Experimental Validation of a Direct Fiber Model for Orientation Prediction
Round 1
Reviewer 1 Report
When considering composites, there is always question of the constituents sizes (fibers and matrix), which emphasize the influence of size-effects on rheological behavior... Numerical approach that has to be considered is de facto multi-scale modeling ... Can you comment on this ? More largely, discussion of results should make a special mention of size-effects on both numerical and validation aspects ...
In line with what preceds, In part 4, authors mentioned "To set up a simulation that matches the SPR experiment, a representative number of fibers was placed inside a shear cell..." How can you be sure that the cell considered is really representative of the material ? Can you develop ?
Author Response
Dear Reviewer,
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
This article presents the results of a validation effort for a direct fiber (particle level / microscale) model for fiber orientation evolution. The article is relatively well written and deserves publication, but the comments below must be addressed.
-- Comments
There are quite a few punctuation errors throughout the article that should be addressed. Taking just the abstract:
Line 12 “This work therefore employs” --> “This work, therefore, employs”
Line 13 “In the model each” --> “In the model, each”
Line 20 “periodic boundary cell and a flow” --> “periodic boundary cell, and a flow”
Line 21 “aligned orientation, however, it” --> “aligned orientation; however, it”
Similar errors appear throughout. I suggest a thorough proofread.
Line 20: The statement “a flow field matching the experiment was applied” is slightly misleading as coupling calculations were not considered. The statement should perhaps be refined to say an “idealized flow field” or “boundary conditions matching the experiment were applied.”
Line 83: What is a segment node? Looking at the referred work of Schmid et al., it seems to be a segment midpoint. About which axes is the angular velocity resolved? Perhaps labelling Figure 1 in more detail is appropriate. Alternately, introducing another figure may be appropriate.
Lines 83-90: Some comments on the potential effects of each neglected physics would be useful.
Page 3: There is some imprecision in the typography throughout this page. Sometimes, variables are introduced with their subscripts. Sometimes, they are not. Sometimes, variables are introduced in parenthesis; other times they are not.
Lines 112-116: It would be useful if the A, B, and D variables were italicized to distinguish them from simply the letters A, B, and D.
Line 113: “A und B” --> “A and B”
How do you guarantee no interactions in the initial condition? Is an excluded body force only calculation performed after initialization of the particles in the domain?
Line 168-170: This statement is quite confusing. Consider rephrasing. At no point in the prior discussion had I imagined a molding scenario where weld lines would be significantly encountered.
Line 176-178: Consider reversing the order of these statements.
Line 188: should be “effectively swapping the a_11 and a_22 components of the orientation tensor.”
Line 191: Were the samples CT scanned to account for the initial condition prior to or after this compression? I assume after the compression, but clarification could be useful.
Figure 2 axes are labeled incorrectly. The orientation tensor
components are not global axes… Line 224: What is a representative
number of fibers? Is it the actual expected number of fibers in the volume or something smaller? The fibers in figure 4 seem quite thick compared to the sample thickness (<100 diameters across the thickness --> > 2 micron diameter fibers..) Section 4.2: How was the orientation state sampled to produce the initial state. There are quite a few ways to do this using Fourier recoveries of the orientation distribution function (Advani and Tucker JOR 1987), using natural or exact closures (Montgomery-Smith et al 2011 JFM, Favaloro and Sommer JOR 2020), using entropy (Breuer, Stommel, and Korte JCS 2019).
Figure 9: The strain levels should be indicated on the figure.
Line 326-329: ???
Line 340-341: This is an incorrect statement regarding anisotropic diffusion. Anisotropic diffusion was particularly to limit the A33 term in shearing flows. As the author previously noted, long fibers tend to be orientation near the 1-2 plane (see Favaloro and Tucker 2019 JCOMA). The slow orientation kinetics model of Wang 2008 predates Phelps anisotropic diffusion 2009.
Line 346: More discussion on the effects of coupling would be useful, in microscale models and macroscale models.
Figure 12: curves should be labeled in the figure.
The conclusions slightly contradict the results section in which it was discussed that the results using the injection molding result neither matched the rate or steady state. In fact, it was stated that steady state was not achieved. This must be reconciled.
Author Response
Dear Reviewer,
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
The paper is well presented. I have some questions that could also be useful for future readers.
- How does the proposed microscopic model compare with previous ones, like that used by Fan, X.; Phan-Thien, N.; Zheng, R. A direct simulation of fibre suspensions. J. Non-Newton. Fluid Mech. 1998, 74, 113-135?
- The comparison with the experimental data is performed using only the 1st order fiber orientation tensor. This implies a considerable loss of information both for the numerical and experimental results. Maybe the authors can also investigate higher order orientation information to validate their microscopic modeling approach.
- How can the proposed microscopic model be used to improve current macroscopic Folgar-Tucker-based fiber orientation models? Can it be used to predict or analyze for instance the phenomenological interaction coefficient Ci?
Author Response
Dear Reviewer,
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The article is fine.
Reviewer 2 Report
The edits to the article seem acceptable, though I am not sure why some of the discussion included in the response letter were not incorporated.
The section regarding slow orientation kinetics still misstates the goal of anisotropic rotary diffusion.
The authors noted that they used Fourier recovery / spherical harmonic recovery of the orientation distribution function to initialize the orientation state but did not include that in the article. This is quite critical for the reader to know especially given that the Fourier based recovery is known to give a poor approximation of the ODF in many conditions, including even negative values of probability. While it is certainly an acceptable method, it must be mentioned, and I suggest the authors investigate other methods for the future.
