The third IAHR/ERCOFTAC workshop on draft tube flows, Turbine-99 III, is based on the experience gained during the first two workshops and the development of the computational capacities. It is expected to be a step towards better understanding of draft tube flow simulation capabilities. Three cases based on the Turbine-99 benchmark were proposed to the participants: steady calculation, unsteady calculation, optimization of the draft tube performance. More than 30 simulations have been performed by the participants with several turbulence models, near wall treatment, grids and boundary conditions. The complexity of the turbulence models ranges from zero equation model to large eddy simulations. The contribution of the different participants, the protocol of their simulations and a comprehensive comparison of experimental data with the simulations are included in the present document.
A method to estimate the radial velocity in swirling flows from experimental values of the axial and tangential velocities is presented. The study is motivated by the experimental difficulties to obtain this component in a draft tube model as evidenced in the Turbine-99 IAHR/ERCOFTAC Workshop. The method uses a two-dimensional nonviscous description of the flow. Such a flow is described by the Squire-Long equation for the stream function, which depends on the boundary conditions. Experimental values of the axial velocities at the inlet and outlet of the domain are used to obtain the boundary conditions on the bounded domain. The method consists of obtaining the equation related to the domain with an iterative process. The radial velocity profile is then obtained. The method may be applied to flows with a swirl number up to about Sw=0.25. The critical value of the swirl number depends on the velocity profiles and the geometry of the domain. The applicability of the methodology is first performed on a swirling flow in a diffuser with a half angle of 3 deg at various swirl numbers, where three-dimensional (3D) laser Doppler velocimeter (LDV) velocity measurements are available. The method is then applied to the Turbine-99 test case, which consists in a model draft tube flow where the radial inlet velocity was undetermined. The swirl number is equal to Sw=0.21. The stability and the convergence of the approach is investigated in this case. The results of the pressure recovery are then compared to the experiments for validation.
The variation of the mechanical energy for the mean, oscillating and turbulent flows are considered. Their mean point out the contribution of the unsteadiness to the viscous losses and the turbulent production. A two-dimensional turbulent pulsating flow is investigated for different unsteady regimes to illustrate the effects of unsteadiness on the overall losses.
The temporal evolutions of small, streamwise elongated disturbances in the asymptotic suction boundary layer (ASBL) and the Blasius boundary layer (BBL) are compared. In particular, initial perturbations localized (δ-functions) in the wall-normal direction are studied, corresponding to an axi-symmetric jet coming out of a plane parallel to the flat plate. Analytical solutions are presented for the wall-normal and streamwise velocities in the ASBL case whereas both analytical and numerical methods are used for the BBL case. The initial position of the perturbation and its spanwise wave number are varied in a parameter study. We present results of maximum amplitudes obtained, the time to reach them, their position and optimal spanwise scales. Free-stream disturbances are shown to migrate towards the wall and reach their (negative) optimum inside the boundary layer. The migration is faster for the ASBL case and a larger amplitude is reached than for the BBL. For perturbations originating inside the boundary layer the amplitudes are overall larger and show the phenomenon of overshoot, i.e. positive amplitudes moving out of the boundary layer. The overall largest amplitudes are obtained for the BBL case, as in other studies, but it is shown that for free-stream disturbances initiated somewhere downstream the leading edge streak growth may be amplified due to suction since in the BBL the disturbance mainly advects above the boundary layer.
Propagating bending waves are studied in plates made of aluminum and wood. The waves are generated by the impact of a ballistic pendulum. Hologram interferometry, with a double pulsed ruby laser as the light source, is used to record the out of plane motion of the waves. Elliptic-like fringes visualize differences in wave speed for different directions in the anisotropic plate and circular ones are obtained for the isotropic plate. The experimental data for the isotropic plate compare favorably with analytical results derived from the Kirchhoff-plate equation with a point impact of finite duration. A similarity variable is found when starting conditions are modeled as a Dirac pulse in space and time, that brings new understanding to the importance of specific parameters for wave propagation in plates. A formal solution is obtained for a point force with an arbitrary time dependence. For times much larger than the contact time, the plate deflection is shown to be identical to that from a Dirac pulse applied at the mean contact time. A method for determining material parameters, and the mean contact time, from the interferograms is hence developed.
The Turbine 99 workshop, held in Porjus, Sweden, 20-23 June 1999, was organized to determine the state-of-the-art of CFD simulations of draft tube flows. A total of 16 groups accepted the invitation to compute the draft tube flow. The following experimental data was available before the workshop: the axial and swirl velocity components at the inlet (with rms-values and one Reynolds' stress component) and the pressure distribution around the outlet cross section. Before the workshop, the groups submitted simulation results (using 12 different CFD codes) which were compiled by the organizers to an extensive set of data available at the workshop. The experimental data for the pressure recovery factor, pressure distributions along the draft tube walls and the detailed velocity field in one downstream cross section was presented during the workshop. In this report a summary of the main results and conclusions of the workshop is given, together with the written reports from the experiments and the simulations. In addition, the report contains all relevant background information for the workshop such as the draft tube geometry, provided data, requested information etc. and is thus the main document from the workshop. As the simulations provided more data than was available for comparison with the experiments, the simulation results will be available in a separate document provided by the organizers. One major conclusion of the workshop is that much attention must (still) be payed to the grid quality and the boundary conditions, factors that need to be strictly specified before a fully relevant comparison can be made between different flow models.
The flow field inside and downstream of an open channel placed near the surface of a free flow (such as the tail water of a turbine) is characterized in detail. The channel cross-section is U-shaped and in the downstream end is placed a ramp on the bottom which accelerates the flow passing through the channel. This flow is intended to catch the attention of fish and improve their entrance to fishways, which has also been successfully demonstrated in field tests.
On considère un mécanisme qui peut conduire à la croissance algébrique, suivie d'une décroissance exponentielle de petites perturbations non axisymétriques dans un écoulement en conduite. Le mécanisme est interprété comme une résonance directe entre les perturbations de pression et de vitesse d'écoulement. Résolution numérique des problèmes de valeur propre pour les modes de pression et de vitesse. Influence du nombre de Reynolds
The development of a small three-dimensional disturbance in plane Poiseuille flow is considered. Its kinetic energy is expressed in terms of the velocity and vorticity components normal to the wall. The normal vorticity develops according to the mechanism of vortex stretching and is described by an inhomogeneous equation, where the spanwise variation of the normal velocity acts as forcing. To study specifically the effect of the forcing, the initial normal vorticity is set to zero and the energy density in the wavenumber plane, induced by the normal velocity, is determined. In particular, the response from individual (and damped) Orr-Sommerfeld modes is calculated, on the basis of a formal solution to the initial-value problem. The relevant timescale for the development of the perturbation is identified as a viscous one. Even so, the induced energy density can greatly exceed that associated with the initial normal velocity, before decay sets in.
The direct resonance mechanism between vertical vorticity and vertical velocity is studied for plane Poiseuille flow. The resonance term, obtained from the initial-value problem, has been determined numerically for various initial disturbances. The amplitude, in wave-number space, is found to decay with time for all resonances. Because of its appearance in the Laplace-transform plane as a double pole, we show that a resonance term will have an initial amplitude which grows with Reynolds number, thus indicating that resonances will dominate the initial development of a disturbance at large Reynolds number.
Based on the normal velocity-normal vorticity (v - η) formulation for the development of 3D disturbances in plane-parallel shear flows, the non-linear terms in the governing equations are derived as convolution integrals of the Fourier-transformed variables. They are grouped in three categories: v - v, v - η and η - η terms, and are expressed in a simple geometric form using the modulus of the two wave-vectors (k′ and k″) appearing in the convolution integrals, and their intervening angle (χ). The non-linear terms in the v-equation involving η are all weighted by sin χ (or sin2 χ). This confirms the known result that non-linear regeneration of normal velocity, necessary for a sustained driving of 3D disturbances, is not possible for stream-wise elongated structures (α = 0), only. It is therefore suggested how transiently amplified η can interact with decaying 2D waves to activate (oblique) waves which may be less damped than the 2D wave. This is shown to be possible for Blasius flow. In the η-equation, non-linear effects are possible for elongated structures resulting in shorter spanwise scales appearing at a shorter time-scale than the (linear) transient growth. A numerical example shows the details of this process in plane Poiseuille flow. From an inspection of the y-dependency (wall-normal direction) of the non-linear terms it is suggested that higher y-derivatives may give rise to non-linear effects in the inviscid development of perturbations. Also, a result for the y-symmetry of the non-linear terms is derived, applicable to plane Poiseuille flow.
A linear mechanism for growth of three-dimensional perturbations on plane Poiseuille flow is investigated. The mechanism, resonant forcing of vertical vorticity waves by Tollmien-Schlichting waves, leads to an algebraic growth for small times. Eventually, viscous damping becomes dominant and the disturbance decays. The resonance occurs only at discrete points in the wave-number space. Nine resonances have been investigated. For these, the phase velocities range from 0.67 to 0.81 of the center-line velocity. The lowest Reynolds number for which the resonance can occur is 25. The strongest resonance appears only above a Reynolds number of 341. Also, two cases of degeneracy in the Orr-Sommerfeld dispersion relationship have been found.
Degeneracies of the Orr-Sommerfeld eigenmodes and direct resonances between the Orr-Sommerfeld eigenmodes and vorticity eigenmodes are studied in water-table flow. The sensitivity of the characteristics of these algebraic mechanisms to flow parameters, such as the Reynolds number (R), the slope of the table $(\theta)$, and the material parameter $(\gamma)$, are investigated. It is found that the mechanisms become operative at subtransitional R, and their damping rates decrease with increasing R. When the mean flow profile is slightly distorted from the ultimate parabolic profile, the characteristics of the direct resonances show remarkable variations. Also, some of the algebraic mechanisms in water-table flow are shown to have the same characteristics and modal structures as some of those in plane Poiseuille flow.
The perturbation velocity field induced by a three-dimensional surface distortion in a boundary layer flow is considered. For small amplitudes, the kinetic energy is shown to be composed of two factors: one associated with the surface structure and the other with the velocity profile. Level curves of the profile factor, in the (alpha, beta) wavenumber plane, are ridge-like and approach the beta-axis as the Reynolds number increases. Thus, in the inviscid limit, the kinetic energy is confined to structures infinitely extended in the streamwise direction. For a certain class of surface structures, also the level curves for the kinetic energy have been determined. It is shown how a spanwise modulation and an aspect ratio of the surface distortion change the position of the level curves and the amplitudes.
The development of localized disturbances in parallel shear flows is reviewed. The inviscid case is considered, first for a general velocity profile and then in the special case of plane Couette flow so as to bring out the key asymptotic results in an explicit form. In this context, the distinctive differences between the wave-packet associated with the asymptotic behavior of eigenmodes and the non-dispersive (inviscid) continuous spectrum is highlighted. The largest growth is found for three-dimensional disturbances and occurs in the normal vorticity component. It is due to an algebraic instability associated with the lift-up effect. Comparison is also made between the analytical results and some numerical calculations. Next the viscous case is treated, where the complete solution to the initial value problem is presented for bounded flows using eigenfunction expansions. The asymptotic, wave-packet type behaviour is analyzed using the method of steepest descent and kinematic wave theory. For short times, on the other hand, transient growth can be large, particularly for three-dimensional disturbances. This growth is associated with cancelation of non-orthogonal modes and is the viscous equivalent of the algebraic instability. The maximum transient growth possible to obtain from this mechanism is also presented, the so called optimal growth. Lastly the application of the dynamics of three dimensional disturbances in modeling of coherent structures in turbulent flows is discussed.
The transition from laminar to turbulent flow in porous media is studied with a new method. To mimic inter-connected pores, a simplified geometry is studied consisting of a pipe with a relatively large diameter that is split into two parallel pipes with different diameters. This is a pore-doublet set-up and the pressure drops over the parallel pipes are recorded by pressure transducers for different flow rates. Results show that the flow in the parallel pipes is redistributed when turbulent slugs pass through one of them. The presence of the slugs is revealed by positive skewness in the pressure signals as well as an increase of the standard deviation of the pressure drops and correlation between the pressure drops of the pipes. A frequency analysis of the pressure drops show that lower band frequency pressure variations in one pipe are communicated to the other pipe while higher band frequencies are filtered out.
When Atlantic salmon (Salmo salar) and sea trout (Salmo trutta) migrates upstream rivers they encounter obstructions such as hydropower plants. To increase the water velocity at the fish inlet of a fishway an attraction channel is used. The channel increases the velocity without using extra attraction water from the reservoir. Field experiments show that fish use the channel, and lab experiments show that the water velocity out of the channel is 38 % higher than the surrounding velocity and the increased velocity lasts for about 18 water depths down stream.
A flow device that accelerates turbine tail water (or any free stream) to act as an attraction for migrating fish is field tested. The device consists of an open (U-shaped) channel which accelerates the incoming flow by a local constriction of the cross-sectional area. The velocity increase has previously been investigated in a lab-scale model and an increase of 38% has been established. In the summers of 2004 and 2005, a full-scale prototype of the attraction channel was tested at the Sikfors hydropower plant in the Pite River in Sweden. The channel was equipped with underwater cameras to monitor and record the fish swimming through it. The tests show that the fish do swim through the attraction channel. During the same time period in 2004 and 2005, 57 and 471 fishes swam through the channel, respectively. The major change of the channel between the two years was that it was painted black for 2005.
Fiber reinforced composite materials often consist of fibers gathered in bundles. Thus, during manufacturing, a liquid resin impregnates a multiscale porous medium. For wetting systems the capillary pressure jump becomes much higher in the smaller pores, i.e., within the bundles, and in addition to any applied pressure gradient there will be a local driving pressure gradient between the small- and large-scale areas. Such gradients will influence mechanisms, such as void formation and particle filtration. Hence, it is of interest to clarify the mechanisms for the wetting in general and the influence from the detailed geometry of the fiber network in particular. In this article, a porous pore-doublet model is studied in order to determine if an overflow of liquid can explain a leading flow in the smaller capillaries, and at which conditions it takes place. Experiments, as well as theoretical calculations on this generic geometry show that the leading front can be in the smaller capillary, as well as in the larger one. The outcome is dependent on the actual permeability of the porous material being a parameter that determines to what extent the larger capillary feeds the smaller one.
The flow properties of the Turbine-99 draft tube have been subject of much interest by the research community. The first part of the draft tube, i.e. the cone, raised many questions about the pressure recovery and the time dependence. A complementary experimental investigation was therefore performed in the model test facility of Vattenfall at Älvkarleby, Sweden, to get time resolved wall pressure data. Good agreement with the old measurements is achieved. However, the positions for the T(n) and R(n)-cases are not as expected. The results give a detailed map of the pressure behaviour in this part of the draft tube and are thus of interest for the validation of CFD simulations. The pressure amplitude at the inlet of the draft tube is 1000 Pa. The results also give new information regarding the cases used in the Turbine-99 workshops.
Isotropic and non-isotropic plates are impacted by a ballistic pendulum. The bending waves that are generated are studied with holographic interferometry using a double pulsed ruby laser as light source. The pulse shape changes with time because of the dispersivity of the waves. Initially the fringe pattern in the isotropic case is cylindrically symmetric and determined from an initial value problem. Later, when the waves have reached the plate rim, in- and outgoing waves gradually develop fringe patterns which in the end will be a combination of eigenmodes of the plate. A solution to the corresponding Kirchhoff plate equation is presented, which in the special case when the impact is modelled as a Dirac-pulse in space and time, is shown to depend only of the distance to the impact point divided by the square root of the time after impact and a parameter containing plate parameters. From the slope of the central deflection material parameters can be determined. Another solution, assuming a finite impact time, is shown to agree better with experiments. Results from investigations of non-isotropic materials are also presented.
To study influences of gas jet instability on tuyere refractory wear, gas injection was performed in an air-water system with a tuyere of 2 mm inner diameter. High-speed photography was used, with a framing rate of 8000 pictures per second, to film the tip region of a free and a half free tuyere. Characteristics of the cavity formed as a result of the jet instability were measured from the films, and the results were used in equations of bubble dynamics to calculate the pressure generated by the cavity motion. The film sequences show that as a result of the distortion of the gas-liquid interface, a throat in the gas jet is formed about 1.5D (tuyere i.d.) downstream of the tuyere tip. Radially moving gas starts to form an expanding cavity. The radius of the throat increases as it is pushed forward by the cavity expansion. With its radius reaching the maximum, the cavity stops growing. When the cavity collapses, bubble swarms are generated in the region near the tuyere. The cavity expands to 2D-4D (tuyere i.d.) within 1-15 ms. The maximum velocity of the expansion is about 10 m/s and acceleration ranges from 20 to 80 m/s2. The pressure calculated by using the cavity expansion data is in good agreement with the pressure measured at back-attack, which is around one half of the absolute pressure for the gas injection. This implies that the back-attack and cavity expansion are the same phenomenon. Passing the transition point of flow regimes the pressure increases very slowly as the injection rate increases. Occassionally, the cavity does not collapse immediately and it contracts after reaching the maximum radius. The cavity contraction generally takes longer than the expansion, with a velocity of about 2 m/s. The pressure reduction from the contraction is less than 0.1 bar which can not cause the formation of vapour bubbles in the liquid. By the cavity expansion, a liquid flow is set up which deforms and disintegrates gas bubbles nearby. At the moment of disintegration, liquid penetrates the concave side of the deformed bubble. The liquid flow may lead to an impact pressure of 30-90 bar in water and 210-630 bar in liquid steel. This pressure may cause refractory erosion with a pattern similar to that previously observed on H3BO3 disk surface. Besides the pitting erosion of the tuyere refractory, the influences of the jet instability on other factors of the refractory wear are also discussed.
The problem of tuyere refractory wear has been studied in an air-water model. Erosion tests, with boric acid (H3BO3) disks as refractory simulators, and measurements of back-attack frequency were carried out. The erosion pattern showed two distinct features: isolated elliptical pits and a continuous irregular shear wear pattern. The influence of surface hardness and gas flow rate was investigated. Pitting was found most frequently on disks formed at the lowest pressure (10 tons), but for pressures greater than 20 tons, little difference was seen between disks. When the gas flow was in the bubbling regime, pitting was observed inside the region closest to the tuyere tip, with a maximum at the transition to jetting flow. Occasionally, pits could still be observed when the gas flow rate was rather high (NMa = 1.82). The irregular wear pattern appeared independent of disk surface properties, however sensitive to the gas flow rate. In the bubble flow regime, wear was seen only outside a certain radius, which corresponds to the radius of the bubbles. In the jetting regime, wear was also observed close to the tuyere. The disk weight loss showed a maximum in the bubbling-to-jetting transition region, where the back-attack frequency also reach a maximum. The results support the idea that cavitation erosion, through liquid microjet pitting, is the main mechanical wear agent. A model for the generation and collapse of cavitation bubbles is proposed. Applied to gas injection into liquid metals, the model suggests greater erosion due to higher cavitation intensity, but also indicates ways of reducing cavitation erosion of tuyere refractory.
The problem of tuyere refractory wear has been studied in an air--water model using a 4 mm diamter tuyere. Erosion tests, with boric acid (H sub 3 BO sub 3 ) disks as refactory simulators, and measurements of back-attack frequency were carried out. The erosion pattern shows two distinct features: isolated elliptical pits and a continuous irregular shear wear pattern. The influence on these features of the surface hardness and gas flow rate was investigated. 24 ref.--