You are on page 1of 6

Scientific Bulletin of the Politehnica University of Timisoara Transactions on Mechanics Special issue

The 6th International Conference on Hydraulic Machinery and Hydrodynamics Timisoara, Romania, October 21 - 22, 2004

THE PRINCIPLE OF NEGLECTING UPSTREAM REACTIONS


rpd . FY, Docent* Department of Hydro- and Thermo-machinery The University of Miskolc

*Corresponding author: Szent Dont u. 33. Pilisborosjen, 2097, Hungary Tel.: (+36) 26 336931, Fax: (+36) 26 336931, E-mail: arpad.fay@axelero.hu ABSTRACT This paper presents a practical principle, which states that in the case of modifying a flow boundary (e.g. adjusting a blade angle), upstream of the modification the flow is affected to a very small extent, what is usually negligible, while the downstream effect of the modification is large, usually not negligible. This is not a new statement; in special situations many researchers used it. Nevertheless, the author felt that an analysis of well-known examples from this aspect may be useful both for research and for teaching. Examples when this principle is applicable: Flow around a body placed into uniform flow (plates, airfoils, etc.) Blade cascades in wind tunnels Testing the hydraulic losses of pipe-like elements (elbows, valves, etc.) Reactions of Kaplan turbine runner blades on their inlet velocity conditions The problem of spontaneous swirl caused by pump impellers on the approaching flow Cases when this principle failed: The guide vane opening of a Kaplan turbine affects the flow in the spiral casing The flow conditions in the pump volute affect the flow in the impeller Pump testing, with special considerations. KEYWORDS Hydro machines, hydraulics, flow losses NOMENCLATURE cr [m/s] Subscripts 1u peripheral direction, upstream of runner, Fig. 4 1m meridional direction, upstream of runner, Fig. 4 2u peripheral direction, downstream runner, Fig. 4 2m meridional direction, downstream runner, Fig. 4 1, 2 front and back side, Fig. 7 small, opt, large small, optimal, large, Fig. 8 ABBREVIATION IEC International Electrotechnical Commission 1. INTRODUCTION In hydro-machinery practice one is frequently faced with flow boundary modifications. Some of these are intentional (e.g. adjusting a blade angle), while some modifications are unwanted (e.g. facing with erroneous blade shapes). In such cases the principle of neglecting upstream reactions is a potential, useful tool. This is based on the fact that upstream of the modification the flow is only slightly affected. Whether a small change in the velocity is negligible or not, depends on the accuracy requirements of the calculation. In many practical cases, however, the upstream reaction of the modification proves to be negligible. This largely simplifies the analysis (e.g. the calculation of the changes in the characteristics caused by the modification). Many researchers may have used this principle in the past. Nevertheless, it seems that focusing attention to its features, and to the cases of its failures, may be useful. 2. INTRODUCTORY EXAMPLES To formulate the principle in a more specific way, some simple flow configurations are shown in Fig. 1. On the upper part of Fig. 1 a uniform flow is seen above a plane. A frictional, turbulent (i.e. a real) flow is considered. In the middle of Fig.1 the flow boundary is modified by a hump. The shape of the 85

v [m/s] r v [m/s]

E H Q

absolute velocity at the runner (Fig. 4) velocity before modification (Fig.1, 2) velocity after modification (Fig.1) [-] error term of modification, Eq. (1) [m] head [m3/s] discharge

3. BODIES IN WIND TUNNELS Blunt bodies placed into nominally uniform flows of wind tunnels, such as the symmetric profile in Fig. 2, have usually considerable wakes. The same applies for such bodies as for the barrier in Fig. 1.

Figure 2. Symmetric profile placed into uniform flow in wind tunnel, blade cascade in the wind tunnel Figure 1. Uniform flow, modification of the flow boundary without and with flow separation hump gives the impression that no flow separation takes place on the hump. At point A (Fig. 1) which is not too near to the hump, the parameter In the lower part of Fig. 2 a blade cascade is placed into the wind tunnel. The flow upstream of the cascade changes only slightly. Downstream of the cascade, however, the flow direction is largely affected. Thus, while the upstream reaction of the blade cascade may be neglected, the downstream reaction surely may not, since the aim of the cascade is just to create the flow deflection. 4. PIPE-LIKE ELEMENTS IN HYDRAULICS In Fig. 3 a bend is shown in a pipeline. A mercury manometer indicates that checking of its loss coefficient is intended. It is known from practice that the manometer tapping should be placed away from the bend. The necessary length upstream of the bend may be rather small. The downstream length should be, however, quite large, in order to allow smoothing out of the flow disturbance caused by the bend. These are in accordance with the above principle. If the discharge in the pipe of Fig. 3 is intended to be measured with an orifice plate, then the minimum straight length of the orifice from the bend can be taken from the standards [1]. For example, using an orifice of diameter ratio 0.6, the zero additional uncertainty length is: 7 diameters from the bend if the orifice is situated upstream of the bend, and 18 diameters if the orifice is situated downstream of it.

r r v v E= r v

(1)

based on the absolute values of the vectors, gives a measure of the effect of the modification. Looking at the flow configuration with this hump, one may have the feeling that the flow can be well approximated with incompressible, frictionless flow. Based on the usual training with such classical flows, it can be stated that in this flow configuration presumably E is small at Point A. With not too strict accuracy requirements, E may be negligible. Point A is upstream of the hump. Point B in Fig. 1, the mirror of A over the symmetry axis, is situated downstream. Since the frictionless flow (in the middle of Fig. 1) is symmetrical, E is the same for B as for A, and so it is similarly negligible at B. The situation is, however, different on the lower part of Fig. 1. In this case the shape of the barrier implies flow separation downstream the barrier. Vortices are shed, the flow character changed. While at point A the effect of the modification may still be neglected, at point B parameter E, calculated with the average (turbulent) velocities, might be quite large, and so in a large area downstream, in the wake and in its neighborhood, the effect of the modification is not negligible. In both modified flows of Fig. 1 the upstream reaction of the boundary modification may be neglected in a large area. This is what the principle states. In contrast to this, it is also seen that the downstream effect of the modification may not be negligible, due to flow separation. 86

Figure 3. A bend in a pipeline, equipped with a manometer for loss measurement With less strict requirements, using 0.5 % additional uncertainty, these lengths are [1]: 3.5 diameters

upstream, and 9 diameters downstream. Thus, the flow region, where the upstream reaction of the bend is negligible (as far as the application of an orifice is concerned) depends on the accuracy requirement. It is also seen from the numerical values that the downstream reaction is much larger. 5. THE EFFECT OF KAPLAN RUNNER BLADES ON THE APPROACHING FLOW In the past the author made velocity traversing tests [2], with 3-hole cylinders, upstream and downstream of a Kaplan model runner, in a test configuration shown in Fig. 4.

speed, nor the blade angle affected the inflow conditions, and not even the presence of the blades.

Figure 4. Velocity traversing with 3-hole cylinders upstream and downstream of a Kaplan model runner It was important to make such tests in order to check the design parameters of the runner blades. At the best efficiency point of the turbine the velocity distributions of Fig. 5 were obtained. The velocity distributions were measured in many operating conditions of the turbine (covering the range of model efficiency above 75 per cent). In evaluating the vast amount of the test results it was surprising to see that at the same opening of the guide vanes (as the opening of the best efficiency point, for which Fig. 5 is valid), the proportions of the c1u, c1m distributions were the always the same, independently of the runner speed. The tests were made with H = constant, but recalculating the velocities with the similarity laws for Q = constant, just the same velocities were obtained as in the upper part of Fig. 5. Thus the speed of the runner did not affect the velocity distributions upstream of the runner (within the test accuracy). Then the blade angle was changed, and at the same opening and same discharge, again the same velocities were obtained at any speed. Then, the runner blades were removed, and the velocity traversing was made with the bladeless turbine, and at the same opening this resulted again the same velocity distributions. Thus it was concluded that the Kaplan runner did not affect its upstream flow conditions. Neither the runner Figure 5. Test results of the velocity traversing at the best efficiency point of the Kaplan model turbine. The opening of the guide vanes naturally affects the c1u, c1m distributions, but for a given model turbine the opening is the only parameter which affects the proportions of these velocities. A generalization of this conclusion may also be attempted. It seems that the usual blade shapes of Kaplan or bulb turbines (excluding extraordinary cases) do not affect their inlet flow conditions. This conclusion largely simplifies both the design and the experimental work. Knowing this, one needs to measure the upstream velocity distributions only at a few openings. From these, the inlet velocity triangles can be determined, and these are valid for any blade shape, at any operating condition. Thus, in this case the upstream reactions of the Kaplan blades were negligible. 6. THE PROBLEM OF THE SPONTANEOUS SWIRL UPSTREAM OF PUMP IMPELLERS In pump impeller design it is a crucial question whether spontaneous swirl (or prerotation) should be assumed upstream of the impeller or not. Some of the designers have a philosophy that the impeller 87

creates a spontaneous rotation upstream of the blades. The smaller the specific speed the larger the spontaneous prerotation what they assume. Other designers use zero prerotation upstream of the impeller. Csanady [3] presented an elegant proof that rotating blades cannot create spontaneous swirl. The basic idea follows from Fig. 6. Apply the moment of momentum theorem for the fluid mass bounded by the intake pipe and Sections 1 and 2. At Section 1 the fluid enters without rotation. On the pipe wall the fluid friction is very small, and its direction is axial. Therefore the moment of the momentum over Section 2 should also be zero. (Some fluctuations may appear in the velocities at Section 2 due to the passing blades but the average torque of the impulsive forces should be zero.) This implies that the rotating impeller blades cannot create prerotation upstream of the impeller.

special geometry the clearance flow can be reduced dramatically.) Then no prerotation occurs. Thus the rotating impeller blades does not create spontaneous swirl upstream of the impeller. The prerotation, if it appears, is due to other reasons. It is seen again that the upstream reaction of the rotating impeller blades is negligible, like for the turbine blades above. The prerotation, when it occurs, is caused by the downstream effect of the clearance flow, and by the downstream effect of the backflow.

Figure 6. Intake pipe section of a pump impeller The test results, however, showed small but measurable prerotation upstream several radial-flow impellers [4]. This contradiction between theory and tests created a puzzle for the author for a long time, solved at last by the following considerations: Upstream of the pump impeller the clearance flow enters into the inlet pipe with considerable rotation, and carries a moment of momentum into the inlet pipe. Mixing with the main flow this creates some prerotation. When the pump operates at a very small discharge, below a certain limit value a backflow appears flowing from the impeller into the inlet pipe at its periphery. This rotating backflow brings a moment of momentum into the inlet pipe, where mixing with the main flow creates considerable prerotation. The limit discharge of the backflow is usually small. In most cases the backflow region is found below the normal operating range of the pump. Above the limit value only the clearance flow causes prerotation. Cs. Fay [5] and Lnzmann [6] discussed of how different clearance geometries affected the pump characteristics. Their results are in accordance with this concept of the prerotation. As a theoretical exercise, let us assume that the pump operates in the normal discharge range (without backflow) and the clearance flow is zero. (With 88

Figure 7. Flow boundaries of a Kaplan spiral casing 7. THE GUIDE VANE OPENING AFFECTS THE FLOW IN A KAPLAN SPIRAL CASING One of the cases when the upstream reaction is not negligible concerns Kaplan turbines. In Fig. 7 a semi-spiral casing of a model turbine is shown. At one of its sections a mercury manometer indicates the intention to make index test by the Winter-Kennedy method [7]. This method assumes that the discharge is proportional to the square root of the differential pressure. However, it is known [8], that this is not satisfied for any section of the spiral casing. Following the recommendations of the IEC model test code [8], several sections were prepared on the model turbine for the pressure measurements. As expected, it was found that the pressure differentials depended not only on the discharge but also on the opening of the guide vanes. Thus, it was concluded that the upstream effect of the guide vanes cannot be neglected. At design point the discharge from the spiral into the stay-vane row is rather uniform (along the periphery). Thus, at equal arcs the same flow enters: Q2 = Q1, (Fig. 7). However, at an opening away from the design

value, experimental evidence shows that these discharges are not equal: Q2 Q1. This is due certainly to the fact that the flow resistance of the guide vanes varies along the periphery. Therefore the flow velocities at the casing wall, and also the pressure differentials at the various sections are a function, besides of the discharge, also of the guide vane opening. Fortunately, for the turbine tested, the required proportionality was nearly satisfied at one of the sections, and this section was suitable for the Winter-Kennedy method. It may be concluded that the guide vane opening largely affected the flow in the spiral casing, though it is situated upstream, and this effect was not negligible if the application of the Winter-Kennedy method was in view. 8. THE FLOW CONDITIONS IN THE VOLUTE AFFECTS THE FLOW IN A PUMP IMPELLER Another case, when the upstream reaction is not negligible, is the effect of the flow conditions of a pump volute on the flow in the pump impeller.

Figure 9. Pump testing 9. PUMP TESTING In pump testing (Fig. 9), when the valve body is adjusted to a new position, then the pump works at a new operating point on its characteristics. Thus, in this case the upstream effect of the valve may not be neglected. However, in the pipe between the pump and the valve the upstream effect of the valve is small. The proportions of the velocity distribution at the discharge flange of the pump are affected only slightly. It is only the pressure level variation which affects the pump. If a booster pump were applied to produce the same pressure variation at the suction flange of the pump, then the original flow conditions are essentially restored in the pump (unless cavitation disturbs the picture). Thus, in this case there exists a means to compensate for the pressure variation, and the principle of neglecting upstream reactions is still valid. 10. CONCLUSION In many flows several structural elements can be distinguished, and the fluid passes through these elements in a sequence. The surface between neighboring elements is called here as interface. The modification of the flow conditions is produced in this paper mainly by a change in the flow boundary (Sections 2 to 7 and 9). However, one example is shown here, when not the boundary but the flow angle changed at the interface (Section 8). Several practical examples are shown when the upstream reactions may be neglected (Sections 3 to 6). Some of the fundamental technical tasks are included. It is particularly important that rotating blades do not affect their inflow conditions. There are, however, warning examples when the upstream reactions may not be neglected (Sections 7 to 9). It seems that the size and the shape of the interface matters. In the examples when the principle is applicable (Sections 4 to 6) the interface is a simple circle, while when the principle fails (Sections 7 and 8) 89

Figure 8. A volute pump and its characteristics at constant speed When the volute pump of Fig. 8 operates at its optimum point (Q = Qopt), assuming proper volute design, at points A,B,C, and D (Fig. 8) the pressures are more or less the same. This uniform pressure distribution assures uniform discharge from the impeller into the volute. In design, the spiral is regarded sometimes as a single vane of a vaned diffuser. When operating at large discharge (Q = Qlarge, Fig. 8), the attack angle of the tongue is larger than its optimum value. Therefore the pressure at A is larger than at C. This means on one hand that the discharge from the impeller is not uniform, and on the other hand that a radial force arises on the impeller. Similar statement can be made for the operation at small discharge (Q = Qsmall) but the direction of the pressure differential and the radial force is opposite. The author has seen a case in the past when this radial force was large enough to destroy the bearings of the pump. Thus, in this case, the flow conditions in the volute affected the flow conditions in the impeller, and so the upstream reaction of the volute was not negligible at all.

the interface is a large circumferential surface. In Section 9, for the first sight the principle fails, but with some manipulation the validity of the principle may be assured. The main conclusion of this paper is that the principle of neglecting upstream reactions may be used in many cases simplifying both the theory and the experimental work. However, the application of the principle should be verified in each case either by tests or by estimates. REFERENCES 1. International Standard, ISO 5167-1 (1991) Measurement of fluid flow by means of pressure differential devices Part 1: Orifice plates, nozzles and Venturi tubes inserted in circular crosssection conduits running full. Publ. by ISO. 2. Fay A. A. (1969) Explanation of the difference in the moment of momentum of axial-flow impellers or runners and the shaft torque of axial-flow turbomachines. Ganz Mavag Publications No. 42. Budapest, pp 89-95.

3. Csanady G.T. (1964) Theory of Turbomachines, McGraw-Hill Book Company, New York, San Francisco, Toronto, London. LCCC Number: 6321475, 23456789-MP-10987, 14877 4. Hajd S. (1957) Prerotation upstream radial-flow pump impellers (in Hungarian), Ph.D. thesis, Hungarian Academy of Sciences 5. Fay Cs. (1995) The clearance geometry of pump impellers affects their cavitation performance, Proceedings of the 10th Conference on Fluid Machinery Budapest. Ed. Szab ., Budapest, by GTE, Paper 23, pp 187-190 6. Lnzmann K. (1988) Untersuchungen an Kreiselpumppen mit Schrgspalt. Pumpentagung Karlsruhe 1988, Section B4 7. International Standard, IEC 41 (1991) Field acceptance tests to determine the hydraulic performance of hydraulic turbines, storage pumps and pump turbines. Publ. by IEC 8. International Standard, IEC 60193 (1999) Hydraulic turbines, storage pumps and pump-turbines Model acceptance tests. Publ. by IEC

90

You might also like