THE UNIVERSITY OF MICHIGAN INDUSTRY PROGRAM OF THE COLLEGE OF ENGINEERING CAVITATION PERFORMANCE OF A CENTRIFUGAL PUMP WITH WATER AND MERCURY Fio Go Hammitt R. Ko Barton V, Fo Cramer Mo J, Robinson September, 1961 IP-528

ACKNOWLEDGEMENTS The authors wish to acknowledge the assistance of various present and past project personnel in gathering and reducing the data for this report: especially Co L. Wakamo, P. T. Chu, Jo Schmidt, Lo Eo Hearin, and T. A. Sheehano ii

TABLE OF CONTENTS Page ACKNOWLEDGEMENTS..... o o. o o....... * - o o O. O o ~ ~ * ii LIST OF TABLES..............................ao o o o o o o........... iv LIST OF FIGURES 0........0000000.0.000000..00000o00000....00 V NOMENCLATURE.......,,,,,,, o o o o o o o....................... vi ABSTRACT..oooo.oo. oo.o. oo.oooooo.o........ 0 0000000 0 00000000 vii Io INTRODUCTIONoo oooo0..o0.0...........OOO 0............ 1 II SYSTEM DESCRIPTO........... 0. 0 0 0 0 0 0 0 0 0 0 0 0 0 Instrumentation.... o.................................. 3 IIIo PROCEDURE OF EXPERIMENT.......0000....................... 3 IVo DEFINITION OF PARAMETERSo..............00o000 0000...... 7 V. DISCUSSION OF RESULTS...............00......000.............. 7 Scale Effects for Thoma's Cavitation Parameter.oo....... 7 Non-Cavitating Head vs Flow............... 0000000...... 12 Normalized Head vs Normalized NPSHO.........0......0....0. 12 Hysteresis Effect 000000.00....000.............000000.... 16 VIo APPENDIXo...........0000000.....o...oo..o o................ 21 VII. BIBLIOGRAPHY......................................000000.......00 24 iii

LIST OF TABLES Table Page I Summarization of Results - Mercury....... 19 II Summarization of Water Results and Standard Deviation - Water........................................ 20 iv

LIST OF FIGURES Figure Page 1. Sketch of Over-All Loop Layout........................ 3 2. Schematic of Pump Test Set-up.......................... 4 3. Thoma Cavitation Parameter vs Normalized Pump Speed.... 8 4. Thoma Cavitation Parameter vs Normalized Reynolds' Number,.,,,,,,,,.,,,,,,.,,,,,,,,,,,,,,,,,,,,,,,,.... 10 5. Suction Specific Speed vs Normalized Pump Speed for Water and Mercury at Two Different Normalized Flow Conditions Using Berkeley Model 1-~ WSR Centrifugal Pump.,.,...,,,...,.,,,,,.,,,,,.,, 11 6. Typical Normalized Pump Head vs Normalized Suction Head Curves for Water for Two Flow Coefficients and Constant Temperature (-.85~F) and Pump Speeds of 2420 and 3000 RPM...,,,.,,,,..,,,,,.,,,,,,,,,,... 13 7. Normalized Pump Head vs Suction Head for Water at Two Basic Flow Rates, Constant Temperature (-1650F) and Pump Speeds of 1750, 2420 and 3000 RPM....,..,.. 14 8. Typical Normalized Pump Head vs Normalized Suction Head Curves for Mercury for Two Flow Coefficients, Constant Temperature (- 60~F) and Pump Speeds of 900, 1200, 1500, and 1750 RPM........................ 15 9. Thoma Cavitation Parameter vs Normalized Pump Speed Based on First Break Due to Cavitation.............. 17 10, Net Positive Suction Head vs Head Across Pump With Increasing and Decreasing Suction Head (To Illustrate Hysteresis Effect) - Berkeley Model 1-~ WSR Centrifugal Pump With Mercury as the Working Fluid... 18

NOMENCLATURE P Pressure S Suction Specific Speed N RPM of Pump NPSH Net Positive Suction Head (total head at impeller C, above vapor) V Velocity of Fluid cTT Thoma Cavitation Parameter P-. Density of Vapor PL Density of Liquid z Height f Friction Factor of Pipe L Length of Pipe D Diameter of Pipe Hpump Pump Head h f Friction Head vi

CAVITATION PERFORMANCE OF A CENTRIFUGAL PUMP WITH WATER AND MERCURY F. Go Hammitt R. K. Barton V. F. Cramer M. J. Robinson ABSTRACT The cavitation performance of a given centrifugal pump with water (hot and cold) and mercury is compared. It is found that there are significant scale effects with all fluids tested, with the Thoma cavitation parameter decreasing in all cases for increased pump speedor fluid Reynolds' number, The data for a fixed flow coefficient fall into a single curve then plotted against pump speed (or fluid velocity), rather than against Reynolds' number, Conversely, the Thoma parameter for a given Reynolds' number is approximately twice as large for mercury as for water. The direction of this variation is as predicted from consideration of the cavitation thermodynamic parameters which vary by a factor of 107 between these fluids. No difference in cavitation performance between hot and cold water ( 160~F and 80~F) was observed. However, the thermodynamic parameters vary only by a factor of 5. vii

Io Introduction The purpose of the tests described in this report is to compare the cavitation performance of a given centrifugal pump operating with a liquid metal (mercury) with its performance operating with water. Cavitation initiation, arbitrarily defined as that operating point where the pump head has been reduced to 95% of the non-cavitating head for conditions of constant pump speed and system resistance, has been selected as the condition for comparison0 Tests with water for the same pump have previously been reportedl1 However, the significant portions are repeated herein for convenience, and the experimental data for mercury, also previously given2 are listed and compared with the water data. It was found from the previous water tests that a significant scale effect existed for a given flow coefficient when the Thoma cavitation parameter (or suction specific speed) was plotted against either normalized Reynolds' number or velocity (pump speed and fluid velocity are proportional for fixed flow coefficient)~ It is shown here that similar relations exist for mercury. The curves for a given coefficient as a function of pump speed for water and mercury appear identical, whereas those for Reynolds' number are somewhat displaced. II. System Description Loop The cavitation tests were conducted in the closed-loop facility, previously described3o Designed for cavitation testing of a venturi with various fluids, it consists essentially of a closed loop of 1 1/2 inch pipe of about 20 fto total length~ It includes two throttling valves, heater, -1

-2cooler, flow-measuring venturi, and centrifugal pump, The test venturi was replaced by a straight pipe section. for these pump tests to allow higher flow rates. Pump The tests were conducted on the Berkeley Pump Company Model. 1 1/2 WSR centrifugal pump ordinarily used to power the loopo This is a sumptype centrifugal pump with shaft overhung from a bearing housing located above the sump, The impeller fluid passages are parallel to and 5~5" above the lower horizontal, loop-piping centerline. The pump design point at its 1800 RPM maximum design speed is 40 GPM and 40 feet of head. These flow and RPM values will be designated by No and Qo respectively, throughout the report, The 6-vaned impeller is 7 3/8 inches O oD,, with eye diameter of 1 1/4 inches and inlet passage width of 3/4 inches. Its specific speed is 1040 in GPM units~ The sump is sealed from atmosphere by a stuffing box which is necessary in the present tests to obtain the required sump vacuums (and pressures) o For water a substantial vacuum is required, Because of the uncertain behavior of the stuffing-box, the experimental data obtained with water1 is less precise than that with mercury. The pump drive is through a variable-speed fluid coupling, so arranged that continuous speed variation up to about 3200 RPM is possibleo The facility has been previously described in detail3 and is shown in Figure 1. Figure 2 is a schematic pump layout.

_____ ~ DRIVE PULLEY BEARING HOUSING ] | | | / X THROTTLING VALVE MEASURING VENTURI TEST SECTION VAL>E-~STUFFING - BOX SUMP TANK THROTTLING VALVE G // \ Skt o OWA TER IN COOLING WATER OUaT Figure 1. Sketch of Over-All Loop Layout

-4SCHEMATIC OF PUMP TEST SET- UP ij_ DRIVE PULLEY FROM VARIABLE _,, g SPEED FLUID COUPLING BEARING HOUSING STUFFING BOX THROTTLE VALVE TO ATM. z-NEEDLE VALVE VALVE TO VACUUM PUMP DISCHARGE PRESSURE TAP ___ ___ CLOSED LOOP =_~.-~~~LIQUID LEVEL i~~_ _ l- |_ ~PUMP SUMP PUMP IMPELLER LONG RADIUS ELBOW SUCTION PRESSURE TAP FIGURE 2

-5Instrumentation The discharge pressure tap was located near the spot where the discharge pipe emerges from the sump casing; the suction pressure tap just before the long radius bend upstream of pump inlet (Figure 2)o The relatively small corrections for friction and elevation were made so that the pressures are referred to impeller centerline elevation. The flow was measured by a calibrated venturi through a manometer, and the pump speed by magnetic pick-up feeding an electronic countero For the mercury tests, pressures were measured by two stainless steel Heise gages with ranges of -15 psig to 45 psig and 0 to 400 psigo For water1 the pressures were read by manometers in some tests and highresponse-rate piezoelectric transducers in otherso The transducers were necessitated by the difficulty of obtaining steady-state with the substantial sump vacuums requiredo They resulted in poor accuracy of absolute pressure measurement because of transducer drift, but reasonable precision in the location of the cavitation break point. Temperature was measured by a thermocouple inserted into the stream slightly downstream of the pump discharge~ Air content for the water tests was measured in some of the initial tests using a Van Slyke instrument. Although it' varied between about 30% and 120% of saturation no effect was apparent within the precision of the data. III. Procedure of Experiment The pump was run at speeds of 1750, 1500, 1200, and 900 RPM for mercury and 3000, 2400, and 1800RPM for water1o The higher water speeds

-6were necessitated by the difficulty of obtaining an NPSH in the same range with water as is easily obtained with mercury in this facility. Thus water tests suffer somewhat in precision by the difficulty of maintaining speeds in excess of the design speed over appreciable periods as well by the difficulty.of'maintaining steady-state sump pressure0 Flow coefficients, defined as Q/Qo, of 1o2 and 0~93 were used. For a given pump speed, with the pump in a non-cavitating condition, the flow coefficient was set with the throttle valves. Then maintaining RPM and valve-setting constant, the sump pressure was lowered until significant cavitation resulted. Pump discharge and suction pressure readings were taken throughout the tests at short intervals,* The sump pressure was then increased until non-cavitating performance was attained, and pressure readings were taken continuously well into the noncavitating region. The procedure was repeated several times for most cases to afford a DT vs NPSH plot with a reasonably large number of points. The entire procedure described above was followed for each of the pump speeds and flow coefficients mentioned, several runs being made for each case0 Water runs were made for "low temperature" (r 800F) and "high temperature" (a 170~F). For mercury the vapor pressure and viscosity are relatively insensitive to temperature within the attainablerange, so only ambient temperature was used0 Additional data to better define the noncavitating conditions were obtained by running conventional SiH vs Q curves for several speeds. * In the case of some of the water testsl, these were recorded automatically from transducer output0

-7 - IVo Definition of Parameters The definition of the Thoma cavitation parameter depends upon the definition of the NPSH corresponding to cavitation initiation. This was arbitrarily specified as that NPSH for which the pump head had been reduced by 5% from the non-cavitating condition. The effect of this definition will be discussed later. The normalized Reynolds' number was defined to be unity for a pump speed of 1800 RPM and a flow rate,with 60~F water, of 40 GPMo Thus the normalized Reynolds' numbers refer to no particular point in the flow passage and are not a direct indication of degree of turbulence. A sample calculation is given in the Appendix*, The NPSH is defined for this report as the difference between the dynamic head and vapor head at pump impeller (t above vapor pressure. Vo Discussion of Results Scale Effects for Thoma's Cavitation Parameter It was found for water and mercury, considered together, that Thoma's cavitation parameter decreased virtually on a single smooth curve as normalized pump speed, N/N0, increased, for fixed flow coefficient. Although the pump speeds with mercury and water did not overlap due to equipment limitations, it appears from these data that the Thoma cavitation parameter for a given flow coefficient is a function solely of pump speed, regardless of fluid (Figure 3), * This definition conflicts with the definition of the normalized Reynolds' number previously useSd in that it was not previously referred to the pump design speed and flow but to an entirely arbitrary operating point0 A correction factor of lo69 must be applied to the normalized Reynolds' number of Reference 1 to compare with this report~ This has been done for the curves presented~

-8-.40 STANDARD DEVIATION (WATER).35. x x JSTANDARD DEVIATION (MERCURY).30 X Ix 0/0=1.2.25 x r ~~~~~X (~X x \ 03.20 v v V~~~~~~~ V V ~8.15 V~~~~ V Q/Q -.93 I,.10' X MERCUR''NH Q/0 z~i.2 o WATER 0I El V MERCURY".05 WATER Q/Q=.93 -. MERCURY - - - WATER 0.4.8 1.2 1.6 2.0 2.4 Figure 3. Thoma Cavitation Parameter vs. Normalized Pump Speed.

-9The Thoma cavitation parameter also decreased for increasing normalized Reynolds' number for both water and mercury, when considered separately, (Figure 4), although the curves for the two fluids did not coincide. For a given flow coefficient and Reynolds' number, the Thoma cavitation parameter is about twice as large for mercury as for water. This variation is in the direction predicted by the thermodynamic para4 meters, although the magnitude of the thermodynamic effect cannot be predicted. It may be that the apparent correlation in terms of velocity is actually a result of opposing separate effects due to Reynolds' number and thermodynamic parameters as suggested in Reference 1. As mentioned previously, no difference was noted between "hot" and "cold" water (- 160oF and 800F). However, the thermodynamic parameter as used in Reference 4 (equilibrium ratio of vapor volume to liquid volume formed per unit head depression) differs by a factor of about 5 from "hot" to "cold" water, but by a factor of about 107 from "cold" water to mercury. Hence the existence of a significant effect between mercury and cold water may not be surprisingo Figure 5 is a plot of suction specific speed versus normalized pump speed, It, of course, shows simply the inverse trend from the Thoma parameter plots, ranging from about 2500 in GPM units for low speed with mercury to about 4000 for high speed with water. These values appear unusually low, However, the pump is designed for reliable operation with liquid metals rather than good cavitation performance. Also the piping elbow immediately upstream of the pump suction distorts the inlet flowo

-10-.40 STANDARD DEVIATION (WATER).35 ISTANDARD DEVIATION (MERCURY).30 Q/Q0: 1.2 x x.25 IN' — Q/Q0 =1.2 0 x 2 020.15 -'\ —.10: 10.\oX/Q93 X MERCURY 0 WATER C,, —O/-.9V MERCURY.05 0I WATER 0/ 93 MERCURY ---- WATER 0 2.0 4.0 6.0 8.0 10 12 RE/REo - Figure 4. Thoma Cavitation Parameter vs. Normalized Reynolds' Number.

-115000 I X MERCURY j 0 WATER 0 9Q]0~.93 4500 AL MERCURY} 0 WATER Q/00 1.2 0 4000 0 ~~~~~b 3~500~~0 3ooo 30Normalized Flow Conditions Using Berkeley.2 Model 1 1/2 WS Centrifugal Pump. 2000 0.4.8 1.2 1.6 2.0 N/No " Figure ~. Suction Specific Speed vs. Normalized Pump Speed for Water and Mercury at Two Different Normalized Flow Conditions Using Berkeley Model i 1/2 WSR Centrifugal Pump.

-12Non-Cavitating Head vso Flow It was noted from the mercury and water data for non-cavitating conditions, that the affinity laws held closely only for flow-rates close to the design rate. For example, a maximum deviation of about 10% was noted for a flow coefficient of 1o2. This deviation from the affinity laws (which As in opposite directions for water and mercury and is presumably a result of Reynolds' number effects)may to some extent influence the conclusions regarding cavitation scale effects, since the assumption of comparable conditions for constant flow coefficient is based on the affinity lawso However, since the same general scale effect trend occurred for both high- and low-flow coefficients, the deviation from the affinity laws does not in itself explain the observed scale effects. Normalized Head vso Normalized NPSH Figures 6 and 7 show typical water data and Figure 8 mercury data, plotted in terms of normalized head and normalized NPSH (normalized in both cases by dividing through by [RPM] 2) According to ideal theory, a single curve should result, The deviations from this expectation for the noncavitating portions of the water curves are mostly (especially Figure 7) the result of drift in the transducers. Also there are the deviations from the affinity laws which were previously mentioned for either water or mercury. The purpose of these plots was to ascertain to what extent the arbitrary definition of cavitation-initiation point affected the observed scale effects. Since the slope of the cavitating portion of the curves is somewhat steeper at low pump speed (especially noticeable in the mercury

25 oN-SQ/Qo.~93 13 4 5 6 7 8 and'3000 RPM Z 3000 RPM 2426 RPM LOW TEMPERATURE- WATER,~, ~0 95 %/OF NON-CAVITATING PUMP HEAD A FIRST BREAK DUE TO CAVITATION 0 I 2 3 4 5 6 7 (NPSH/Nz) I0- FT H20/(RPM) Figure 6. Typical Normalized Pump Head vs. Normalized Suction Head Curves for Water for Two Flow Coefficients and Constant Temperature ( 8'85~F) and Pump Speeds of 2420 and 17000 RPM.

30 -25~~Q/Qo 393 K 1750 RPM 25 a- 20 o I- 2420 RPM, Q/Qo.93 I 15 T 3000 RPM, O/Q o 1.2.10 1Q I T I I I I I II ~~~HIGH TEMPERATURE WATER 0 95% OF NON-CAVITATING PUMP HEAD ____ ____ ____ ____ ~A FIRST BREAK DUE TO CAVITATION 0 I 2 3 4 5 6 7 8 9 (NPSH/N2)lO -- FT H20/(RPM)2 Figure 7. Normalized Pump Head vs. Suction Head for Water at Two Basic Flow Rates, Constant Temperature ( = l65~F) and Pump Speeds of 1750, 2420, 3000 RPM.

14 13 1500 RPMx 4900 RPM 1200RPM IIZ. 6 l t < M 1N1750 RPM 3, 10 a. 591500 RPM 8'-1200 RPM Q/0- 1.2 # I 7 _]/ - PM5 0 RPM tv 6 0 95 /a OF NONCAVITATING HEAD 4 A FIRST BREAK DUE TO CAVITATION 0 I2 3 4 5 6 7 8 (NPSH/N2)IO -FT Hg /( RPM)2 Figure 8. Typical Normalized Pump Head vs. Normalized Suction Head Curves for Mercury for Two Flow Coefficients, Constant Temperature (-" —600F) and Pump Speeds of 900, 1200, 1500 and 1750 RPM.

-16curves) it is apparent that the scale effect will be greater if the cavitation-initiation point is defined to correspond to a greater proportionate head losso However, even if cavitation initiation is defined in terms of the point of first head decrease, there will still be a substantial scale effect. This is shown in Figure 90 It is further noted from Figures 6, 7, and 8 that the water (either hot or cold) and mercury curves are generally similar in shape with fairly similar slopes in the cavitating portions when compared for the same speed. This may appear somewhat surprising in view of the large difference in thermodynamic parameter (head differential required to produce a given vapor volume under equilibrium conditions4). It is believed that any meaningful explanation of the detailed shape of these curves can only be accomplished by a careful study of the flow in the impeller as reported for example for different impellers in References 5 and 60 Hysteresis Effect A hysteresis loop in the bHi vs. NPSH curves has been noted for both water and mercury. The pump head tends to be higher for a given NPSH while NPSH is being increased, rather than decreased, through the pump cavitation region. A typical curve from the mercury data (Figure 10) illustrates the effect, Since the average passage time for fluid around the loop is only about 10 seconds (and the time between readings and reversal of pressure variation for the runs much longer), no explanation is readily apparent. Again, it is felt that only a detailed study and visualization of the flow in the impeller could shed light on this phenomenon~

-17-.40.35 x.25.20 V -V O-Q/QOo.93.15.10 0 X MERCURY' 1. 2 0 WATER /Q/%1.2.05 V MEPCURY~ Q/Qo=.93 El WATER N/N - Figure 9. Thoma Cavitation Parameter vs. Normalized Pump Speed Based on First Break Due to Cavitation.

70 65 60 70.., I, J 13'1 LU55 ___ _ _ _ / I -- - DECREASING SUCTION HEAD a-. l | | S1 l~ |INCREASING SUCTION HEAD 50 50/ I E/ 1 1200 RPM < l~~ /3 [ [ \ |FLOW = 32GPM,,, 55./F 45 El /001.2 40 i 12 13 14 15 16 17 18 19 20 21 22 23 24 25 NET POSITIVE SUCTION HEAD - PSIA Figure 10. Net Positive Suction Head vs. Head Across Pump with Increasing and Decreasing Suction Head (to Illustrate Hysteresis Effect)Berkeley Mcdel 1 1/2 WSR Centrifugal Pump with Mercury as the Working Fluid.

-19TABLE I Summarization of Results -Mercury Mercury a N/No Q/Qo RE/RE o T S.500.93 4o 41 0o 1640 2550 "t " 0. e1872 2370 1.2 5.60 0 3450 2540 tt " 0.3430 2540 tt tt " t 0o 3200 2690.667.93 5 77 0 o 1390 3030 ti tt " 0.1495 2890 "t tt " 0 o.1385 3040 "i 1.2 7.47 0.2700 2800 tt ". ttt0 o 02950 2770 II II " I 0.3200 2730:833 o93 7,24 0,1620 2795 It " "t 0.1450 2943 tt 1.2 9~36 0,2330 2930 t " tt it 0 2630 2795 II It"t 0 o.2500 2820 I I"t tt 0.2660 2765 tt If I" 0.2420 2880 o971.93 8.41 o.1635 2840 tIt " i t 0.1500 2963 ti t IT 0 o.1430 3020 i" 1.2 10 o60 0.2600 2680 "f " f tt 0.2160 2980 AVERAGE STANDARD DEVIATION - Mercury a =.o0161 a = 101.0 S Temperature - 80~F for all runs.

TABLE II Summarization of Water Results and Standard Deviation - Water Tempo N/No Q/o F. Tha11Re0 eT S XF. RaT S 0.97 0093 166 2.58 0.1732 2351,, f1.2 162 2. 523 0,232 2572 1.343 0.93 83 1. 69 0 802 4144 f" 1.2 85 1.73 0.209 2927 0.93 167 3,60 0.1071 3437 "t 1.2 162 3.48 0.2065 3033 1.665 0.93 88 2.22 0~ 0865 3930 "i 1.2 97 2.45 o.1846 3192 " O 0.93 166 4.425 00687 4935 f" 1.2 161 4.29 0 1599 3516 f" 1.2 93 2,365 0o192 3747 1.343 0.93 120 2049 0.1214 3200 "t 1.2 110 2. 28 1925 36 50 1.665 0.93 125 3.225 o0o884 4240 "t 1.2 125 3.225 0 o1618 4040 AVERAGE STANDARD DEVIATIONS - Water a =.0218 T = 387

-21VI. Appendix A. Standard Deviation Using conventional procedures, the standard deviation was computed for the points obtained from the A H vs. NPSH curves, giving a standard deviation for the Thoma cavitation parameter and suction specific speed at each given flow rate and speed. An average value of standard deviation for all points is shown on the various graphs. It was found that the standard deviations for the mercury were much less than those for the water, some to the extent of the third magnitude. This was in accordance with expectations based on the test arrangement and instrumentation which could be used, B. Data Processing The working equations in reducing the data obtained are as follows: NPSH = The net positive suction head aT NPSH/A H (1) T pump 1 s N(G.PoM) (2) NPSH3/4 P in V2 Pvapor (3) NPSH = -- PL 2gc PV P = (P ) -A- -f__L V 2 (4) in static in D 2gc n 2 2a 2 (Xi - x) x n- i-=l where x. = Data x = Average of xi n = No. of runs a = STANDARD DEVIATION x

L V2 P (P A Z - f out static out D 2gc The following is a representative calculation: Pump Head = 62.0 Pv 0 for mercury Barometric pressure = 29.50 inches of mercury Flow Rate = 32.0 GPM from the venturi calibration curve Pump Speed = 1200 RPM (1) ID of Pipe = 1.61 inches Velocity of the fluid = 4092 fps Re = 5.07 x 105 f = 0.0203 for the pipe of the type used and above Reynolds' number (2) Suction side pressure correction A Z = 14 in. = 1.166 ft. Equivalent length of piping = 4ft. Ah f fL - 0.228 fto D 2gc in P(static) + AZ - hf + Patm in P (static)in + 19.96 ----------------------------— psia (3) Density of mercury = 844 lbm/ft3 V2/2gc =.376ft.; Hsv =.l708Pin + o376 ft.

-23(4) Thus the working equations for 1200 RPM and a flow rate of 32 GPM are Pin = P static)i + 19o96 -------------------— psia NPSH = o1708 Pin + 0.376 ---------------------— ft. Hpump = (Pout - Pin) (o1708) ----------------— fto Thus if P(static)i = -2)40 psi for cavitation initiation and Pout - Pin = 62 0, then Pin = 17o56 psia and NPSH = 3o366 ft H pump = 620(0(1708) = 10o55 -ft, 3_366 NiJ T 10 = = 320 AND S = PSH374 = 2730

-24VIII. Bibliography 1. Hammitt,F. G., "Observations of Cavitation Scale and Thermodynamic Effects in Stationary and Rotating Components", Internal Report No. 7, ORA Project 03424, University of Michigan, May, 1961. 2. Barton, R. K., "Cavitation Performance of Berkeley Model 12 WSR Centrifigual Pump with Mercury and Water Comparison", Term Paper for ME 108, Mechanical Engineering Department, University of Michigan, June 5, 1961. 3. Hammitt, F. Go, Chu, P. T., Cramer, V. F., Travers, A., Wakamo, C. L., "Fluid Dynamic Performance of a Cavitating Venturi - Part II", UMRI Report 03424-3-T, University of Michigan, December, 1960. 4. Hammitt, F. G., "Liquid-Metal Cavitation - Problems and Desired Research", Paper No. 60-HYD-13, ASME, April 11, 1960. 5. Hartman, M. J., "Observations of Cavitation in a Low Hub-Tip Ratio Axial-Flow Pump", Paper Noo 60-HYD-14, ASME, April, 1960o. 6. Wood, G. M,, Murphy, J. S., Farquhar, J., "An Experimental Study of Cavitation in a Mixed Flow Pump Impeller", Trans ASME Series D, No, 4, Jour. of Basic Engr., Vol 82, ppo 929-939, (December, 1960)