DebrisInterMixing-2.3: a finite volume solver for three-dimensional debris-flow simulations with two calibration parameters – Part 2: Model validation with experiments

von Boetticher, Albrecht; Turowski, Jens M.; McArdell, Brian W.; Rickenmann, Dieter; Hürlimann, Marcel; Scheidl, Christian; Kirchner, James W.

doi:https://doi.org/10.5194/gmd-10-3963-2017

Articles | Volume 10, issue 11

https://doi.org/10.5194/gmd-10-3963-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

https://doi.org/10.5194/gmd-10-3963-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

Articles | Volume 10, issue 11

Model evaluation paper

|

03 Nov 2017

Model evaluation paper |

| 03 Nov 2017

DebrisInterMixing-2.3: a finite volume solver for three-dimensional debris-flow simulations with two calibration parameters – Part 2: Model validation with experiments

Albrecht von Boetticher, Jens M. Turowski, Brian W. McArdell, Dieter Rickenmann, Marcel Hürlimann, Christian Scheidl, and James W. Kirchner

Download

Final revised paper (published on 03 Nov 2017)
Supplement to the final revised paper
Preprint (discussion started on 28 Feb 2017)
Supplement to the preprint
Companion paper

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Review of "DebrisInterMixing-2.3: a finite volume solver for three-dimensional debris-flow simulations with two calibration parameters - Part 2: Model validation", by v. Boetticher et al.', Guillaume Chambon, 21 Apr 2017
RC2: '-----', Anonymous Referee #2, 06 May 2017
AC1: 'Final Authors Response', Albrecht von Boetticher, 21 Aug 2017

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

AR by Albrecht von Boetticher on behalf of the Authors (21 Aug 2017) Author's response Manuscript

ED: Referee Nomination & Report Request started (23 Aug 2017) by Jeffrey Neal

RR by Anonymous Referee #2 (05 Sep 2017)

Suggestions for revision or reasons for rejection

The MS has been improved. Here are some suggestions for further improvements.

P2 -
L4: Confusing! --> The model treats the ambient air and debris phases separately, (?)

L29: allows reliable numerical studies --> allows numerical studies
[would be better this way, because not all flows might be reliably modelled with just 2-parameters].

P3 -
L7: as defined in (Yu et al., 2013). --> as defined in Yu et al. (2013).

L21: full --> large

P4 -
L2-3: Why so, explain.

L6: core --> ad-hoc

L18: gradients and smaller than ten.: Sounds strange, re-formulate?

L31: Math symbols.

P5 -
remove , in eqn.

P6 -
L5: equation 1. --> (1).

L13-14: The simulated flow depths reproduce the laser signal with respect to both time and amplitude (Figure 3) and predicted run-out deposits replicate the water content sensitivity (Figure 4) but underestimate the effect.
--> The simulated flow depths tend to reproduce the laser signal for higher water content with respect to both time and amplitude (Figure 3) and predicted run-out areas replicate the water content sensitivity (Figure 4) but over- and under-estimate the effects.

L16: Figure 3 top --> Figure 3a
[also throughout the text indicate figure panels with a, b, c, etc., and refer in the text accordingly. This will help the reader to match fig. with text].

is accurately --> is more accurately

L19: the flow depth --> the exp flow depth

P8 -
L26: , or even a coupled Lagrangian particle simulation: Remove, or provide Refs.

P10 -
L33-34: For a physically meaningful comparison with SD it would be better also to mention error bars in experiments.

P11 -
L9: a widely applicable modeling framework --> a potentially widely applicable modeling framework

L12: to accurately --> to rather accurately

P12 -
L23: phase separations due to grain-size sorting --> phase separations and grain-size sorting

simulated --> simplified ?

L25: lower curve, upper curve --> inner curvature, outer curvature
[would be better to call this way, see, e.g., Pudasaini et al., 2005].

L33: partial slip --> the slip

P13 -
L32: to be taken into account --> to be taken into account as in Pudasaini and Fisher (2016) with the mechanical separation-fluxes, or

L34: However, the corresponding model extension would introduce new model parameters and higher numerical costs. As a consequence of: Remove.
[Such statements are not helpful from a physical point of view of naturally complex mixture debris flows].

P14 -
L1: our reduced approach without grainsize sorting effects, --> Our reduced approach without phase-separation and grainsize sorting effects,
[bacause, phase-separation and grainsize sorting are not the same, one or both can be present in mixture flows].

L7: grain-size sorting --> phase-separation and grain-size sorting

P15 -
[Some of the statements in L5-20 are not consistent with the complex nature of debris flow, so need to be re-formulatted].

L9: not to gain a perfect representation of the experiment, but: Remove.

P11-12: depend on many free parameters --> depend on several physical parameters

L13-14: Two-phase approaches, on the other hand, involve high numerical costs. --> Multi-phase approaches involve high numerical costs.

L14-16: In the case of two-phase coupling by drag between grain and fluid, the uncertainty in the drag between granular and fluid phases necessitates parameters that are difficult to quantify in case of the non-Newtonian suspension and non-spherical gravel grains. As a consequence, no previous: Remove.
[Applications might need simplifications. But, this is unnecessarily negative to the reality of complex mixture flows as the phases are not locked-up and evolve with interphase momentum exchanges].

L16-17: modeling approach has succeeded in predicting debris-flow behavior across such diverse experimental settings as those examined here,
--> Our modeling approach has succeeded in predicting debris-flow behavior across a diversed experimental settings as those examined here,

P16 -
L12: than a highly parameterized model (perhaps because the latter is over-fitted),
--> than a model with more parameters,
[In physical model the parameters emerge from the natural need and characterize the material and flow properties. But, again, as mentioned earlier, such statements do not help so much].

L13-18: Because such changes ... ... such information: Remove.
[This can in principle be done in any mechanics-based model as these statements basically concern with the initial and boundary conditions and erosion-deposition].
[The two sentences in L18-20 already explain enough about the main essence of the paper, from an application point of view].

L21: due to --> and

References:

P17-18 -
Check references, e.g., Pudasaini, S. --> Pudasaini, S. P., etc.

Figures:
Fig. 3: [also applies to other figures]

- increase the font sizes in label, legend, etc.,
- increase the curve thickness.
- in caption, mention which exp. it corresponds.

Fig. 8, 11:

Put a, b, c, d, ..., in panels and explain each panel clearly.

Fig. 9:

Still confusing the correspondnce between test and simulation.

Hide

RR by Guillaume Chambon (11 Sep 2017)

Suggestions for revision or reasons for rejection

The authors extensively revised their paper to account for referees’ comments. Globally, the different issues raised in the reviews are addressed in a satisfactory manner. In particular, the newly-added parts concerning sensitivity to grid resolution and contribution of gravel rheology are highly relevant, both from a physical point of view and for potential users of the model. I do not necessarily agree with the response concerning the necessity to consider the calibration parameter $\tau_{00}$ dependent on grid resolution. To me, this tuning rather constitutes an artifact related to the impossibility of simulating the considered flows with a sufficient 3D resolution. In an “ideal world”, simulated shear rates should not depend on grid resolution! But since this issue is now explained more carefully in the paper, every reader shall have the possibility to make his own opinion on it. I was also surprised not to find in the revised paper the interesting discussion concerning the role of pressure-dependent viscosity (and the accompanying figure) that the authors mention on p.1 of their response file. Is it an omission?

Provided the authors correct this probable omission, I consider the paper now as almost ready for publication. I recommend nevertheless that the authors try to further improve the structure of the text, notably in the presentation of results, to make the reading easier. I provide below some suggestions in this sense, as well as point out some few issues that would still require clarification.

Detailed comments

*Following the request of one of the reviewers, the authors added numerous remarks throughout the text concerning multiphase models and their advantages/drawbacks with respect to the presented monophasic model. I would rather suggest gathering all these remarks in one single place in the paper, in order to have a more structured and well-identified discussion on this issue (possibly in subsection 4.6).

*p.2, l. 26: Is there a “in” missing before “the interstitial suspension”?

*p.2, l.27-28: Please explain a bit better why the assumption of damped grain collisions justifies a Coulomb-viscoplastic rheology for the gravel phase.

*p.3, l.6-7: The sentence starting with “In case of dense mixtures…” is not very clear. Does it mean that the user-specified value of $\tau_{00}$ is automatically modified by the model in cases of high concentration?

*p.3. The fact that $\tau_{00}$ needs in general to be considered as dependent on grid resolution, due to changes in computed shear rate, could already be indicated at the end of section 2 (together with a reference to the more detailed discussion later in the paper).

*p.4, l.11: “does not fulfill this requirement…” What requirement?

*p.4, section 3.2. To make the text easier to read, I suggest moving the detailed elements relative to the grid resolutions used for the various test cases, to the respective sections describing these test cases. Only the general description of the criteria that contributed to the selection of these grid resolutions, needs to kept here.

*p.4, l.31-32: the x and y directions are not defined.

*p.6, l.4-30. Describing first the case with a reduced water content, then the calibration experiment, and then the case with an increased water content, does not seem optimal. I would suggest instead to first describing in detail the calibration case, and then discussing the differences obtained when water content is decreased or increased.

*p.6-7: The data presented in Table 1 would need to be described with a bit more details in the text. In particular, all what concerns the description of raw results in section 4.2 could probably be moved to 3.1.1.

*p.7, l.13: What the authors mean by “the second group of experiments” is not fully clear.

*Section 3.5: I still find it strange that the authors do not properly describe the results obtained with the SG mixture. Devoting a specific subsection to this case, as it is done for the two other cases, would be more consistent. Some of the descriptions currently presented in 4.5 could be moved to this new subsection. This would also allow the authors to specifically comment Fig. 7 and Fig. 12-e,f, which are currently only briefly mentioned in the text (if at all for the latter).

*Section 3.5.1. Again for the sake of consistency, and to facilitate comparisons with the two other cases, I suggest adding and describing a plot showing the evolution of front position over time for the smooth-SGM case.

*p.11, l.30-31. The sentence starting with “The enhanced underestimation…” is not very clear.

*p.12, 34. I do not think that a multiphase approach would be required to account for the wetting of the walls in the bend experiment. It should be perfectly possible to handle partial-slip boundary conditions in the frame of the present monophasic approach.

*p.13, l.9. The statement concerning the “severe change of channel roughness” should probably be removed here, since the comparison between the SGM and SG mixtures was performed for the same roughness condition (smooth bottom).

*p.14, l.30. The statement “we removed …” is not clear: did you completely remove these cells from the computation, or did you simply exclude them from the computation of the average viscosity?

*Fig. 12. It would probably be clearer to put panels e and f, which do not pertain to the same case as panels a-d, in a devoted figure. In addition the current caption is unclear, as it leads to think that the SG experiment was modeled with the same $\tau_{00}$ value as the rough-SGM experiment (82.8 Pa).

Hide

ED: Publish as is (26 Sep 2017) by Jeffrey Neal

AR by Albrecht von Boetticher on behalf of the Authors (28 Sep 2017)

Short summary

The open-source fluid dynamic solver presented in v. Boetticher et al. (2016) combines a Coulomb viscosplastic rheological model with a Herschel–Bulkley model based on material properties for 3-D debris flow simulations. Here, we validate the solver and illustrate the model sensitivity to water content, channel curvature, content of fine material and channel bed roughness. We simulate both laboratory-scale and large-scale debris-flow experiments, using only one of the two calibration parameters.