Application Research of a High Turbulence Numerical Simulation Technique in a USBR Type III Stilling Basin

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The study introduces a high turbulence numerical simulation for the USBR Type III stilling basin, providing valuable insights into flow behavior and energy dissipation that can be used for practical applications in hydraulic engineering.

However, there are several flaws in the paper, and the flaws restrict the theoretical and practical engineering value of the paper. The details are listed as below:

1. It looks like commercial software was used. Can you provide the name of the commercial software? It can increase the credibility of the article.

2. The absence of validation through physical experiments or real-world data weakens the reliability of the simulation results. Without experimental confirmation, it is difficult to ensure the accuracy of the simulation outcomes in practical applications, especially under high-turbulence conditions.

3. The paper lacks a thorough discussion on the limitations of the turbulence model (e.g. k-epsilon model..) and boundary conditions. This may lead to misunderstandings about the precision and applicability of the model, particularly concerning its performance under different fluid conditions or complex boundaries.

4. In CFD simulations, grid independence verification is a fundamental step to ensure that the results are not dependent on the grid resolution. The lack of grid independence verification reduces the credibility of the simulation outcomes, especially in regions where fluid characteristics change significantly.

5. The study focuses specifically on the USBR Type III stilling basin under specific conditions, which might make it difficult to generalize the findings to other types of stilling basins or hydraulic structures without further adaptation.

Author Response

We greatly appreciate your review and valuable feedback on our manuscript. All comments were taken into account during the revision of the manuscript.

In addition, we prepared a detailed, point-by-point list of our replies to all of the comments, which is presented below.(attachment)

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

A numerical simulation study of USBR type II stilling basin. The authors have utilised the RANS k-e turbulence model to study the basin and managed to reduce the length of the basin by 12%. However, it still requires a fair number of details to be added to the article as below -

Page 1: line 45 - in the sentence "k-e" is missing in " RNG turbulence model".

Page 2: line 58 - 66, please add some relevant references here. There are a few claims which require citation.

Page 3: it appears that the "Figure 1.1 Diagram of a typical USBR III Stilling Basin" scanned from a book or a paper. Therefore, please add a relevant reference here.

Page 3 & 5: There are many mathematical equations based on Navier Stokes equations for k-e turbulence model, which are vastly available in the literature. Therefore, adding a citation and giving a brief description of the modelling (equations, methods) is sufficient at this stage. Please consider doing that.

General Comment: by looking at Figures 2.3, 2.4, etc..., it looks like the simulation was done in ANSYS, but there are hardly any details about ANSYS. Add full details of the simulation package, Fluent or CFX. 

There is a single Mesh Figure (Figure 2.3) but there is no further details on Mesh sensitivity analysis, what are the number of elements/cells? Add those details. 

Page 5: The Reynolds number (Re) is calculated based on the flow speed and the characteristics length, whereas there is no details on how the Reynolds number(s) is estimated. Re is critical for flow determination; therefore, please consider adding the details here.

For the results and discussion part, i.e. between pages 6 and 10, add some statistical comparison between the normal basin size and the optimised basin size which is claimed to be 12% shorter than the original length. This can be done by comparing the velocity profile of the water flow at a few cross-sections.

Add a few results (brief) on the Conclusion since the way it is written is very general and sounds vague.

 

 

Author Response

We greatly appreciate your review and valuable feedback on our manuscript. All comments were taken into account during the revision of the manuscript.

In addition, we prepared a detailed, point-by-point list of our replies to all of the comments, which is presented below.(attachment)

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for the responses. I still have some concerns for the paper.

1. The authors mentioned that there is no physical model test. However, the authors used the prototype test to observe the operation data. I think the authors can compare their results with the prototype data and verified the established numerical model.

2. Please provide your data for the grid independence verification.  

Author Response

1. However,the author sused the prototype test to observe the operation data.I think the authors can compare thei rresults with the prototype data and verified the established numerical model.

Thank you for your suggestion. The USBR Type III stilling pool studied in this paper is an energy dissipation structure of the drainage sluice of the flood storage reservoir of the project, which generally operates under abnormal conditions. , according to we to the build phase of the trial run and nearly twice the use of the process, monitoring the stilling basin into the pool velocity 13.60 m/s, 13.65 m/s, 13.50 m/s, respectively corresponding to the pool velocity 3.52 m/s, 3.55 m/s, 3.50 m/s, the overall prediction results conform to the model, And the appearance of energy dissipator is normal, no obvious damage, cavitation phenomenon, the downstream anti-scour groove structure is stable, no erosion damage phenomenon. Considering that this project started construction at the end of 2018, the construction period is 40 months, and the impact of the global epidemic in the middle, the actual operational monitoring data that can be referred to are not sufficient, so the specific value of prototype test is not shown in the paper. In addition, the simulation results of the paper have also passed the expert review and approval of the project organization in the early stage, and the theoretical research of the subject has also been technically checked by well-known scientific research institutes in China. In the future, we will continue to monitor and collect the actual application data of USBR Type III stilling pool, establish a more complete and scientific long-term analysis and judgment basis, and deeply study this topic through new articles. At the same time, we look forward to maintaining long-term communication with you and listening to your guidance.

 

2.Please provide your data for the grid in dependence verification.

Thank you for your question. Fluent software is one of the world's leading CFD software. It is a special software used to simulate and analyze fluid flow and heat exchange in complex geometric regions. As long as engineering problems involving fluid, heat transfer and chemical reactions can be solved by Fluent. The grid division of USBRⅢ deadening pool is spatial discretization. In terms of controlling equation discretization, the finite element volume control method is adopted by Fluent fluid computing software. The finite element volume control method divides the whole computing domain into several control bodies, and solves and integrates the control equations of each control body, so it is suitable for structured and unstructured grid elements. Users can customize the choice of using unstructured or structured grids to divide complex geometric areas. Unstructured grids are easier to deal with the computation area of complex shapes than structured grids. Fluent software has a good acceleration convergence technology, and for each kind of physical problem, the best solution accuracy can be achieved by using the appropriate numerical solution. Because it has the advantages of wide application, high efficiency and time saving, and can reach two-dimensional accuracy, it has been widely used in various engineering design problems related to fluid mechanics calculation.

Gambit is a dedicated CFD pre-processor independently developed by Fluent Company, which can complete the modeling of computational objects, grid generation, etc. The geometric and network models established in the pre-processor are mathematical models in the same environment as the previous test models, and the simulated values obtained are compared with the test values, and a good result is obtained. This verification helps to extend Fluent software to other complex flow field simulations, making it possible to conduct comprehensive and accurate numerical simulation of complex flow field structures. (From Lu Qing; Verification of 3-D Numerical Model for Open-channel Flows Basedon Fluent; Science Technology and Engineering; Vol.12No.32Nov. 2012)

In this paper, in the grid independence verification of USBRⅢ deadpool model, our team has also done specific work. Due to the urgency of time, a simple description is as follows:

1. 1.select the appropriate mesh size

According to the evaluation index of grid division and the result of trial operation, the optimal mesh size is determined after balancing the relationship between calculation accuracy and calculation cost. That is, the grid is continuously encrypted until the calculation results change within an acceptable range, so that the calculation results are considered to be close to the real results with high accuracy.

In order to determine the best mesh division method, the structural tetrahedral mesh and unstructured hexahedral mesh were divided respectively, and the mesh number, partition quality, model calculation time and simulation results of the two were compared. The mesh quality can be evaluated according to the skew coefficient or orthogonal coefficient of the model mesh. The smaller the skew coefficient is, the better the mesh is, while the orthogonal coefficient is vice versa. By systematic comparison, the skew coefficient of tetrahedral mesh division is about 0.80. The skew coefficient of the non-structural hexagon grid is about 0.69. After hexahedral mesh method is used to partition, the number of optimal grids should be determined to ensure the accuracy of simulation results. The standard section (length 1m) at the entrance of the deadpool model was divided into four mesh numbers: A (1430), B (8316), C (28322) and D (81250).

 

Figure 1    Two grid divisions, A and B (coarse grid)

  

Figure 2    Two types of grid divisions, C and D (fine grid)

2. Conduct simulation and statistical results

In this paper, the standard k-ε two-equation model and non-equilibrium wall function are used to calculate the model. The rate of change of each encrypted grid is greater than 30%, in case the change of the solution is overwhelmed by the numerical calculation (such as incomplete iteration error). By comparison and selection, the voltage drop β per unit length of the period segment is used as the standard to evaluate the grid independence. Define Δ beta/beta = | A - B | beta beta/beta A x 100%. The beta and beta B respectively A and B is the use of grid and calculate the cycle period of unit length head. For the same model, the larger the number of Re, the more grids are required and the denser the grids. This may be due to the fact that under the condition of high Re number, the flow changes more dramatically, and the mesh needs to be denser to capture the flow characteristics and get the correct calculation results. (Excerpted from Wang Yuemei, Study on the Grid Generation and Independence of the Complicated Geometric Model Mumerical Calcuation, JOURNAL OF ZHONG YUAN UNIVERSITY OF TECHNOLOGY, Vol.18No.7Feb.,2007)

When Re is (4~10) e+05, the grid is checked. Table 1 shows the grid evaluation results. From the examination results, it can be seen that when the grid division changes from A to B, the change of Δβ/β is 3.59%, while when the grid division is encrypted from B to C, the change of Δβ/β is only 5.52%, and if the grid C is encrypted to grid D, the Δβ/β is 3.60%. Therefore, if the calculation model of the standard segment adopts the same type of grid division, it is passed Grid independence test. Grid C can be used as the grid division of the computational model.

Table 1      Standard Segment Grid Independence Assessment

Grid	β	△β/β （%）	v (m/s)
A	449.07		14.50
B	432.97	3.59	13.96
C	409.08	5.52	13.55
D	394.36	3.60	13.22

 

3. Analyze the trend of the results

As can be seen from Table 1, the rate of grid cell calculation decreases with the increase of grid density. A rough grid is used for numerical simulation, and the velocity values obtained are too large. By using precision mesh, the calculated results are more consistent with the actual values. In order to ensure the calculation accuracy and save the calculation cost, the C grid is suitable for the actual situation.

4. Determine grid independence

(1) Grid independence test

According to the time step of 0.01s, the number of four kinds of grids A, B, C and D was tested for irrelevance. After the simulation reached steady state, the results were obtained as shown in Table Table 2      Standard Segment Grid Independence Assessment

Grid	Number of grids	Maximum flow velocity （m/s）	Reynolds number
A	1430	14.50	6.22e+5
B	8316	13.96	5.64e+5
C	28322	13.55	5.12e+5
D	81250	13.22	5.01e+5

 

When the number of grids in the standard segment is less than 20,000, there is a big difference between each detection item and the calculated value of the high grid number, while when the number of grids is more than 20,000, the simulation results have little difference.

The distribution of pressure and velocity under different grid numbers is further analyzed. It is generally believed that the splitter pier mainly plays the role of separation, while the baffle is mainly used to block the fluid. Therefore, this paper encrypts the velocity distribution by the height of the abrupt junction surface, and the results show that the number of grids has a great influence on the velocity distribution and Re. When the number of grids exceeds 20,000, the simulation results are very close, that is, the calculation results tend to be stable when the number of grids increases to a certain extent. Considering the calculation time and simulation error, it can be approximated that the grid number C can meet the requirement of grid number independence.

(2) Time independence test

With the grid number C as the grid model, the time independence test of the selected grid number was carried out according to the time step of 0.05, 0.01 and 0.001s.

After the simulation is carried out to steady state, the calculated results are shown in Table 3. It can be seen that when the number of grids reaches a certain level, there is little difference in the simulation results of speed and flow by time step. The speed distribution and flow distribution under different time steps are further analyzed. The results show that the time step has a certain influence on the calculation results of the velocity distribution. In the free flow field region, the time step has little effect on the calculation results, but in the sudden flow field region, the time step has a more obvious effect on the calculation results of the velocity distribution. After comprehensive consideration, the time step size of 0.01s is more suitable. (See Feng Jing an; Reliability verification method of numericalsimulation based on grid independence and time in dependence; Journal of Shihezi University: Natural Science; Vol.35No.1Feb.201

Table 3            Standard time independence verification

Time step （s）	Number of grids	Maximum flow velocity （m/s）	Flow （kg/s）
0.05	28322	13.53	0.952
0.01	28322	13.55	0.957
0.001	28322	13.56	0.958

5. Conclusion

Through the above verification and evaluation of the accuracy and reliability of the results, the unstructured hexagon grid was finally selected for discretization, and the complex boundary and water flow area were encrypted. Its advantage is that it can simulate the boundary and complex working conditions more accurately, especially for the irregular flow field area, but its disadvantage is that it takes a large amount of calculation and takes a long time. In the model grid analysis, the number of grid cells of the model is 1090179,1cell zone, 5 face zones. The simulation results of this model are well supported. By the way, the reason why the analysis of grid independence is not described in this paper is that for commercial Fluent software in many literatures, its default mesh partitioning method has the applicability and accuracy of fluid and thermodynamic simulation. On the other hand, through our simple testing and communication with the technical staff of the product side, when the mesh density division exceeds a certain data density, its accuracy is capped, which is closely related to the amount of data required by the professional category. Finally, thank you very much for your valuable suggestions. Our team is considering planning a new paper based on this aspect. Thanks again

Author Response File: Author Response.pdf

Round 3

Reviewer 1 Report

Comments and Suggestions for Authors

The paper can be accepted in the present form.

Comments on the Quality of English Language

The paper can be accepted in the present form.