THE SHORTCOMINGS OF USING PUMP SUCTION SPECIFIC SPEED ALONE TO AVOID SUCTION RECIRCULATION PROBLEMS by Allan R. Budris Director of Engineering ITT A-C Pump inet Allan R. Budris is presently Director of Engineering at ITTA-C Pumpin Cincinnati, Ohio. He received his MME. (Fluids Major) from Newark College of Engineer- ing (1968) and has attended management courses at American Management Associa- tion, Dresser Management Developmentand the Goldratt Insitute. Before joining 1TT ALC Pump, he was Manager of Engineering for ITT Marlow in Midland Park, New Jer- ‘sey. He previously held various postions with Dresser Pump Division, Dresser Industries ine He as had! several journal publications. He isalso a member of| ASME and has been Committee Chairman Hydraulic Institute NPSH ‘Committee and ASME/ANSI B73 Commiree, Mechanical Seals, ABSTRACT ‘There isa mistaken belief inthe pump industry Fueled by many’ publications) that the best way t0 avoid suction recirculation problems isto design and/or apply only pumps having low suction specific speed values (below 8,500 1 11,000). Attemptsare made to correc this overly simplified treatment of « complex problem, which if used alone, can still lead to field problems, or, unneces” sary and expensive over design. A more accurate, but tll simple, ‘method is presented as an alternative for identifying pumps which ay be susceptible o suetion recirculation problems (damage) during reduced capacity operation. Pump type, impeller inlet tip speed, impeller vane overlap and fluid specific gravity have been added 10 pump suction specific speed to greatly improve the prediction process, INTRODUCTION There are many published articles which infer a direct tink between suetion specific speed and suction recirculation damage, with litle or na recognition of other equally important factors. Suction specific speed is defined as: nid (NpstiR)™ (@ BEP Capacity & Max. Impeller Dia. (US Units)) Hallam [1 states that “centrifugal pumps with suction specific speed (5) greater than 11,000 failed at a frequency nearly twice that of centrifugal pumps with suction specifie speed ess than 11,000." He blames thishigher failure rate on“A centrifugal pump impeller with a high S, as a large impeller eye or an inlet vane angle, B such that the impellers susceptible to inlet eye recireu- lation.” Lobanoff and Ross [2] state that one of the steps to avoid cavitation ina centrifugal pump is “not (to) select a pump with 2 nati, Ohio suction specific speed above 10,000." Fraser [3], tates that “There is no question that many pump installations are operating today either continuously or intermittently in suction or discharge recie- culation. This is especially true for pumps designed for high suction specific speeds. The history of these pumps will show a pattern of cavitation damage, noise, rotor oscillation, shaft break- Age or surging in varying degrees depending on the pump design and application. Many oF these problems can be avoided by ‘designing the pump for lower suction specific speed values a limiting the range of operation to capacities above the point of recitculation.” The 14th edition of the “Hydraulic Institute Stan: dards” (4) states that “Increased pump speeds without proper suction conditions can result in abnormal wear and possible failure from excessive vibration, noise, and cavitation damage. Suction Specific Speed Available, SA, has been found fo be a valuable criterion in determining the maximum permissible speed. The curves presented in this standard are based on SA of 8500; this represents a practical value" Istates that "Suction specifi speed ‘equited, 8, must equal or exceed the suction specific speed available, SA, to prevent cavitation.” Suction specific speed alone has not, however, been able to explain why pumps with suction specific speed values as ow as 6,900 have experienced recirculation damage, while pumps wi sition specific speeds as high as 18,000 and higher can oper over their entite flow range without any detrimental effects. The preceding publications are correct to the extent thatthe suetion Specific speed of a pump is one factor in determining whether or nota pump will experience suction recirculation damage, but itis not the only factor. Budrs[5] lists four other (equally important) factors which, when prop:rly combined with the pump suction specific speed, have proven to beable to predict suction recireu tion damage with a higher degree of accuracy. The additional factors are the pump type, determined by the number of right angle turn the liquid must makeat the inlet the pump, the tip speed of | the impeller inlet, the spesifie gravity of the fluid pumped, an existence of impeller vane overlap. Even though there isa certain relationship between the suction specific speed of a pump and the sizeof the impeller inlet, s suggested by Hallam, thete are other design methods for improving the NPSH, of a pump than by just increasing the impeller eye diameter, softs only @ loose te in. Further, the suction speifie speed ofa specie pump remains {generally constant, regardless ofthe speed of operation, while the ‘impeller inet vane tip velocity changes directly with pump speed. ‘Thisisespecially importa intoday’sageof variable speed drives. ‘The impeller eye diameter (tip speed) must, therefore, independent factor from suction specific speed in determi likelihood of damage from suction recirculation. For example, & particular pump with a sucion specifie speed of, say, 12,000 may not experience any suetion recirculation problems at fow flowrates \when operated at 1200 rpmor 1800 rpm, However, when this same pump is operated above 3000 rpm it may begin to experience Suction recirculation damage, even though the suction specific speed has basically not changed, 92 PROCEEDINGS OF THE TENTH INTERNATIONAL PUMP USERS SYMPOSIUM. PUMP TYPE ‘Therelativesusceptbilty ofa centrifugal p uetion recirculation at low flow operation is affected by, among, ‘other things, the numberof right angle warns the li inthe inlet ofthe pump. Thus, an axial flow impeller and inducer (Figure 1), which has no turns at the vane inlet has the best performance. An end suction radial flow impeller (Figure 2), Which has one right angle cur as the liquid is picked up by the ‘vanes, issecond best. Finally, a radial suetion, radial low impeller (Figure 3) is the worst performer, as it has two right angle tutns, fone in front ofthe impeller, and one asthe liquid is picked up by the vanes. The effect of pump type isshown in Figure 45] forthe susceptibility ofa pump to damage from suctio reduced capacities, Above the upper suetion recirculation factor (SRE) lines, many pumps ean be expected to experience problems ifallowed to operate at or below the start of suction reitcalation ‘without providing an adequate NPSH safety margin, while below the lower SRF curves, minimum low restrictions are not required, {except to prevent thermal build up. Between the “upper limit of ne ‘pump problems" and he “Towser limit of potential pump problems” ‘sa gray (shaded) area where damage cannot be accurately prec te due to the many influences not included in the SRF method, ‘Mean lines, based on plots of approximately 200 actual pumps (as gutes 6, 7, and 8),arealso presented foreach pumptype in Figure 4. Its interesting to note thatthe mean radial stetion| pump inecrossesthe lower suetionrececulaton Fator (int ine forradial suction pumpsat a suetion specific speed of about 9,000, lose to the Hydraulic Institut 8,500 recommended timit, Further, the mean end Suction pump line crosses the lower suction recireu lation factor fine forerd suction pumps at an $value of 12,000, not far from Hallam’s 11,000 start of increased field problems with APLype pumps. Most pumps in a refinery are of the end suction type. Figure 3, Radial Suction Impeller Figure 2. End Suction Impeller. Figure 4, Suction Recirculation Minimum Flow Restriction Mapto Avoid Damage TING VORTEX Figure 5. Suction Recirculation.‘THE SHORTCOMINGS OF USING PUMP SUCTION SPECIFIC SPEED ALONE TO AVOID SUCTION RECIRCULATION PROBLEMS 93 Figure 6, End Suction Inducer, Suction Recirculation Damage/ Vibration Guide Line Figure 7. End Suction Impeller, Suction Recirculation Damage/ Vibration Guide Line. nl ener a7 aa ala Figure 8 Radial Suction Impeller, Suction Recirculation Damage/ Vibration Guide Line. Not all pumps fall into one of these three basie pump types, however. A vertical in Hine or elf priming pump has aright angle ‘umn front ofthe impeller, ike a radial suetion pump, but des not havea shaft through the eye. The performance ofthese in-between ‘pump types would be expected to fall somewhere between that of the two basie types, depending on how sharp the inlet turns really INLET TIP SPEED Within any given pumptype, the inlet tip speed ofthe impeller vanes isthe most ertical single factor which determines revireu- lation intensity ina pump. the tip speed (energy imparted 0 the Tiquid) isto0 high in relation tothe through flow velocity momen tum (energy) ofthe approcching liquid, the liquid in contact with ‘he outer portion of the ine: vanes wil be thrown back intothe inlet ‘of the pump (Suction recir-ulation), Figure 5. Many publications ‘substantiate this basie fae’, The point of contention is not that a large impeller eye will ease suction recirculation and possibly ‘damage, but that tip speed is independent, toa large degree, of suction specie speed. Ill pumps Fell on the mean punp lines shown in Figure 4, then inlet tip speed would not have to be ‘considered separately. However, as shown in Figures 6, 7, and 8, hot all pumps fll on these mean lines. There is wide spread in performance, due in part tothe fact thata given pump, witha given ‘Suetionspecifie speed, canoperate atmore than one speed inet ip velocity). Further, such factors as vane blockage, inlet vane angle (slow angle), and infet throat area can lsostect the NPSH, ul suction specie speed ofa pump, without affecting the size othe inet eve or ier tip speed. For a given value of suetion specific speed, the spread in tip speed is such thatthe highest tip speed is proximately dauble or greater than the lowest value. For exat ple, given 10,000 suction specific speed radial suction impeller pump (Figure 8) witha low inlet tipspeed, say 40 fysee,one Would hot expect to experience any significant damage from suction recitculation, On the otherhand, given hada second radial suetion impeller pump, also witha suction specific speed of 10,000, but with an inet tip speed of 80 see, there would be a high probabil ity of experiencing problems Trom suction recieulation at low capacities with this second pump. Schiavello [6] shows several «examples of eavtation/suciion recirculation field problems, wher {the solution was actually an increase of the purnp suction specific speed. He accomplished tis with better matching of the impeller inlet vane angle with that of the incoming flow, while actually reicing the inlet tip spel VANE OVERLAP, For most impeller designs, the inlet pesformance is unaffected by the discharge ofthe impeller, except when there is very ‘no vane overlap, such as ina two vane solids handling impel an impeller witha severe impeller cut down (trim), Figure 9. With this ype of design and under low flow conditions, portion ofthe impeller discharge flows back between the vanes 10 the suction, casing suction recirculation. This explains why end suetion 1we ‘vane impellers are prone to suction recigeulation damage, at 3 ‘much lower suction recircelation factor value, as shown in Figure 7. than more typical end suetion impeller designs, SPECIFIC GRAVITY Given the fact that virtually all centrifugal pumps will expeti- ‘ence suction recirculation t reduced capacities (normally inthe 50 percent 10 85 percent of best efficiency capacity range), the likelihood of damage from suction reciteulation becomes matter ‘ofenergy level atthe suction ofthe pump. Asalready indicated, the tipspeed ofthe leading edge ofthe impeller vanes isa measure of the inlet energy. Further te numberof turns the liquid must make atthe inlet adds turbulence (energy), andthe lack of vane overlap allows the high energy cischarge liquid to enter the suction.96 PROCEEDINGS OF THE TENTH INTERNATIONAL PUMP USERS SYMPOSIUM