Multiblock-Parallel Computation of Turbulent Flows and their Effects on Cohesive Sediment in Intake Channel

 

S. Ushijima N., Yoneyama and N. Tanaka

 

Central Research Institute of Electric Power Industry (CRIEPI)

1646 Abiko, Abiko-shi, Chiba-ken, 270-1194, JAPAN

email: ushijima@criepi.denken.or.jp

 

 

Abstract

A multiblock-parallel computation technique, which is advantageous to deal with complicated-shaped hydraulics structures and to improve computational efficiency as well, is applied to the three-dimensional turbulent flows in an intake channel. From the predicted friction velocity arising on the bottom surface, resuspension areas of the cohesive sediment in the channel are specified with the critical friction velocity, which has been investigated in flume tests. As a result, it is shown that the present numerical method is applicable to evaluate the effects of flow fields on bottom sediments as well as to predict three-dimensional turbulent flows in complicated-shaped hydraulics structures.

 

Keywords: multiblock grid, parallel computation, body-fitted coordinates, mud resuspension

 

Introduction

The accurate numerical prediction for the flow fields in hydraulics structures is necessary to design their geometries and to decide their operating conditions. In the present study, a new computational technique, multiblock-parallel computation method [1], is applied to the turbulent flows in an intake channel of a power station. The computational domain, consisting of the channel and the sea area in front of its intake slit, is decomposed into multiple blocks having simplified geometries. The boundary shapes of the blocks are adequately represented by generating body-fitted coordinates. The individual flow fields within the blocks are calculated simultaneously with an EWS cluster, which enables us to improve the computational efficiency.

In addition to the computation of cooling-water flows, the resuspension areas, where the cohesive bottom sediment is eroded by the flows, are also estimated when the flow rate increases from the present 17 to 27 due to the future replacement with more efficient generators. The critical friction velocity for the resuspension is evaluated with flume tests using the bottom material sampled in the channel and the results are compared with an empirical formulation proposed by Otsubo [2]. The resuspension areas are specified through the comparison between the distribution of the friction velocity, which was predicted in the multiblock-parallel computation, and the critical values investigated in the experiments.

In a series of these investigations, it is demonstrated that the present numerical method can be utilized to evaluate the effects of flows on bottom sediments as well as to predict the three-dimensional flow fields in complicated-shaped hydraulics structures.

 

Numerical Prediction Method and its Application

 

Multiblock-Parallel Computation Method

Since the hydraulics structures usually have complicated-shaped geometries, it is essential to deal with their boundary shapes adequately in the computation of the flow fields. As one of the effective techniques for these problems, multiblock-parallel computation method has been proposed [1]. In this method, a computational domain is decomposed into multiple blocks having more simple geometries and the flows in the blocks are predicted with parallel computation. This parallel computation is performed in an EWS cluster, consisting of four workstations connected with LAN.

In the present computation method, the individual block geometries are represented by generating body-fitted coordinates as done by Ushijima [3]. The computations of turbulent flows in the blocks proceed simultaneously and they are synchronized in the iterative computation of a pressure field and at the final stage of each computational step to exchange the variables among the related blocks.

 

Governing Equations

Since the flows in actual hydraulics structures have sufficiently high Reynolds numbers, a standard model is employed. The momentum equation is given as the following forms in the transformed space:

 

(1)

 

Here , , , , and are average velocity and external force in direction, pressure, fluid density and eddy and kinematic viscosities, respectively.

The transport equations for turbulence energy and its dissipation rate are written in the transformed space as

 

(2)

 

and

 

(3)

 

The coefficients of the turbulence model are = = 0.09, = 0.075, = 1.44, = 1.90 (Rodi [4]).

The transformed governing equations are discretized on a Lagrangian scheme in the computational space [3]. The submodules of the computational method have been applied to the flows in various conditions and verified in detail with experimental results [3], [5], [6].

 

Application to Intake Channel Flow

The three dimensional flow field in an intake channel of a power station is numerically predicted with the present computational method. Fig.1 shows the plane view of the channel, in which cooling water of 27 is taken from the sea area through an intake slit located under a curtain wall. The computational domain, consisting of the intake channel and the sea area, is decomposed into four blocks as shown in Fig.2.

 

Figure 1 : plane view of intake channel

 

Figure 2 : decomposition of computational domain

 

The total grid number is 50,274. The boundary conditions for the intake velocity, which directs outside of the computational domain, are set up on the intake port of the block-1. The block-4, corresponding to the sea area, has three boundary walls in which free boundary conditions are given and sea water is taken through them.

The calculated flow patterns near the bottom surface are shown in Fig.3 as a perspective view.

 

Figure 3 : predicted flow patterns near bottom surface

 

The water flows from the sea area in block-4 are taken through the intake slit located under the curtain wall and then direct to the intake port along the intake channel. The non-uniform velocity distributions in the transverse direction within the channel are caused by the asymmetric geometry of the cross section.

 

Resuspension Area of Cohesive Sediments

 

Basic Properties of Bottom Sediment in Intake Channel

Since the resuspension process of bed material largely depends on its grading distribution, moisture ratio and other factors, it is important to take some samples from the site and investigate their properties. In the intake channel dealt with in this study, bed materials were collected at No.1 to No.5 points shown in Fig.1.

The samples were applied to the soil tests to measure shear strength, moisture ratio, grading distribution, viscosity obtained from flow curves and other basic properties. As a result, it was proved that the bed material, which includes 70 % silt and 25 % clay, shows the non-Newtonian rheology having yield stresses in the flow curves and that the average moisture ratio is about 200 %. This means that the bed material can be taken as cohesive sediments.

 

Hydraulics Experiments

While the resuspension and erosion processes of cohesive sediments have been extensively studied, Otsubo [2] proposed an empirical formulation concerning the critical shear stress for resuspension of the mud which has yield points in flow curves:

 

(4)

 

Here is the viscosity of mud.

In order to confirm the validity of Eq.(4) for the present bed material, the critical value was investigated in the hydraulics experiments using a flume shown in Fig.4.

 

Figure 4 : experimental flume

 

The effects of the horizontal jet flows on the surface of the cohesive material were visualized and the concentration of sediments was measured in the downstream region. As a result of these experiments, the relationship between the critical friction velocity for resuspension and moisture ratio was obtained as shown in Fig.5. Fig.5 also indicates Eq.(4) as a solid curve, which was evaluated from . While these two results were derived independently, the agreement is satisfactory. From these results, the validity of Eq.(4) was confirmed and it was concluded that the occurrence of resuspension can be judged with this formulation.

 

 

Figure 5 : critical friction velocity for resuspension of cohesive sediment

 

Evaluation of Resuspension Area

The distribution of friction velocity is evaluated on the bottom surface of the channel from the numerically predicted three-dimensional flow field as shown in Fig.6. The relatively large friction velocities are found in the areas where the geometries of the channel and flow directions are rapidly changed.

 

Figure 6 : predicted friction velocity on bottom surface

 

With the distribution of and Eq.(4), the resuspension regions are specified for three moisture ratios as shown in Fig.7; regions A, B and C in Fig.7 correspond to the resuspension areas of the bottom sediment with moisture ratios of 350 %, 300 % and 250 % respectively. In these results, no resuspension area has appeared for moisture ratio of 200 %. Conclusively, although the surface of the bed material, which have relatively high moisture ratio, might be entrained when the flow rate of cooling water increases to 27 , it is expected that the most of the cohesive sediment, having the average moisture ratio of 200%, is not eroded by the flows.

 

Figure 7 : predicted resuspension area of mud on bottom surface

 

Concluding Remarks

In the present study, a multiblock-parallel computation method has been applied to the turbulent flows in an intake channel. With the predicted friction velocity, resuspension areas of the cohesive sediment were specified according to its moisture ratio on the basis of the critical friction velocity , which was confirmed with flume tests. As a result, it was demonstrated that the present numerical method can be utilized to evaluate resuspension areas for cohesive bottom sediments as well as to predict three-dimensional flows in complicated-shaped hydraulics structures.

 

References

[1]    S. Ushijima, N. Yoneyama and N. Tanaka. Multiblock-parallel computations of 3D flows in complicated-shaped hydraulics structures. Proc. 7^th Int. Symp.on Flow Modeling and Turbulence Measurements, pages 347-354, 1998.

 

[2]    K. Otsubo and K. Muraoka. Physical properties and critical shear stress of cohesive bottom sediments. Journal of JSCE, 363(II-4):225-234, 1985.

 

[3]    S. Ushijima. Prediction of thermal stratification in a curved duct with 3D boundaryfitted co-ordinates. International Journal for Numerical Methodes in Fluids, 19:647-665, 1994.

 

[4]    W.Rodi. Turbulence models and their application in hydraulics. A state of the art review presented by the IAHR section on fundamentals of division II experimental and mathematical fluid dynamics, 1980.

 

[5]    S.Ushijima, T. Shimizu, A.Sasaki and Y. Takizawa. Prediction method for local scour by warmed cooling-water jets. ASCE Journal of Hydraulics Engineering, 118(8):1164-1183, 1992.

 

[6]    S. Ushijima. Arbitrary Largrangian-Eulerian numerical prediction for local scour caused by turbulent flows. Journal of Computational Physics, 125:71-82, 1996.