Can Stanmod be used to model this scenario?
Can Stanmod be used to model this scenario?
I took undisturbed soil cores (30 cm length) from long-term field plots where organic amendments were applied annually for nine years. I used syringe pump (5 cm/day), flow cells, fraction collector, and vacuum chamber (-65 mbar) to leach nutrients and soluble salts and obtain breakthrough curves. I obtained asymmetrical breakthrough curves for nitrate and chloride with peak concentrations before one pore volume. My question is can I use CXTFIT in Standmod to fit the observed and predicted breakthroughs to estimate macopore flow (i.e. mobile water fraction)? If yes, what should the boundary (BVP), initial (IVP), and production (PVP) value problem conditions be? I am unsure of what to use since this scenario is different from a typical pulse application where you apply tracer-free solution followed by pulse of the solute. In my case, the solute is already in the soil from mixing of the amendment with the soil and the solution applied is always tracer-free. I am a new user of Standmod. Thanks for any advice.
Jim, I think it may worth while to try the MIM in CXTFIT for your asymmetrical BTCs. If the solute is initially in the soil, the distribution of the solute should be described as IVP. If the distribution is uniform throughout the soil column, choose constant initial concentration. If not, try to describe the distribution using the multiple stepwise distribution. The rest of the setting should be:
1) Concentration mode: fluxed averaged concentration for effluent concentrations
2) BVP: solute free input water
3) PVP: zero production.
If any further questions, please do not hesitate to ask me or send me your CXTFIT project.
Nobuo
1) Concentration mode: fluxed averaged concentration for effluent concentrations
2) BVP: solute free input water
3) PVP: zero production.
If any further questions, please do not hesitate to ask me or send me your CXTFIT project.
Nobuo
Thanks Nobuo.
We incorporated the organic amendments to a depth of 20 cm but the soil cores are 30 cm in length, so I cannot assume uniform distribution throughout column and cannot use constant initial concentration. Therefore, I tried using the stepwise distribution. I used # of pulses=1. For stepwise initial distribution, I used concentration=0.904 (relative concentration C/Co at start) and start position of 30 (30 cm length of column). I am not sure if the # of pulses, concentration, and start position are correct? I varied the start position and the model seemed to give the same answer. The parameters I used in the model were the following for the test case.
Model type=inverse
Code-nonequil CDE
Input and output data code=time and position dimensionless
Time/space units=dimensionless
Conc mode=flux-averaged conc
Parameters=no constraints/no estimate of total mass
Type non-equil model=2-Region
Parameters-V=4.9 (water applied at 4.9 cm/day), D=2 (fit), R=1 (Fit), B=0.9 (fit), omega=10 (fit), mu1=0, mu2=0
BVP=solute-free
IVP=stepwise, # pulses=1 (?)
Stepwise initial distribution=conc=0.904?/start position=30? (30 cm core)
PVP=0
Inverse data structure=T,C for fixed depth (breakthrough)
Data file=time=pore volumes, concentration=relative concentration (C/Co, I used Cmax for Co)
I obtained a good fit (R2=0.994) with CXTFIT but the mobile water fraction or beta value was very small (0.0012), so I am not sure if model is giving the right answer since peak concentrations before one pore volume suggests preferential flow in macropores? I estimated the mobile water fraction from observed breakthrough curve as number of pore volumes at C/Co=0.5 (Singh and Kanwar 1991, JEQ, 20:295), and this was 0.30. So CXTFIT predicted a beta value of 0.0012 and from the observed BTC the beta value was 0.30?
We incorporated the organic amendments to a depth of 20 cm but the soil cores are 30 cm in length, so I cannot assume uniform distribution throughout column and cannot use constant initial concentration. Therefore, I tried using the stepwise distribution. I used # of pulses=1. For stepwise initial distribution, I used concentration=0.904 (relative concentration C/Co at start) and start position of 30 (30 cm length of column). I am not sure if the # of pulses, concentration, and start position are correct? I varied the start position and the model seemed to give the same answer. The parameters I used in the model were the following for the test case.
Model type=inverse
Code-nonequil CDE
Input and output data code=time and position dimensionless
Time/space units=dimensionless
Conc mode=flux-averaged conc
Parameters=no constraints/no estimate of total mass
Type non-equil model=2-Region
Parameters-V=4.9 (water applied at 4.9 cm/day), D=2 (fit), R=1 (Fit), B=0.9 (fit), omega=10 (fit), mu1=0, mu2=0
BVP=solute-free
IVP=stepwise, # pulses=1 (?)
Stepwise initial distribution=conc=0.904?/start position=30? (30 cm core)
PVP=0
Inverse data structure=T,C for fixed depth (breakthrough)
Data file=time=pore volumes, concentration=relative concentration (C/Co, I used Cmax for Co)
I obtained a good fit (R2=0.994) with CXTFIT but the mobile water fraction or beta value was very small (0.0012), so I am not sure if model is giving the right answer since peak concentrations before one pore volume suggests preferential flow in macropores? I estimated the mobile water fraction from observed breakthrough curve as number of pore volumes at C/Co=0.5 (Singh and Kanwar 1991, JEQ, 20:295), and this was 0.30. So CXTFIT predicted a beta value of 0.0012 and from the observed BTC the beta value was 0.30?
Jim:
According to the CXTFIT manual (p.73), the starting position of the first step Z1 is always 0 (see also Fig.2.2 in p.10). I guess the program automatically replaced the starting position of 30 cm to 0 cm because you chose # of pulse =1. That may be a reason to have the same result regardless of the position value.
Can you try a following setting?
# of pulses =2 with
Conc., Starting position
0.904 , 0
0, 20
If the fitting is good and beta is very small, you may need to try the equilibrium CDE. However, please firstly try the above IVP setting. If any further problems, please let me know.
Nobuo
According to the CXTFIT manual (p.73), the starting position of the first step Z1 is always 0 (see also Fig.2.2 in p.10). I guess the program automatically replaced the starting position of 30 cm to 0 cm because you chose # of pulse =1. That may be a reason to have the same result regardless of the position value.
Can you try a following setting?
# of pulses =2 with
Conc., Starting position
0.904 , 0
0, 20
If the fitting is good and beta is very small, you may need to try the equilibrium CDE. However, please firstly try the above IVP setting. If any further problems, please let me know.
Nobuo