Experimental Investigation and Modelling of the Droplet Size in a DN300 Stirred Vessel at High Disperse Phase Content Using a Telecentric Shadowgraphic Probe
Round 1
Reviewer 1 Report
This is a very interesting paper dealing with modeling of the particle size distibution during mixing using a novel, innovative approach. In my opinion, the paper has some minor issues which need to be addressed:
Abstract - L16-17: The sentences state that droplet sizes were extracted and size distributions were determined by a convolutional neural network. Maybe the two sentences could be combined together into one? (Just a sentence style regard, not mandatory).
P1, L30-35: Please rewrite the whole paragraph. It is completely unclear, there are words which are used in multiple places, the verbs are in the wrong places in the sentences and are grammatically incorrect.
P2, L 87: "especially at which phase fraction measured" - there seems to be a word missing in this sentence. Please correct.
P8, L 300 - 301: when calculating the volume distribution from the number distribution, why werent the shape factors taken into account? Excluding the shape factors can lead to errors.
P7, L251-253: Why didn't you use design of experiments (e.g. 3 factors at 3 or more levels) ? The experiment you conducted is perfect for the application of any kind of DOE software. It would have reduced the number of experiments (10 experiments for each impeller speed is a lot) and time required to perform the experiments, plus, you would have gotten direct ANOVA analysis and optimisation of the whole procedure via several software clicks. Or maybe you wanted to collect as much data as possible because of the CNN models?
P7, L271: Model fitting and CNN were both done in MatLab?
P9, L314-315: Please provide a reference for this claim.
P11, L360-390: Do the models show larger prediction errors for smaller Sauter diameters or bigger? Also, are the observed/predicted differences bigger for smaller impeller speeds or for larger ones? Please expand this part of the discussion section.
Author Response
Dear Reviewer,
Thank you for your helpful comments. We considered them and changed the document as following:
Point 1: Abstract - L16-17: The sentences state that droplet sizes were extracted and size distributions were determined by a convolutional neural network. Maybe the two sentences could be combined together into one? (Just a sentence style regard, not mandatory).
We merged both sentences to one
“High-resolution droplet size distributions are extracted from the images using a convolutional neural network for image-analysis in order to investigate the influence of impeller speed and phase fraction (up to 50 vol.‑%).”
Point 2: P1, L30-35: Please rewrite the whole paragraph. It is completely unclear, there are words which are used in multiple places, the verbs are in the wrong places in the sentences and are grammatically incorrect.
We rewrote the paragraph to:
“Stirred batch vessels are commonly used apparats in process engineering applications for tasks like polymerization, emulsification or liquid-liquid extraction. These apparats are applied in a variety of fields e.g. in the chemical, pharmaceutical, biological, petrochemical or food industry [3, 4]. In all these fields the droplet size is a key parameter, which critically influences the desired mass transfer area or is even a product property, e.g. in the production of stable emulsions. Although the droplet size plays a vital role in many processes its measurement, control and prediction is still a challenging task.”
Point 3: P2, L 87: "especially at which phase fraction measured" - there seems to be a word missing in this sentence. Please correct.
We changed the sentence to:
“Therefore, the significance of any models to predict the DSD is depends highly on the type of measurement method used in the specific environment (e.g. phase fraction).”
Point 4: P8, L 300 - 301: when calculating the volume distribution from the number distribution, why werent the shape factors taken into account? Excluding the shape factors can lead to errors.
The drops are circular, hence, no shape factor is needed. In addition, our CNN detects circles and measures their diameters, therefore is the extraction of a shape factor not possible. The following sentence is added in L311 to make clear on that.
“No shape factor was used as the drops in the images are almost perfect circles (shape factor equals 1).”
Point 5: P7, L251-253: Why didn't you use design of experiments (e.g. 3 factors at 3 or more levels) ? The experiment you conducted is perfect for the application of any kind of DOE software. It would have reduced the number of experiments (10 experiments for each impeller speed is a lot) and time required to perform the experiments, plus, you would have gotten direct ANOVA analysis and optimisation of the whole procedure via several software clicks. Or maybe you wanted to collect as much data as possible because of the CNN models?
Our goal was to determine as much data as possible, because our project partners and co-authors are simulating the apparatus with PBE and are interested in a wide span of experimental data. Additionally, we want to provide the reader with experimental data; therefore, we added all our data (q0, q3 distribution, Sauter mean diameters) to the supplementary material. As to that, other authors can use these data for further analysis (for example d10 or d90 to describe the width of the distributions).
These are data are new and very detailed and can be used for more sophistical models. Hence, other authors can try their own modeling approaches with our data, if they like to.
Point 6: P7, L271: Model fitting and CNN were both done in MatLab?
The model fitting was done in MatLab, but the CNN is based on a Python architecture. To be exact and provide fluent reading we added the following sentence in Line 280 (with markup):
“The CNN is based on a U-net architecture using PyTorch with pre and post processing by the corresponding OpenCV functions. The CNN is specially trained for this use case as described elsewhere [46]”
Point 7: P9, L314-315: Please provide a reference for this claim.
Two books covering the fundamentals of liquid-liquid dispersing in stirred vessels were added, which support the statement:
- Zlokarnik, M. Stirring: Theory and practice, Wiley-VCH, Weinheim, Chichester (2010).
- Kraume, M. Transportvorgänge in der Verfahrenstechnik, Springer Berlin Heidelberg, Berlin, Heidelberg (2012).
Point 8: P11, L360-390: Do the models show larger prediction errors for smaller Sauter diameters or bigger? Also, are the observed/predicted differences bigger for smaller impeller speeds or for larger ones? Please expand this part of the discussion section.
Thank you for this interesting idea. We analyzed our data and found, that the errors for the approach with Laso et al. are totally random distributed. It does not matter if either large or small Sauter mean diameters are predicted or at which impeller speed (interestingly!). The same was found for the standard approach with Doulah, with the exception of 5 vol.-% phase fraction. The reason is, that this is a linear approach and the influence of the phase fraction is comparably strong at 5 vol.-% phase fraction. This was found for the relative errors (%), as well for the absolute errors (µm). The discussion was extended by the sentences:
“The errors are randomly distributed with the approach of Laso et al. [2] over all data. The standard approach by Doulah [1] has the highest errors for φ = 5 vol.-%, which is a result of the linear dependence of the phase fraction in the model.”
in Line 390-393. Thank you!
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Recommendations:
The authors have investigated steady-state droplet size distributions in a DN300 stirred batch vessel with a Rushton turbine impeller using an insertion probe based on the telecentric transmitted light principle. In the process of considering these peculiarities, the practically significant approach was formulated. This is a good reason to recognize the need to publish such a work. However, there are some minor problems in the current version of the manuscript. Thus, some important updates are still needed before I can recommend the publication. In this connection I should suggest some recommendations:
- It would be very interesting to know the authors' views on how realistic the prospect of implementing their methods is, bearing in mind their competitive ability compared to existing methods, which might be simpler and more effective. I think that short enough estimations would be very useful and could increase the significance of the work.
- The authors' competent opinion on the prospects of applying their approach in other fields of knowledge, for example, in the method of liquid membranes in the crushing of droplets of immiscible liquids, or for the study of bubbles from a mixture of vapour and gas during cavitation, etc. is also very important. If there are original fruitful ideas on this subject, it would be very appropriate to mention this in the introduction or conclusion. In my opinion, this would significantly increase the interest of readers and the significance of the work done.
- The practical application of new methods is always accompanied by certain difficulties and costs. Can the authors assess the advantages and disadvantages of the approach they are developing in terms of comparison with existing practical technologies? A brief remark would be very useful to interested parties.
This paper is well enough written to understand main results. The manuscript seems to be suitable for publication. I incline therefore to consider that such a work corresponds to the content of the Journal of Applied Sciences and can be published there after minor corrections.
Author Response
Dear Reviewer,
Thank you for your helpful comments. We considered them and changed the document as following:
Point 1: It would be very interesting to know the authors' views on how realistic the prospect of implementing their methods is, bearing in mind their competitive ability compared to existing methods, which might be simpler and more effective. I think that short enough estimations would be very useful and could increase the significance of the work.
A detailed overview of available techniques and their drawbacks are detailed described in L48-L92. The measurement principle of the telecentric shadowgraphic probe is detailed described in section 1.2, L104-164, also considering advantages and disadvantages (for example increased space requirement). Nowadays, in industry 4.0 more sensors are implemented giving insight to more local process details. So in process industry besides P and T, sometimes concentration are measured, but maybe other features, like particle size distribution, may be of interest.
We added a detailed report about a project covering a comparison between a commercial available probe to the literature and added the following sentence in L164:
“A detailed comparison between this technique and a commercially available probe for droplet detection can be found elsewhere [47].”
Becker, K. et al. (2019), Abschlussbericht;
Point 2: The authors' competent opinion on the prospects of applying their approach in other fields of knowledge, for example, in the method of liquid membranes in the crushing of droplets of immiscible liquids, or for the study of bubbles from a mixture of vapour and gas during cavitation, etc. is also very important. If there are original fruitful ideas on this subject, it would be very appropriate to mention this in the introduction or conclusion. In my opinion, this would significantly increase the interest of readers and the significance of the work done.
The probe has been successfully tested also with crystallization, bubble and extraction columns, phase separation in gravity settlers, entrainment in distillation and absorption columns and detailed literature covering the applications and are described in L145-149. Nevertheless, we added the following sentence to provide the reader with possible applications and projects for the future in L421-424:
“Possible future applications are manifold, like the detection of cavitation bubbles, the measurement of bubbles in jet-loop-reactors or the use in fixed bed reactors for the characterization of the flow regime to name a few.”
Point 3: The practical application of new methods is always accompanied by certain difficulties and costs. Can the authors assess the advantages and disadvantages of the approach they are developing in terms of comparison with existing practical technologies? A brief remark would be very useful to interested parties.
This question is difficult to answer, since this is a scientific approach. Calculations or concrete comparisons in the field of commercial measurement techniques have not been made so far. Until now, there are just scientific comparisons as mentioned above.
Thank you for the ideas and honest questions!
Author Response File: Author Response.pdf
