F12-1 EXPERIMENT F12 PARALLEL SERIES PUMP TEST SET Objective: To obtain performance data for typical centrifugal water pumps in single, series and parallel operation Background: A schematic of the apparatus, manufactured by G. Gilkes and Gordon Ltd., is given in Figure 1. Refer to Ref. 1 for more detailed information on the series/parallel test set. The main components of the system are: (i) (ii) two centrifugal pumps with 5" diameter impellers, a rectangular "V" notch weir for flow rate measurement, (iii) a venturi type flow meter having a 1.182" throat diameter, (iv) two 3 H.P. variable speed-d.c. drive motors with variable speed control systems, (v) suction and delivery pressure gauges for each of the pumps, (vi) a tachometer supplied for measurement of the speed of the pumps.
F12-2 (1) Pumps (2) Dynamometer Motors (3) Venturi (or similar) Discharge Meter (4) Valves closed for Series operation (5) Cress connection Valve closed for Parallel operation (6) Head Control valve Figure 1: Schematic of test system.
F12-3 Theory: Referring to Figure 2, a theoretical head discharge curve may be obtained from a derived general expression This equation can be rearranged to give: H = U 2 V 2 cos α 2 (1) g H = u 2 v u2 u 1 v u1 (2) g but pumps are usually designed such that the angular momentum of the fluid entering the impeller is small, and consequently: V u2 = V 2 cos α 2 (3) = u 2 V R2 cotβ 2 (4) Neglecting the vane thickness at outlet then the theoretical discharge rate is where b2 is the blade width at R. Q =2π Rb 2 V R (5) Thus eliminating V R2 the following relationship for the Head H is obtained H = u 2 2 u 2 Qcot β 2 (6) g 2π R b 2 g Hence for a pump running at a particular speed the theoretical head H varies linearly with the flow rate Q. For blades which are radial at exit, i.e. β 2 = 90, H is independent of Q. Figure3 thus illustrates H vs Q for β 2 < 90, - β 2 = 90 and β > 90. If the head losses are subtracted from the theoretical head discharge curve, the actual head discharge curve is obtained. The most important subtraction is not an actual loss but a failure of the finite number of vanes to impart the relative velocity with angle β 2 of the blades. The fluid is actually discharged as if the lades had an outlet which is less than β 2 for the same discharge. This inability reduces V u2 and hence, decreases the actual head produced; this is called circulatory flow and is shown in Figure4. Another loss is due to fluid friction. The general equation for friction loss is: h f = f L v 2 (7) 4m 2g
F12-4 where f is a friction factor, L is the length of the channel, m is the hydraulic radius of the channel section and v is the velocity at the section with the hydraulic radius m. The final loss to consider is that of turbulence, the loss due to improper relativevelocity angle at the blade inlet. At the impeller discharge the loss is mostly caused by a high rate of shear due to a low average velocity in the volume and a high velocity at the impeller discharge. There is also a shock loss at the entrance to the impeller due to diffusion. These losses may be minimized by designing a pump to operate at a given discharge (at a given speed) at which the relative velocity is tangent to the blade at the inlet. This is the condition of best efficiency and here the shock and turbulence losses are minimal.
F12-5 Figure 4: Head-discharge relationships. In addition to the head losses and reductions, pumps have torque losses due to bearing and packing friction and disk friction losses from the fluid between the impeller and housing. Internal leakage is also an important power loss, in that the fluid which has passed through the impeller, with its energy increased escapes through the clearances inside the
F12-6 pump and flows back to the suction side of the impeller. (a) Torque (T) (b) Input b.h.p. F = force indicated on torque indicator (8) T = F X 0.525 ft [ft.lbf] (9) b.h.p. = 2 πnt (10) 33000 where N is the speed, r.p.m. (c) Flowrate Q (i) From the weir, Q is read directly from the scale. Notice that the scale is logarithmic. (ii) From the Venturi meter, the differential head can be read. (iii) H (dynamic, suction) = V 2 inlet = H (dyn) (11) 2 g where V inlet = Q/A inlet (12) (iv) A = inlet area = π (0.75/12) 2 = 0.1227 ft 2 (13) Hence: Pi (suction) = Pi (actual) + H (dyn) (14) The calibrations are similar to that for the suction side, i.e. where where then 2 V H ( dyn) = 2g (15) V e = Q/A exit (16) A exit = π(0.5/12) 2 = 0.00545 ft 2 (17)
F12-7 where Ze is the exit elevation P (delivery) = Pe (abs) ± Ze + He (dyn) (18) (v) Total Head H H(Total) = P del - P suction (19) (vi) where P and H are in ft. of water. Water Horse Power w.h.p. w.h.p. = where ρ = water density ρ Q H(total) 33000 (20) (21) (vii) Efficiency (η) η = w.h.p./b.h.p. (22) Safety Rules: 1. Before running the experiment or turning power on, check all electric cords and hoses and make sure that there are no cracks or exposed areas. 2. This experiment is equipped with a mercury manometer. Extreme care should be taken not to break it. Experimental Procedure: Two speeds for the tests will be selected (the actual values will be decided on by the Instructors). Prior to starting the motors, the zero position of the 'V notch gauge should be adjusted. The exhaust valve (3) being closed, the drive motors are switched on with the control variac set at zero. The variac is then increased to almost 50% full scale and the air bleed valves on the venturi system and the delivery pressure gauges are vented until the 'air is excluded from the system. The motors are then switched off and the remaining pressure head in the system will be sufficient to air vent the suction pressure gauge bleeds. The single pump tests are then commenced. Motor A is then started and the speed adjusted to the required value with no delivery and the gauge readings noted. The exhaust valve (3) is then fully opened for maximum delivery rate and the speed again adjusted; the appropriate readings on the venturi and V notch as well as the pressure gauge noted. A number of intermediate flow rates, say five in number, should be run in order to give a completely defined curve. Since the curve of total head against flow rate is not linear, the intermediate steps must be carefully selected, if a useful accurate performance curve is to be obtained. This procedure must be repeated for each speed. The procedure is repeated for pump B and the resulting data plotted on a curve of total head vs delivery rate for constant speeds, noting the efficiency at each data point. The general procedure is basically the same for parallel operation with both pumps, similar graphs of total head against delivery rate be plotted. Graphs of efficiency against
F12-8 delivery rate at constant speed should also be plotted for each arrangement. For some insight into the results of a typical test, see Ref. 2. After each group has obtained different sets of constant speed data, it will be possible to collate the data to obtain total performance curves for the pump set. Proper analysis of this experiments data can be found in Chapter 10 of Fox and MacDonald. Students are expected to fit data to a polynomial such as H = A-BQ 2 and give appropriate error estimates for these findings. Simple addition of head and flow rate values for series and parallel pump tests and comparison to single pump A and pump B characteristic curves should be discussed. Causes for error should be related back to theoretical losses shown above. Bonus analysis: Students should be able to, with the help of a textbook, nondimensionalized data to fit any pump analysis presented to them in the tutorial questions. References: 1. "Parallel-Series Pump Test Set", operating and maintenance instruction, by Gilbert Gilkes and Gordon Ltd., Ins. No. 78. 2. "The Testing of Water Turbines and Centrifugal Pumps", by Paul N. Wilson, published by Gilbert Gilkes and Gordon Ltd. 3. Introduction to Fluid Mechanics, Fox and MacDonald