Centrifugal Pumps

The nearly inexhaustible variety of flow phenomena - from the flow through blood vessels, the flow in centrifugal pumps to global weather events is based on a few basic physical laws only. In this chapter these will be briefly reviewed and their general nature illuminated. Emphasis will be on the phenomena which are of special interest and significance to the pump engineer. Basic knowledge of the terms of fluid dynamics is taken for granted. There is a wealth of textbooks and handbooks on fluid dynamics, only a few can be quoted: [1.1 to 1.7, 1.14].

Centrifugal pumps are turbomachines used for transporting liquids by raising a specified volume flow to a specified pressure level. The energy transfer in turbomachines is invariably based on hydrodynamic processes for which characteristically all pressure and energy differences are proportional to the square of the circumferential rotor speed. By contrast, positive displacement pumps (e.g. piston pumps) essentially deliver the same volume Vstroke at each stroke independently of flow velocity or rotor speed n. The flow rate then becomes Q = nxV_stroke; the pressure rise results solely from the imposed back pressure.

This chapter deals with the calculation methods essentially common to all impellers and diffusing elements regardless of the specific type. The details of calculation and design of the various types of impellers and collectors are discussed in Chap. 7.

Following the changing operational requirements, practically all pumps temporarily operate away from the design point which is defined by q* ≡ Q/Qopt = 1. Overload corresponds to q* > 1, while operation at q* < 1 is called “partload”. The pump characteristics describe the behavior of head, power consumption and efficiency as functions of the flow rate (the behavior of NPSH = f(Q) is discussed in Chap. 6). The shape of these performance curves over the range from shut-off (or zero flow Q = 0) to the maximum possible flow rate is important for the operational behavior of the pump in the plant – for instance when operating in parallel or during start-up (see Chap.11). The majority of applications require a Q-H-curve steadily falling with increasing flow rate, i.e. H/∂Q < 0. This is termed a “stable characteristic”. In contrast, if the Q-H-curve has a range with ∂H/∂Q > 0, the characteristic is said to be “unstable”.

A pump working significantly below the best efficiency flow rate is said to operate at partload. At low specific speeds this can be roughly assumed at q* < 0.8, at high nq below q* < 0.9. Since blade inlet angles and channel cross sections are too large for the reduced flow rate, flow patterns during partload operation fundamentally change compared with the design point. The flow becomes highly 3-dimensional since it separates in the impeller and the collector. Finally, recirculationsare observed at impeller inlet and outlet at sufficiently low flow. An easymeans to obtain information on the impeller flow are stroboscopic observations oftufts. Flow patterns in a radial impeller of nq = 22 gained in this way are shown inFig. 5.1 [B.20]. It can be seen that the flow is attached at q* > 0.8 while increasinglylarge zones with separation and recirculation are observed at a lower flow rates. Similar flow patterns were found on impellers of nq = 26 and 33.

This chapter deals with one-dimensional calculation procedures and design methods for impellers, volute casings, diffusers and inlet casings. For developing these components, the main dimensions and blade angles are calculated in a first step. Subsequently, the hydraulic contours are designed based on certain rules and methods. Many pump manufacturers employ computer programs for this work, and the drawings are generated on 2D-CAD systems. However, these methods are also increasingly replaced by 3D-CAD systems with which fully three-dimensional geometry models of a component can be created, Fig. 2.2A and Fig. 7.45. The complex hydraulic channels can be evaluated better with such models than with the conventional two-dimensional representations in various sections and views. Even more importantly, 3D-CAD systems provide the capability to directly manufacture the hydraulic components (or the casting patterns) by NC milling, stereo lithography or other fast-prototyping processes, [7.1]. The advantages of such processes are evident in terms of geometrical accuracy and lead times (not in the least also for model tests with milled or stereo-lithographed components). Since manual designs on the drawing board are rather the exception, the subsequent discussion of the design methods emphasizes the fundamental aspects of the design processes rather than a very detailed description of geometrical operations.

Real flows are described by partial differential equations which cannot be solved analytically in the general case. By dividing a complex flow domain into a multitude of small cells, these equations can be solved in an approximate manner by numerical methods. Because of their wide range of application, numerical flow calculations (“computational fluid dynamics” or “CFD” for short) have become a special discipline of fluid dynamics.

As explained in Chap. 5, the flow at the impeller outlet is non-uniform. The diffuser vanes or volute cutwaters are thus approached by an unsteady flow. The flow at the stator vanes acts back on the velocity field in the impeller. The related phenomena are called “rotor/stator interaction” (RSI). As a consequence of the RSI, hydraulic excitation forces are generated. These give rise to pressure pulsations, mechanical vibrations and alternating stresses in various pump components. The vibrations transmitted to the foundations spread as solid-borne noise throughout the building. The pressure pulsations excite the pump casing to vibrations. They travel as fluid-borne noise through the piping system, where they generate vibrations of the pipe walls. The vibrating walls and structures radiate air-borne noise.

A positive displacement pump delivers at a fixed speed a nearly constant flow rate independently of the back pressure. In contrast, the flow rate of a centrifugal pump depends on the pressure difference Δp = ρ×g×HA imposed by the system on the pump. The pressure rise Δp generally depends on the flow rate because of hydraulic losses. Thus the system characteristic HA = f(Q) is understood as the difference in total pressure which must be supplied by the pump to maintain a specific flow rate through the system, Eq. (T 2.2.6). The operation point of a centrifugal pump is given by the intersection of the characteristics of system asnd pump, Fig. 11.1.

Reverse-running centrifugal pumps can be used as turbines for energy recuperation (in [N.6] termed “Hydraulic Power Recovery Turbines”, HPRT). Applications are processes where a large amount of fluid energy is dissipated in valves or other throttling devices. In some processes dissolved gases separate from the fluid or liquid is flashed into steam during the expansion. Higher energy differences are then available for power recovery than in pure liquid flow, Chap. 13.3.

When a centrifugal pump is used for transporting a fluid with a viscosity much higher than cold water, additional losses impair the performance. Therefore, the pump characteristics determined for water must not be applied without correction to pumping highly viscous fluids as encountered for example in the oil and process industry.

Degradation or failures of materials due to fatigue, corrosion, abrasion and cavitation erosion time and again cause costly problems to pump operators. This could be avoided in most cases by a careful material selection. Frequently one of two causes can be made responsible for a wrong material selection: (1) the corrosive properties of the liquid pumped are not clearly specified (or unknown), or (2) for cost reasons (competitive pressure) the least expensive material is chosen which appears just to be able to do the job.

Operational problems can often be traced back to a non-optimum - or even downright wrong - pump selection. An unsuitable selection may result from insufficient knowledge of pump operation and installation conditions or from not giving them careful consideration and analysis.