An Overview of Computational Fluid Dynamics as a Tool to Support Ultrasonic Flow Measurements
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe reviewed manuscript "An Overview of Computational Fluid Dynamics as a Tool to Support Ultrasonic Flow Measurements" provides a novel review of the latest years of research regarding CFD and its application to UFMs.
The authors introduce UFMs technology using the transient-time method for flow measurement, utilising wetted and clamped-on transducers. Additionally, a clear and accurate description of the physical principles and underlying mathematical simplifications used in the current UFM analysis is done. Different uses of CFD to aid the design and improvement of UFMs are described and discussed, highlighting advancements' barriers and opportunities based on the state-of-the-art.
Finally, a simple conclusion is reached in which the authors discuss the pros and cons of their study, yielding interesting suggestions to the research and industrial community to further improve UFMs by the use of numerical simulations.
I would like to thank the authors for the effort and time placed into such a necessary study, which is clearly lacking in the current literature. The manuscript is written in correct and clear scientific English. The topic of study and "red thread" are clearly stated and followed. Educated insights are provided throughout all sections, and a clear conclusion is reached.
In its current state, I think the manuscript is in a state of acceptance with minor revisions. I would like the authors to reflect and incorporate the following comments and suggestions, which I believe, would improve the quality of the text even further:
1. In fundamental fluid mechanics research, the main flow direction is generally represented from left to right on paper. In the manuscript, the figures show the streamwise direction from right to left. I would suggest that the flow direction gets changed, however, there is nothing wrong with the current state of the figures.
2. Lines 173-174: It is mentioned that t_{ba} > t_{ab} because of the acceleration of the sound wave. Does the sound wave also decelerate if sent in the counter-flow direction? Including the answer to this question could be helpful.
3. Line 206: Typo in "TThere"
4. Eq. 8: The term t_{0} is included as the time of the pulse through the transducer. Why is this term only included in the clamp-on equations and not in the wetted configuration equations?
5. Lines 278-279: I would suggest that a reference is placed here that supports the definition of the profile factor k and its mathematical description.
6. Eq. 15 and Eq 18.: I would suggest adding a reference to their definition and original papers.
7. I would greatly suggest the authors add figures from the literature in each/some of the sections to provide visual insights and support the text. Furthermore, I am sure that the authors cited in this manuscript would be very pleased to have a representative figure from their work placed in this manuscript.
8. Section 4.1.: I do have some reservations about the way the authors approach such a complex topic as turbulence modelling. I do understand that turbulence modelling is a broad subject that has its own line of research. However, I do have two main points of discussion:
8.1.: I do not fully agree where the authors state that hybrid RANS-LES models are preferred to pure RANS and LES. The complexity of hybrid RANS-LES meshes, the unsteady nature of the simulation, and the parameter tinkering to obtain an accurate result do not provide an obvious advantage to e.g. RANS k-omega SST or LES WALE.
8.2.: When stating RANS models and their qualities, the authors do so too broadly. Inside RANS there are a multitude of approaches to turbulence (e.g. linear eddy-viscosity models, non-linear models, Reynolds Transport Models, Data Augmented, EARSMs etc.) and I believe that these should be addressed accordingly. Specifically, in Lines 515-156 where the authors state that (all) RANS models overestimate streamwise vorticity at great distances. I recommend the authors slightly extend the discussion in section 4.1. mentioning the variety and complexity of RANS models, and their strengths and weaknesses. And the difficulty of having an all-in-one turbulence model solution when performing cost-effective RANS simulations vs LES. Depending on the RANS model chosen, one can obtain underestimated or overestimated quantities, mostly in adverse pressure gradient regions or areas affected by Reynolds Stress Anisotropy.
9. Section 4.3.: I would recommend that the statement in lines 598-601 be made clear at the beginning of the section. In its current state, it seems like AI and ML methods are going to take over CFD analysis for UFMs until the reader reaches the abovementioned lines.
I would like to thank the authors again for their work and time to address my comments.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThis paper describes how UFM works, discusses how CFD can be used as a tool to support improvements, and shows relevant trending areas that warrant further research. It also provides an excellent summary of previous work.
Before this paper can be accepted, it is recommended that the author revise the paper to address the following issues in order to improve the quality of the paper.
1.In Eqs. 7, what is the actual physical meaning of αUFM and cUFM? What is the speed of sound and the refraction angle in an ultrasonic flow meter? Similarly, why are αf and c replaced by αUFM and cUFM in Eqs. 12?
2.t0 is the time that pulses takes to travels trough the transducer. Why does the wet transducers not need to consider t0? If t0 is considered, it seems that t0 cannot be eliminated in the calculation process and will still appear in the speed measured by the UFM.
3.How to calculate k using Eqs. 18? What is the relationship between UUFM, Um and umax, um? For a fully developed turbulent flow, since the velocity is only related to the radius r, does it mean that the velocity distribution is centrally symmetric, and k is obviously equal to 1?
4.The discussion of RANS, LES and DNS methods is superficial, without a more in-depth analysis of the applicability, accuracy and limitations of different turbulence modeling methods under specific industrial conditions. Combined with actual industrial application scenarios, and provide reference suggestions in practical applications.
5.Add recent research on the combination of ML/AL and CFD, especially how to optimize profile factors or reduce simulation costs through AI.
6.It is best to concise the description of the working principle of ultrasonic flowmeters and focus on the specific methods and results supported by CFD.
7.Line 206, spelling error, "TTHERE"; line 291, the bracket after wall is redundant
Comments on the Quality of English LanguageCould be improved.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf