matter). My advice with reactive transport is to always use the default tolerances. When I use the defaults (by commenting out your NEWTON_SOLVER block), the code converges on stol (basically a very small relative update in concentration). I normally say that converging on stol is bad, but your maximum time step size is so small (1.d-3 day) that stol (in that case) is likely fine.
Please try using the default convergence criteria and compare the results.
Glenn
To view this discussion on the web visit https://groups.google.com/d/msgid/pflotran-users/CAN8T_N2qyDGTf6cZhoFy%3DR%3DyzpqQLTauns48q0xJBn_nLzt6pQ%40mail.gmail.com.
ping chen
To view this discussion on the web visit https://groups.google.com/d/msgid/pflotran-users/PH0PR09MB7836401E10FBDEF6AD227C1A9A569%40PH0PR09MB7836.namprd09.prod.outlook.com.
ping chen
To view this discussion on the web visit https://groups.google.com/d/msgid/pflotran-users/PH0PR09MB7836401E10FBDEF6AD227C1A9A569%40PH0PR09MB7836.namprd09.prod.outlook.com.
Hammond, Glenn E
From: pflotra...@googlegroups.com <pflotra...@googlegroups.com>
On Behalf Of ping chen
Sent: Sunday, October 2, 2022 11:52 AM
To: pflotra...@googlegroups.com
Subject: Re: [pflotran-users: 6820] Tracer diffusion
Dear Glenn
Thank you very much. the strange flux problem has been solved.
i have other questions. theoretically, if the concentration gradient is assumed to decrease linearly, the flux should increase non-linearly at first and reach a steady state after specific time. the flux at steady state is proportional to concentration gradient and effective diffusion coefficient. the time it needs should relate to the porosity or the rock capacity factor. the bigger the porosity is, the fast the flux reach steady state. while in this model, only effective diffusion is able to adjust (i have tried different value for porosity and D0, the fitted flux compare to experiment flux is one of it. the input file is tracerHTO_ping), i dont know which variable is able to make the time of flux reaching steady state big or small. could you kindly teach me what might be the solution.
According to the governing equations,
http://doc-dev.pflotran.org/theory_guide/mode_reactive_transport.html#governing-equations
diffusive transport is a function of porosity, saturation, tortuosity and the molecular diffusion coefficient. Increase any of those parameters, and the flux reaches a steady state faster, assuming that the gradient does not change or reverse direction over time.
another question is that is the possible to assign different effective diffusion coefficient to different species. i have tried the AQUEOUS_DIFFUSION_COEFFICIENTS block (tracerTc_ping), while it give error info "Species-dependt diffusion may not be used with aqueous speciation since fluxes are currently implemented based on the total aqueous component concentration and the diffusion of secondary complexes is lumped", how to deal with this problem.
Species-dependent diffusion coefficients are not support when SECONDARY_SPECIES are included in the input file as each species has a different coefficient, but PFLOTRAN transports total component concentrations (a summation of the primary species concentration and its stoichiometric contribution to each of the secondary aqueous species). We would have to split out the flux of each individual species and apply the corresponding diffusion coefficient…, and we do not do that.
Glenn
To view this discussion on the web visit https://groups.google.com/d/msgid/pflotran-users/CAN8T_N0_5VKFXJZSt97EkbYWQSjcm%2BCwsveuH9ber1-nM79boA%40mail.gmail.com.
Hammond, Glenn E
You can use Excel. I prefer Matplotlib and there are example scripts in the short course exercises that demonstrate how to do so.
Glenn
To view this discussion on the web visit https://groups.google.com/d/msgid/pflotran-users/CAN8T_N23a9_2QqwcTNH-nrdcM5vLrXPaByhJqMALF7ZNvvqvQA%40mail.gmail.com.