the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 04 Oct 2022
CompLaB v1.0: a scalable pore-scale model for flow, biogeochemistry, microbial metabolism, and biofilm dynamics
Heewon Jung
Hyun-Seob Song
Abstract. Microbial activity and chemical reactions in porous media depend on the local conditions at the pore scale and can involve complex feedback with fluid flow and mass transport. We present a modeling framework that quantitatively accounts for the interactions between the bio(geo)chemical and physical processes, and that can integrate genome-scale microbial metabolic information into a dynamically changing, spatially explicit representation of environmental conditions. The model couples a Lattice-Boltzmann implementation of Navier-Stokes (flow) and advection-diffusion-reaction (mass conservation) equations. Reaction formulations can include both kinetic rate expressions and flux balance analyses, thereby integrating reactive transport modeling and systems biology. We also show that the use of surrogate models such as neural network representations of in silico cell models can speed up computations significantly, facilitating applications to complex environmental systems. Parallelization enables simulations that resolve heterogeneity at multiple scales, and a cellular automata module provides additional capabilities to simulate biofilm dynamics. The code thus constitutes a platform suitable for a range of environmental, engineering and – potentially – medical applications, in particular ones that involve the simulation of microbial dynamics.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(2044 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2044 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Heewon Jung et al.
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1016', Anonymous Referee #1, 09 Nov 2022
Review of egusphere-2022-1016
The manuscript “CompLaB v1.0: a scalable pore scale model for flow, biogeochemistry, microbial metabolism, and biofilm dynamics” by Jung et al. (egusphere-2022-1016) introduces a modular numerical model approach for reactive transport and microbial growth in fully water saturated pore structures. The model can implement/consider high resolution scans of porous media as well as microbial metabolic reaction networks from databases, two growing sources of information on subsurface environment and the processes therein. Model linking this information to reactive transport simulations are still scarce which makes the presented model a potentially very useful tool. The manuscript is well written and introduces the model and its different features. Results on the accuracy and on the model performance are shown.
I suggest publication of this manuscript after some moderate revisions. In addition to my comments below, these revisions should clarify which parts of the model have been introduced and verified before and which parts are new and need to be verified in the manuscript (if not done already). I have no worries regarding the technical accuracy of the model but more information on this would be good. This would then also allow determining where the presented model is more advanced than previous models. I also think it would help to show more results of the presented simulation examples (in the manuscript or in some supplement) to a) demonstrate the model performance and b) to allow putting the discussed results in a better context.
Specific comments:
Introduction: I am missing a bit some statements on what exists already for modeling reactive transport and microbial processes at the pore scale There are several rather recent reviews on this (e.g., König et al., 2020, doi: 10.3389/fevo.2020.00053; Golparvar et al., 2021, DOI: 10.1002/vzj2.20087; Pot et al., 2022, DOI: 10.1111/ejss.13142).
L 79: Clarify if “based on the LB method implemented in Jung and Meile” means you are using the previously established code and implement the new features or you have code new flow and transport modules based on the same LB concept. This determines which parts of the model need to be verified in this manuscript and for which parts a verification is given already in previous publications.
L 120: How is this combination achieved?
L 124: Clarify if the considered microbial dynamics are limited to specific example processes (and their kinetic expressions) or if any arbitrary (user defined) processes/rate expressions can be used.
Section 3.2.1: Related to my comment above, in case any arbitrary set of rate expressions can be considered the approach is a) not limited to microbially controlled reactions and b) would conceptually not make a difference between the concentration of biomass and chemical compounds. It is thus not clear to me why there is a distinction between these concentrations at this stage. At the end Eq. 5 is just a specific version of Eq. 6 in case of R is given as gamma*B.
L 148-162: Is there any specific reason why these variables must have the given units?
L 213-220: Were the imposed initial and boundary conditions comparable to the reference models and the experiments?
L 228-229: How is it shown that the composition of the bacteria is stable? Fig. 3 shows that after 48 all approaches exhibit approximately the same composition but not that this composition will not change later on.
L 251-253: How well could the results of the FBA simulations be fitted by a kinetic approach using a Michaelis-Menten consumption rates with the parameters given here and a constant growth yield fitted to the FBA results?
L 273: To which length scale does the Peclet number refer to?
L 312: What was the time step size?
Fig. 5 and associated text passages: It would have been interesting to see how the “traditional” KNS approach without the CA performs compared to the ANN approach. This would also show how much additional computation time the CA requires (besides the larger time for the flow (and transport?) simulation. Since most of the shown examples consider steady state conditions for the flow field but transient conditions for the transport I am wondering why only the computation time for flow and transport together is shown.
L 323-324: Following my comment above: Is this shown somewhere?
L 328 and Figure A1: Clarify that this a relative error for the biomass/metabolites in the system.
L 329-331: In this context it would be interesting to know how much computational effort was needed to run the Monte Carlo simulations for training the ANN model. I am aware that quantifying the training effort is not straight forward but some words on this would be helpful.
L 340-343: I do not get this argument. If I assume that each methods invest a given time for computation in the biomass cells and an and a given time for computation (wasted) in the non-biomass cells why should they differ in their scaling behavior? Furthermore, since detachment and attachment of the biomass is considered there would be biomass also ending on initially uninhabited surfaces. Perhaps it would be good to actually show some results of these simulations and not only the computation times. Similarly to my comment above: How is the scalability of the KNS approach without CA?
L 370: “Soil” might not be the best key word here since soils are typically only partially water saturated while the presented code considers fully saturated conditions. Better to use e.g. “porous media”.
Citation: https://doi.org/10.5194/egusphere-2022-1016-RC1 -
RC2: 'Comment on egusphere-2022-1016', Maria De La Fuente Ruiz, 19 Dec 2022
This paper presents a novel modeling platform CompLaB for simulating 2D pore-scale reactive transport processes while accounting for microbial metabolism and biofilm dynamics. The manuscript is well-written and offers a fair description of the mathematical framework of the model. However, it lacks fundamental information on the physical processes that are modeled. As well as critical conceptual information, such as, which are the species, phases, and reactions considered in the formulation, how are the model parameters defined and which are their effects on the model performance and outcomes, which specific scientific problem/s are willing to be addressed, or which temporal and spatial scale the model can be extended to. Hence, I highly encourage the authors to add a section for a conceptual description of the model (see for instance Section 3 at https://doi.org/10.1029/2011JB008290 or Section 2 at https://doi.org/10.3390/en12112178). In addition, although is stated in the text (e.g. paragraph 370, “CompLaB facilitates simulating dynamic flux balance analysis capturing the microbial feedback on flow and transport in porous media”) the authors do not seem to explore/present here such feedback, which to me would be one of the strongest points of the model.
Finally, please find below a few comments/suggestions that the authors should address to clarify and improve the manuscript:
Figure 1: What do pore geometry changes stand for? Do you mean pore-clogging by biomass growth? That needs a bit more development within the text, especially because the term porosity does not appear at all or is not included in the few equations presented. Also, why are the boundary conditions only applied at the end of the simulation?
What is the (gdw) unit used in paragraph 150 and thereafter
According to paragraph 155, the time step is measured in hours. How quick are the processes modeled here? How is this time step chosen? And, is there any time-step adaptation process to avoid running out or exceeding biomass concentration within a biomass cell (so that mass conservation is reached)?
Paragraph 175: the excess biomass is redistributed here to a randomly selected neighboring grid cell. However, shouldn’t that depend on how much concentration of Biomass there is on the neighboring cells, or perhaps influenced by solid diffusion through the biofilm?
Paragraph 175: How is Bmax defined? And is its value consistent with changes in the pore space geometry along the simulation?
Figure 2: Is fluid flow in pores used to simulate biomass transport in the porous media? or are they considered immobile when the pore is not fully clogged? (check https://doi.org/10.1016/j.csr.2015.04.022)
Figure 2: How can the biofilm in excess be located in a cell where the flow of metabolites is not allowed? As far as I understand dissolved species are only transported through pores, right, and not within biofilm cells? (see general comment for conceptual clarification)
Why in Eq. 9 and thereafter there is no correction for sediment porosity (at least for the non-solid species)? How is the flow field “u” estimated/calculated here?
Paragraph 290: Is organic matter mineralization accounted for changes in available pore space?
Paragraph 290: How is the solid (particle surface characteristics) accounted for describing biofilm attachment/detachment? Which factors are actually controlling this process here?
Paragraph 310: Why is the flow field only updated every 10 timesteps for CA?
Paragraph 330: Please specify what heterogeneous porous media means in this context. How is heterogeneity influencing the model outcomes?
Paragraph 350: Why is the model not accounting for biomass concentration limitations on the diffusion (D) and flow field (u) values?
Citation: https://doi.org/10.5194/egusphere-2022-1016-RC2 - AC1: 'Comment on egusphere-2022-1016', Christof Meile, 14 Jan 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1016', Anonymous Referee #1, 09 Nov 2022
Review of egusphere-2022-1016
The manuscript “CompLaB v1.0: a scalable pore scale model for flow, biogeochemistry, microbial metabolism, and biofilm dynamics” by Jung et al. (egusphere-2022-1016) introduces a modular numerical model approach for reactive transport and microbial growth in fully water saturated pore structures. The model can implement/consider high resolution scans of porous media as well as microbial metabolic reaction networks from databases, two growing sources of information on subsurface environment and the processes therein. Model linking this information to reactive transport simulations are still scarce which makes the presented model a potentially very useful tool. The manuscript is well written and introduces the model and its different features. Results on the accuracy and on the model performance are shown.
I suggest publication of this manuscript after some moderate revisions. In addition to my comments below, these revisions should clarify which parts of the model have been introduced and verified before and which parts are new and need to be verified in the manuscript (if not done already). I have no worries regarding the technical accuracy of the model but more information on this would be good. This would then also allow determining where the presented model is more advanced than previous models. I also think it would help to show more results of the presented simulation examples (in the manuscript or in some supplement) to a) demonstrate the model performance and b) to allow putting the discussed results in a better context.
Specific comments:
Introduction: I am missing a bit some statements on what exists already for modeling reactive transport and microbial processes at the pore scale There are several rather recent reviews on this (e.g., König et al., 2020, doi: 10.3389/fevo.2020.00053; Golparvar et al., 2021, DOI: 10.1002/vzj2.20087; Pot et al., 2022, DOI: 10.1111/ejss.13142).
L 79: Clarify if “based on the LB method implemented in Jung and Meile” means you are using the previously established code and implement the new features or you have code new flow and transport modules based on the same LB concept. This determines which parts of the model need to be verified in this manuscript and for which parts a verification is given already in previous publications.
L 120: How is this combination achieved?
L 124: Clarify if the considered microbial dynamics are limited to specific example processes (and their kinetic expressions) or if any arbitrary (user defined) processes/rate expressions can be used.
Section 3.2.1: Related to my comment above, in case any arbitrary set of rate expressions can be considered the approach is a) not limited to microbially controlled reactions and b) would conceptually not make a difference between the concentration of biomass and chemical compounds. It is thus not clear to me why there is a distinction between these concentrations at this stage. At the end Eq. 5 is just a specific version of Eq. 6 in case of R is given as gamma*B.
L 148-162: Is there any specific reason why these variables must have the given units?
L 213-220: Were the imposed initial and boundary conditions comparable to the reference models and the experiments?
L 228-229: How is it shown that the composition of the bacteria is stable? Fig. 3 shows that after 48 all approaches exhibit approximately the same composition but not that this composition will not change later on.
L 251-253: How well could the results of the FBA simulations be fitted by a kinetic approach using a Michaelis-Menten consumption rates with the parameters given here and a constant growth yield fitted to the FBA results?
L 273: To which length scale does the Peclet number refer to?
L 312: What was the time step size?
Fig. 5 and associated text passages: It would have been interesting to see how the “traditional” KNS approach without the CA performs compared to the ANN approach. This would also show how much additional computation time the CA requires (besides the larger time for the flow (and transport?) simulation. Since most of the shown examples consider steady state conditions for the flow field but transient conditions for the transport I am wondering why only the computation time for flow and transport together is shown.
L 323-324: Following my comment above: Is this shown somewhere?
L 328 and Figure A1: Clarify that this a relative error for the biomass/metabolites in the system.
L 329-331: In this context it would be interesting to know how much computational effort was needed to run the Monte Carlo simulations for training the ANN model. I am aware that quantifying the training effort is not straight forward but some words on this would be helpful.
L 340-343: I do not get this argument. If I assume that each methods invest a given time for computation in the biomass cells and an and a given time for computation (wasted) in the non-biomass cells why should they differ in their scaling behavior? Furthermore, since detachment and attachment of the biomass is considered there would be biomass also ending on initially uninhabited surfaces. Perhaps it would be good to actually show some results of these simulations and not only the computation times. Similarly to my comment above: How is the scalability of the KNS approach without CA?
L 370: “Soil” might not be the best key word here since soils are typically only partially water saturated while the presented code considers fully saturated conditions. Better to use e.g. “porous media”.
Citation: https://doi.org/10.5194/egusphere-2022-1016-RC1 -
RC2: 'Comment on egusphere-2022-1016', Maria De La Fuente Ruiz, 19 Dec 2022
This paper presents a novel modeling platform CompLaB for simulating 2D pore-scale reactive transport processes while accounting for microbial metabolism and biofilm dynamics. The manuscript is well-written and offers a fair description of the mathematical framework of the model. However, it lacks fundamental information on the physical processes that are modeled. As well as critical conceptual information, such as, which are the species, phases, and reactions considered in the formulation, how are the model parameters defined and which are their effects on the model performance and outcomes, which specific scientific problem/s are willing to be addressed, or which temporal and spatial scale the model can be extended to. Hence, I highly encourage the authors to add a section for a conceptual description of the model (see for instance Section 3 at https://doi.org/10.1029/2011JB008290 or Section 2 at https://doi.org/10.3390/en12112178). In addition, although is stated in the text (e.g. paragraph 370, “CompLaB facilitates simulating dynamic flux balance analysis capturing the microbial feedback on flow and transport in porous media”) the authors do not seem to explore/present here such feedback, which to me would be one of the strongest points of the model.
Finally, please find below a few comments/suggestions that the authors should address to clarify and improve the manuscript:
Figure 1: What do pore geometry changes stand for? Do you mean pore-clogging by biomass growth? That needs a bit more development within the text, especially because the term porosity does not appear at all or is not included in the few equations presented. Also, why are the boundary conditions only applied at the end of the simulation?
What is the (gdw) unit used in paragraph 150 and thereafter
According to paragraph 155, the time step is measured in hours. How quick are the processes modeled here? How is this time step chosen? And, is there any time-step adaptation process to avoid running out or exceeding biomass concentration within a biomass cell (so that mass conservation is reached)?
Paragraph 175: the excess biomass is redistributed here to a randomly selected neighboring grid cell. However, shouldn’t that depend on how much concentration of Biomass there is on the neighboring cells, or perhaps influenced by solid diffusion through the biofilm?
Paragraph 175: How is Bmax defined? And is its value consistent with changes in the pore space geometry along the simulation?
Figure 2: Is fluid flow in pores used to simulate biomass transport in the porous media? or are they considered immobile when the pore is not fully clogged? (check https://doi.org/10.1016/j.csr.2015.04.022)
Figure 2: How can the biofilm in excess be located in a cell where the flow of metabolites is not allowed? As far as I understand dissolved species are only transported through pores, right, and not within biofilm cells? (see general comment for conceptual clarification)
Why in Eq. 9 and thereafter there is no correction for sediment porosity (at least for the non-solid species)? How is the flow field “u” estimated/calculated here?
Paragraph 290: Is organic matter mineralization accounted for changes in available pore space?
Paragraph 290: How is the solid (particle surface characteristics) accounted for describing biofilm attachment/detachment? Which factors are actually controlling this process here?
Paragraph 310: Why is the flow field only updated every 10 timesteps for CA?
Paragraph 330: Please specify what heterogeneous porous media means in this context. How is heterogeneity influencing the model outcomes?
Paragraph 350: Why is the model not accounting for biomass concentration limitations on the diffusion (D) and flow field (u) values?
Citation: https://doi.org/10.5194/egusphere-2022-1016-RC2 - AC1: 'Comment on egusphere-2022-1016', Christof Meile, 14 Jan 2023
Peer review completion
Journal article(s) based on this preprint
Heewon Jung et al.
Model code and software
CompLaB v1.0 Heewon Jung and Christof Meile https://doi.org/10.5281/zenodo.7095756
Heewon Jung et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 385 | 117 | 14 | 516 | 3 | 1 |
- HTML: 385
- PDF: 117
- XML: 14
- Total: 516
- BibTeX: 3
- EndNote: 1
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2044 KB) - Metadata XML