8.34

Pump Controls F. B. HOROWITZ

(1970)

B. G. LIPTÁK

(1985, 1995, 2005)

S. BAIN

Centrifugal

(2005)

Rotary

Reciprocating

Flow sheet symbols

INTRODUCTION This section describes the basic operation and controls of pumps and pumping stations, while the next section (8.35) concentrates on the optimization of the unit operation of pumping. This section is divided into three main parts. The first provides a brief discussion of the processes into which the fluid is being transported by the pumps. The second part describes the types of pump designs, including centrifugal, rotary, and positive displacement designs and their basic methods of controls. The third part discusses some aspects of pumping system commissioning and operation. The general discussion in this section is somewhat abbreviated, because related topics are also covered elsewhere in the handbook. Pumps, pumping stations, and metering pumps are also discussed in Section 7.4, and variable-speed drives in Section 7.10 in Chapter 7 of this volume. In addition, metering pumps are also covered in Section 2.14 in Chapter 2 in the first volume of this handbook. Some of the pumping system-related terms, abbreviations, and conversion factors are described in Table 8.34a.

TABLE 8.34a Pump Terms, Abbreviations, and Conversion Factors* Term

Abbreviation

Multiply

By

To Obtain

Length

L

ft

0.3048

m

Area

A

ft

2

0.0929

m

Velocity

v

ft/s

0.3048

m/s

Volume

V

ft

3

0.0283

m

Flow rate

Qv

gpm

0.2272

m /h

gpm

0.0631

l/s

psi

6890

Pa

psi

6.89

kPa

psi

0.069

bar

ft

0.3048

m

ft

0.3048

m

0.7457

kW

Pressure

P

Head (total)

H

NPSH

H

Output power (pump)

Po

water hp (whp)

2

3 3

Shaft power

Ps

bhp

0.7457

kW

Input power (driver)

Pi

kW

1.0

kW

Pump

Ep

—

—

—

Equipment

Ee

—

—

—

Efficiencies (%)

THE PROCESS A pump is a liquid transportation device that must develop enough pressure to overcome the hydrostatic and frictional resistance of the process as it delivers the required fluid. These resistance components are unique characteristics of the process served and can be described by system curves. The system curve of a process relates the pressure (head) required and the amount of fluid flow that is being delivered. System Curves The characteristics of the system that is served by a pump or pumping station can be represented by a head-capacity

© 2006 by Béla Lipták

Electric motor

Em

—

—

—

Utilization

Eu

—

—

—

Variable-speed drive

Ev

—

—

—

System efficiency index (decimal)

SEI

—

—

—

Speed

N or ω

0.1047

rad/s

Density

ρ

16.04

kg/m

Temperature * From Reference 1.

rpm lb/ft °F

3

—

°C

3

8.34 Pump Controls

System head curve

Head

Friction losses

Total static head

Capacity

FIG. 8.34b The system head curve is the sum of the static head and the friction losses that have to be overcome in order to pump liquid into the process.

system curve (Figure 8.34b). The head at any one flow capacity is the sum of the static and the friction heads. The static head does not vary with flow rate, as it is only a function of the elevation or back-pressure against which the pump is operating. The friction losses are related to the square of flow and represent the resistance to the flow caused by pipe and equipment friction. A system curve tends to be flat when the piping is oversized and steep when the pipe headers are undersized. The friction losses also increase with the age of the plant. Therefore, the system curve for old piping tends to be steeper than for new piping. A generalized equation describing the system curve of a process is given below: P = H + Ff (Q ) x

8.34(1)

where P is the head pressure required to pump liquid into the process H is the static or elevation head of the process Ff is the friction factor of the process Q is the flow rate of the incompressible fluid x is an exponent that varies between 1.7 and 2.0, usually 2.0 is used Static and Friction Pressures Constant static head (H) is the difference in pressure between the pump’s intake and its discharge at zero flow, if the discharge piping is full. Usually it is measured between the pump’s intake and the discharge side of its check valve, thereby correcting the measurement for any elevation differences between the two. The constant static head can be the difference in elevation between the piping system’s intake and discharge (corrected for liquid density if necessary). If the pump is discharging

© 2006 by Béla Lipták

2085

into a pressurized system such as a potable water distribution network or boiler, it will be the difference between the pump intake pressure and the system’s pressure. In any case, the constant discharge head is the pressure the pump works against when the piping system is full and pressurized, but there is no flow through the pumps’ discharge pipes. As its name implies, it is comparatively constant. However, it may change, if for example the pump takes suction from a well and the well level drops, or if pumping to a water storage tower and the water level in the tower rises. Therefore, the constant discharge head is constant only in relation to variations in one process variable: flow. If other process variables (pressure, temperature, density, level) change, it will be affected; it is not constant. The friction pressure component in Equation 8.34(1) is the pressure that is lost due to friction between the liquid and the pipe. It includes losses from turbulence in bends and in the conversion of velocity pressure to static pressure in pipe expansions. However, for practical purposes they are combined into a single term. The value of exponent x in Equation 8.24(1) is not critical. However, the controls can be improved by an accurate knowledge of the system curve, so the recommendation is to measure the value of x during commissioning. Types of System Curves Figure 8.34c illustrates the system curves of three different types of processes. Curve 1 corresponds to the closed-loop circulation of a fluid in a horizontal plane. Here, there is no static head component at all, and the parabola that describes the system starts at zero. Curve 2 is the system curve for a condenser water circulation network. Here, a limited amount of static head is present, because the pump must return the water to the top of the cooling tower. This curve also illustrates that the friction losses tend to increase when material builds up on the inside of the pipe, because it is no longer new. Curve 3 gives an example of a process dominated by static head. This is the case when feedwater is being pumped into a boiler drum. This curve is flat and is relatively insensitive to changes in system flow. As will be discussed later in more detail, when the system curve is flat, there is little advantage to variable- or multiple-speed pumping, and the usual response to system flow variations is the stopping and starting of parallel pumps. Inversely, if the system curve is steep, substantial energy savings can be obtained from the use of booster, multiple-, or variable-speed pumps. Open and Loop-Type Systems Hydraulic systems can be open (noncirculating) or loop-type, as illustrated in Figure 8.34d. Water supply and distribution systems in cities and buildings are typically open systems, whereas hot-and-chilled-water heating and cooling systems of plants are typically loop-type systems.

2086

Control and Optimization of Unit Operations

System head

Storage tank

Curve #1 (Horizontal liquid circulation) Water distribution system to residences

Pump

Open system Pump

Heat exchanger

System flow

Control Heating or cooling coils

Old piping

System head

Friction head of condenser and piping Curve #2 (Cooling tower water)

Valves

Loop type system

New piping

Static rise to top of cooling tower System flow Pipe friction

FIG. 8.34d Pumping systems can be of the open or the loop type.

In actual systems, a single-system head curve may not exist. As illustrated in Figure 8.34e, what often exists is a system head band. This is because the distribution of active loads shifts the system curve within a wide band, as this figure shows.

System head curve

System head

10 ft of pipe friction between each load Heat exchanger

Boiler pressure

P

Curve #3 (Boiler)

100 100 100 100 100 100 100 100 100 100 GPM GPM GPM GPM GPM GPM GPM GPM GPM GPM

100 Active loads remote from pump

90 80

System flow System head, FT

FIG. 8.34c System curves vary as the static-head components change in various processes. (Adapted from Reference 1.)

Uniform loading

70 60 50 40 30 20

Hydraulic systems must also be evaluated as to whether their flow is restricted or unrestricted. Restricted-flow systems are those that include valves that regulate the flow through the system. Hot-and-chilled-water systems, for example, are restricted-flow systems, because manual or automatic valves control their flow. Unrestricted-flow systems include sewage and stormwater lift stations as well as the pumping of municipal water into elevated storage tanks.

© 2006 by Béla Lipták

Active loads near pump

10 0

0

200

400

600

800

1000

System flow, GPM

FIG. 8.34e The system head drops when active loads are near the pump and it 4 rises when the most remote loads are active.

8.34 Pump Controls

PUMP TYPES There are many types of pumps; however, almost all flow is pumped by just two types: centrifugal pumps and reciprocating positive displacement (PD) pumps. As a result, this section will focus on these two types, and particularly on centrifugal pumps, which probably pump more flow than all other types combined. Displacement and Centrifugal Designs The two main pump designs are the positive displacement and centrifugal pumps. Vertical centrifugal units can pump water from depths up to 2000 ft (600 m), and horizontal units can transport process fluids from clear water to heavy sludge at 3 rates up to 100,000 gpm (6.3 m /s). The centrifugal designs are of either the radial-flow or the axial-flow type. Liquid enters the radial-flow designs in the center of the impeller and is thrown out by the centrifugal force into a spiral bowl. A number of impeller designs are illustrated in Figure 7.4b in Chapter 7. The axial-flow propeller pumps are designed to push rather than throw the fluid upward. Mixedflow designs are a combination of the two. In positive displacement pumps, a piston or plunger inside a cylinder is the driving element as it moves in reciprocating motion (Figures 7.4n and 7.4o). The stroke length and, thus, the volume delivered per stroke is adjustable within a 10:1 range. Rangeability can be increased to 100:1 by the addition of a variable-speed drive. The plunger designs are capable of generating higher discharge pressures than the diaphragm types, because of the strength limitation of the diaphragm. The strain on the diaphragm is reduced if it is not attached directly to the plunger but is driven indirectly through the use of a hydraulic fluid. Because solids will still settle in the pump cavities, these designs are all limited to relatively clean services. For slurry service, the hose-type design is recommended. This design eliminates all the cavities, although the seating of the valves can still be a problem.

2087

Peristaltic pumps handle similar liquids as do the progressing cavity designs but are generally more economical in small (ml/s) sizes. The process industries use peristaltic pumps in real-time wet chemistry analyzers that depend on accurately metering small flows. Gear, lobe, and vane pumps deliver a relatively constant flow at constant speed with large changes in discharge pressure and, therefore, approximate the characteristics of positive displacement pumps. These pumps cover the viscosity range from less than 1 centipoise up to 500,000 centipoises. The usual application of this type of pump is for viscous liquids and slurries that are beyond the capabilities of centrifugal pumps. Rotary screw pumps are ideally suited for sludge and slurry services, and can safely pump high solids including live fish. A rotary screw pump design is also in development as an artificial heart. Rotary screw pumps behave similarly to centrifugal pumps. In lift pumps, compressed gas, usually air, is blown into the bottom of a submerged updraft tube. The gas bubbles reduce the average density and, therefore, the hydrostatic head inside the tube. As the head is lower on the inside of the tube, the manometer effect induces the surrounding fluid to enter the updraft tube. This is a maintenance-free means of lifting large volumes of slurries over low elevations but is comparatively inefficient if one considers the compressor power required. The Archimedes Screw is the earliest true pump and is still in use. An advantage of a screw is that it naturally changes its capacity to accommodate changing liquid levels without the complexity of a control system. Air pumps use compressed air or steam to displace accumulated liquids from tanks (Figure 8.6llll). Diffuser micropumps are generally specialized for use in small-scale work. Pumps specialized for high-vacuum work, such as diffusion and cryogenic pumps.

CENTRIFUGAL PUMPS Pump Design Variations If one attempted to give a more-or-less complete list of pump designs, the list would include the following: Progressing cavity pumps are suitable for pumping liquids with entrained solids that would jam or abrade normal pump components, and liquids with low-shear requirements. The process industries use progressing cavity pumps for chemicals that may crystallize, polymers that need low-shear pumping, and liquids with some entrained solids. Progressing cavity pumps depend on the process liquid for lubrication, and they are easily damaged by low suction pressure. They behave like positive displacement pumps but are considerably less accurate than reciprocating pumps, and their accuracy deteriorates with use, to the 2–5% level.

© 2006 by Béla Lipták

Some centrifugal pumps use centrifugal force to throw liquid radially outward while others, such as propellers, use a screw-type action that results in axial flow. Between these two extremes, there is a whole continuum of impellers that change their pumping action from highly centrifugal radial to axial flow. The line separating centrifugal and axial-flow pumps is vague, and the behavior of these pumps is usually described by the same laws. Figure 7.4b in Section 7.4 describes a number of impeller designs, including both radial and axial-flow types. Common characteristics of centrifugal pumps are high efficiency (over 90% in case of large pumps); they have only one moving part (the impeller with bearings); they deliver smooth, steady flow; they have a rangeability of about 4:1; and they are

Control and Optimization of Unit Operations

20 (6) 15 (4.5) 10 (3)

Figure 8.34f illustrates the typical pump curves of a single impeller pump. Efficiency The typical range of pump efficiencies is from 60 to 85%. Pump efficiency is the ratio of the useful output power of the pump to its input power. Using the symbols defined in Table 8.34a, it is calculated (in both SI and US 2 units) as follows: pump output SPQH t (SI units) Ep = = Pi Pi Ep =

Efficiency

50 40 Total head 30

NPSHR

Brake horsepower

300 (0.019)

8.34(2)

SPQH t pump output = (U.S. customary units) bhp bhp × 550 8.34(3)

400 (0.025)

500 (0.032)

600 (0.038)

700 (0.044)

Capacity, GPM (m3/s)

FIG. 8.34f Typical characteristic curves of a single-impeller centrifugal pump.

Formulas Table 8.34g provides a summary of the more common formulas that can be used in connection with pumping calculations.

TABLE 8.34g 1 Common Formulas Used in Connection with Pumping Calculation Formula for

Conventional Units

SI Units

Head

H = psi × 2.31/SG* (ft)

H = kPa = 9.8023/SG* (m)

Output power

Po = Qv × H × SG*/3960 (hp)

Po = Qv × H × SG*/367 (kW)

Input power Equipment efficiency, %

Ps =

Qv × H × SG * (hp) 39.6 × E p

Pi – Ps × 74.6/Em (kW) (Constant speed pumps) (Variable speed pumps)

Ps =

Qv × H × SG * ( kW) 3.67 × E p

Pi = Ps × 100/Em (kW) −2 Ee = Ep × Em × 10 −4 Ee = Ep × Em × Ev × 10

QD = design flow Utilization efficiency, % System Efficiency Index [see Eq. 8.34 (4)] *SG = specific gravity

© 2006 by Béla Lipták

15 (11.1) 10 (7.5) 5 (3.7)

bhp = brake horsepower 550 = conversion factor for horsepower to ft lbf /s

where Ep = pump efficiency, dimensionless Pi = power input, kW (kN·m/s) 3 3 SP = specific weight of water, lb/ft (kN/m ) 3 3 Q = capacity, ft /s (m /s). Ht = total dynamic head, ft (m)

Shaft power

60

Efficiency (%)

70 (21) 60 (18) 50 (15) 40 (12) 30 (9)

Brake horsepower (kW)

Pump Curves

8" (200 mm) impeller 6" (150 mm) suction 5" (125 mm) discharge 70

1750 RPM

Total head, ft (m)

relatively insensitive to air-locking, but are susceptible to cavitation. Vertical centrifugal pumps can lift water from depths of up to 2000 ft (600 m) and horizontal pumps can transport process fluids from clear water to heavy sludge at rates up to 2 100,000 gpm (6.6 m /s). The centrifugal pump is the most common type of process pump, but its application is limited to liquids with viscosities under 3000 centistokes.

NPSH

2088

QA = actual flow HD = design head HA = actual head

Eu =

QD × H D × 100 QA × H A

SEI = Ee × Eu × 10

−6

8.34 Pump Controls

Pump head-capacity curve

Drooping (curve no. 1) 80 (24) 60 (18) Normal (curve no. 3)

40 (12)

System head-capacity curve

Flat (curve no. 2) Head

Pump discharge pressure (in feet or m of head)

2089

20 (6) 0

20 (1.3)

40 (2.5)

60 (3.8)

80 (5.0)

100 (6.3)

Friction and minor losses

120 (7.6)

Flow, GPM (l/s)

Total static head

FIG. 8.34h The centrifugal pump’s curve can be drooping (1), flat (2), or normal (3).

Matching the Pump(s) to the Process Figure 8.34i shows both the head-capacity curve of a centrifugal pump and the system curve of a process. When such a system is uncontrolled, the operating point of the system will be the point at which the pump and system curves cross each other. If the process flow is controlled, a new system curve has to be artificially generated. This can be done 1) by generating a new system curve through the introduction of extra pressure drop in a control valve, or 2) by changing the pump curve through changing the pump speed. Figure 8.34j illustrates the case where a particular process flow is established by throttling a valve and, thereby, changing the unthrottled (solid) system curve into a throttled system curve (dotted).

© 2006 by Béla Lipták

FIG. 8.34i The system curve crosses the pump curve at the operating point.

Throttling with a control valve makes the apparent pump curve steeper, so that it will cross the system curve at the desired flow (operating point). This modification occurs at the cost of introducing an artificial pressure drop of (Hp − Hs), which burns up pumping energy and, therefore, reduces operating efficiency. In addition to that energy waste, the pump will also operate in a less-efficient region, as shown in Figure 8.34k by the throttled operating point #1 (72%) and the unthrottled operating point #2 (80%). Adjusting the Pump Speed The affinity laws describe the relationships among changes in speed, impeller diameter, and specific gravity. With a given impeller diameter and specific gravity, pump flow is linearly proportional to pump speed, pump discharge head relates to

Typical throttled operating point

Pump curve Head or pressure

Characteristic Pump Curves The head-capacity curve is the operating line for the pump at constant speed and impeller diameter. The characteristic curve of a pump describes the variation of its discharge pressure with volumetric flow. The discharge pressure is the total of the velocity and static pressures. Three types of head-capacity curves are shown in Figure 8.34h, illustrating various relationships of capacity to discharge pressure. The capacity varies widely with changes in discharge pressure for all curves, but the shape of the curve determines the type of control that may be applied. The shape of the head-capacity curve is an important consideration in pump selection. Curve 1 is referred as a drooping curve, curve 2 is called a flat curve, and curve 3 would be considered normal. For on/off switching control, curves 1 and 2 are satisfactory as long as the flow is above 100 gpm (6.3 l/s). Below this flow rate, curve 1 allows for two flows to correspond to the same head, and curve 2 may drop to zero flow to obtain a small head increase. Both are, therefore, unstable in this region. Curve 3 is stable for all flows and is best suited for throttling service in cases in which a wide range of flows is desired.

Capacity

1

sy

m ste

ed ttl ed ro ottl thr Th Un

Natural Operating point 2

tem s ys

Hp

3 Hs Flow

FIG. 8.34j The addition of a control valve allows the control of flow at the cost of added pressure drop (wasted energy).

Th

ed rottl Unth

2

form another characteristic surface (surface B). This then illustrates how the intersection of surfaces A and B is the operating line on which the variable-speed pump operates. (Section 8.35 will discuss the mathematical definition of the characteristic operating surfaces of pumps.) Flow control via pump speed adjustment is less common than the use of throttling with valves, because most AC electric motors are constant-speed devices. If a turbine drive is considered, speed control is even more convenient. However, the advent of the pulse-width modulated (PWM) adjustablespeed drive with sensorless flux-vector control has brought adjustable-speed pumping into the mainstream of everyday applications. In order to vary pump speeds with electric motors, one of the variable-speed drives described in Section 7.10 in Chapter 7 should be used. The efficiency curves (wire-toshaft efficiencies) of the various variable-speed drives that are used on centrifugal pumps are shown in Figure 7.10z. Variation of the pump speed generates a family of headcapacity curves, as shown in Figure 8.34m. If the impeller diameter is constant, the volumetric flow through the pump is proportional to its speed, and at reduced speeds, family of speed curves determines the flow rate (points 1, 2, or 3).

80 78 75

60 65 70

ed ttl ro

1

80

Head or pressure

100% speed

78

Control and Optimization of Unit Operations

75

2090

3

Flow

FIG. 8.34k The throttled system not only wastes pumping energy through valve pressure drop but also operates at a less efficient point on the pump curve.

(approximately) the square of pump speed, and pump power consumption is proportional to the cube of pump speed. As shown in Figure 8.34l, one can plot the system curves and the variable speed pump curves on a three-dimensional plot (a-pressure, b-flow, c-speed). On such a plot the system curves form one surface (surface A), and the pump curves

10

Pressure

8 6 Pump surface (B)

4 2 0

00Speed 0

0.2 0.4 0.6 0.8 Flow Flow vs. pressure

1

System surface (A)

10

Pressure

8 6 4 10

2 0

00Flow 0

10

0.2 0.4 0.6 0.8 Speed Speed vs. pressure

1 8

0.8 0.7

6 4 2 0

0.6 Speed (c)

Pressure (a)

0

6 5 4 0.2 0.4 0.6 0.8 1 3 00 Speed 2 Speed vs. Flow pressure 1 0

0.5 0.4

4

pressure

Flow

8

0 0 2 2 0 0. .4 0.6 0.4 0. 0 1 0.8 0.6 Flow 0.8 1

0

2

0

0.2 0.6

0.8

0.4 ) Flow (b

The system curve

FIG. 8.34l The variable-speed pump operates on the line where the surface formed by the system curves (A) intersects with the surface formed by the pump curves (B).

© 2006 by Béla Lipták

Pump discharge pressure

8.34 Pump Controls

1,575 1,400

Pump curves System head curves

Head (H)

1,750

rpm

2091

Pump “B”

rpm (1)

rpm

Zone 4

System curve

(2) (3)

Zone 3

Pump “A”

Pump curves

Zone 2

Flow

FIG. 8.34m Variable-speed pump operation can be described by a family of head-capacity curves.

Because the area of peak pump efficiency falls on a parabolic path, speed throttling will usually not reduce the pump efficiency as much as valve throttling (Figure 8.34k). This increases the total energy savings obtained from pump speed control. As shown in Figure 8.34n, when the flow is reduced from F1 to F2, instead of wasting the excess pump head of (P1 − P2) in pressure drop through a valve, that pump head is not introduced in the first place. Thereby, speed throttling saves the energy that valve throttling would have wasted. When to Use Variable-Speed Pumps The shape of the system curve determines the saving potentials of using variablespeed pumps. As has been discussed, all system head curves 2 are parabolas (H ∼ Q ), but they differ in the steepness of these curves and in the ratio of static head to friction drop. As shown in Figure 8.34o, the value of variable-speed pumping increases as the system head curve becomes steeper. Studies indicate that in mostly friction systems (such as zone 4 in Figure 8.34o), the savings represented by variable7 speed pumping will increase with reduced pump loading. If, on the yearly average, the pumping system operates at not more than 80% of design capacity, the installation of variable-

ed

P1

67% speed

Flow (Q) Ht/Hs < 1.2 < 1.5 < 2.0 > 2.0

Zone 1 2 3 4

led ott r h t Un

P2

F2

F1

Pump sizes Same Various Same Various

Pump drives Constant Constant One variable All variable

FIG. 8.34o Pumps and drives should be selected as a function of the steepness of the system curve.

speed pumps can result in a payback period of approximately 3 years. The zones in Figure 8.34o are defined by the Ht /Hs ratio. The higher this number, the higher the zone number and the more justifiable is the use of variable-speed pumps. Figure 8.34p illustrates how the Ht /Hs ratio is calculated. The shaded areas identify the energy-saving potentials of variable-speed pumps. The values of Ht and Hs are identified on the basis of the average yearly flow rate (Fa) and determining its intersections with the pump and system curves. The larger the shaded area in Figure 8.34p, the higher the Ht /Hs ratio will be and, therefore, the shorter the payback for the use of variable-speed pumps is likely to be.

Speed, %

Th ro ttl

Head or pressure

100% speed

Zone 1

Static head

100 90 80 70 60 50 40 30

Flow, %

Horsepower required, %

100 90 80 70 60 50 40 30

100 73 51 34 22 13 6 3

Flow

FIG. 8.34n 6 Instead of wasting the unnecessarily introduced pump energy, speed is reduced so that such energy is not introduced in the first place.

© 2006 by Béla Lipták

2092

Control and Optimization of Unit Operations

Pump head-capacity curve

Ht

System head

Design head Ht = 2.3 Hs

Friction head

System head curve

Hs

Static head 0

System flow Average yearly flow rate (Fa)

Ht Design head Hs

100%

Pump head-capacity curve

Eu =

System head

Friction head

Ht = 1.3 Hs

Static head

0

System flow

100%

FIG. 8.34p The higher the Ht/Hs ratio, the more justifiable is the use of variablespeed pumps.

Variable-Speed System Efficiencies The overall system efficiency index (SEI) of a variable-speed pump installation is determined as follows: SEI = ( E p × Em × Ev × Eu )10 − 6 where Ep = Em = Ev = Eu =

the the the the

8.34(4)

pump efficiency (%) motor efficiency (%) variable-speed drive efficiency (%) efficiency of utilization (%)

Figure 7.10z gave some typical wire-to-shaft efficiency values for variable-speed drives (Ev). The efficiencies of the variable-speed drives are represented as ranges, because each design has a different efficiency. Figure 8.34q provides some individual efficiency data for a number of variable-speed drive designs. At 50% of rated speed, the variable-speed drive efficiency can be as low as 40% or as high as 70%, depending on the

© 2006 by Béla Lipták

design selected. Naturally, the less efficient the variablespeed drive, the less expensive it is. Table 8.34r provides some cost information for variable-frequency induction motor drives. The efficiency of utilization (Eu) is an indicator of the quality of the overall piping and equipment design. For example, in the piping distribution system illustrated in Figure 8.34s, Qr might represent the required water flow and Hr the required pressure head to transport Qr . Because three-way valves are used in this illustration, the actual flow (Qa) is much higher than the required flow, and the head required to transport this actual flow (Ha) is also greater than what is required. Therefore, the efficiency of utilization, which is defined in Equation 8.34(5), will also be rather low for the design described in Figure 8.34s. Qr H r 100 Qa H a

8.34(5)

The numerical value of Eu can be accurately obtained only by testing. Once all four efficiencies are determined, they can be represented by a single system efficiency index curve (Figure 8.34t). By combining different pumps, motors, drives, and system designs, it is possible to arrive at a number of SEI curves. The relative advantages of different devices and designs can be evaluated quantitatively by comparing these curves. Types of Variable-Speed Drives Section 7.10 in this volume provides a detailed description of the various variablespeed drive designs, their features including efficiencies, and costs. Table 8.34u provides a summary of some of the main features of these variable-speed drives. Two-Speed Pumps Two-speed motors can be utilized on simple pumping systems in which accurate control of pump pressure is not necessary. Such motors should not be used in systems with high static head and low friction processes. On the other hand, for most friction systems, they can offer a reasonably efficient and inexpensive alternative to variablespeed pumping. As shown in Figure 8.34v, standard twospeed motors are available with speeds of 1750/1150 rpm (29/19 r/s), 1750/850 rpm (29/14 r/s), 1150/850 rpm (19/14 r/s), and 3500/1750 rpm (58/29 r/s). Cavitation Cavitation can be a severe problem; it can quickly destroy a pump. The cause of cavitation is that the pumped liquid flashes to vapor at one point in the impeller (where pressure is below the vapor pressure), and as the spinning impeller throws the liquid and vapor outward, the formed bubbles collapse as the pressure rises above the vapor pressure. When the collapsing bubbles come to the wall, they collide with extreme force. This gives rise to the characteristic sound of

8.34 Pump Controls

Regenerative 90

Voltage source adjustable frequency

90

2093

1

Static D. C.

2 3

80

Drive efficiency, % (Ev)

70 Pulse width modulated adjustable frequency Wound rotor

60

Eddy current Hydraulic clutch Variable voltage

50

40

30 50

Efficiency of motor and control, %

4 80

70 5 60 6 50 Type of variable speed system 1 Energy recovery 2 Direct current 3 Current source variable frequency 4 Pulse width modulated variable frequency 5 Eddy current coupling resistor reactor (secondary) 6 Fluid coupling

40

30

60

70 80 % Rated speed

90

100

20 50

60

70

80

90

100

% Rated speed

FIG. 8.34q 9,10 Variable-speed drives in the 100-HP and larger sizes offer a wide range of efficiencies.

cavitation, and also to the consequent erosion, usually of the impeller. For this reason, it is important to operate pumps only in their operating region where cavitation does not occur. Cavitation usually occurs when the pump is delivering high flow, but some pumps cavitate at low flow too, when liquid recirculating in the impeller passes through points of both low and high pressures (Figure 8.34w). When one does not have reliable data on the locations of cavitation regions, it might be necessary to (very briefly) force the operating pump into cavitation intentionally. A practical way to force a pump to cavitate is to run it at or above design speed and then to restrict its intake gradually. Cavi-

TABLE 8.34r 2004 Cost of Variable-Frequency Induction Motor Drives Power Rating 10 hp (7.5 kW) 20 hp (15 kW) 50 hp (37 kW) 100 hp (75 kW) 200 hp (150 kW) 500 hp (375 kW) 1000 hp (750 kW)

PWM Including Reactor $1,700 $2,500 $5,300 $7,500 $11,000

Current-Fed ASCI

$8,000 $12,000 $18,000 $32,000

Note: The reactor price adds 50–90% to the base price of the variable-frequency drive electronics.

© 2006 by Béla Lipták

tation has a very characteristic and memorable sound that has been described as rocks rolling around inside the pump, or hammering on the impeller. If a pump cavitates, one might try to eliminate or minimize the cavitation by improving inlet conditions. Just as restricting the inlet will force a pump into cavitation, so removing restrictions will often stop cavitation. Reducing the flow range of the pump usually helps also. If the pump cavitates at high flows, try to avoid those flows, possibly by starting a second pump sooner. If cavitation occurs at low flows, one might turn off the pump at low flows, based on some on/off flow control strategy. Some impellers can accept a part, called an inducer, that reduces the pump’s susceptibility to cavitation. An extreme option is to inject a compressible gas into the impeller. This reduces pump efficiency and capacity, but it can eliminate cavitation, because the gas acts as a spring inside the liquid, absorbing the drastic localized pressure changes, and so avoiding the pressure extremes that the pure liquid would experience. However, cavitation is mainly a design concern and should be dealt with during design by ensuring that ample pressure is available around the impeller. Net Positive Suction Head When liquids are being pumped, it is important to keep the pressure in the suction line above the vapor pressure of the

2094

Control and Optimization of Unit Operations

Remote building Coil

Three-way valves on cooling coils

Coil Central plant

Coil

Remote building

Qr

Coil

Qr

Coil

Qr

Coil

Qr Qr

Qr

Chiller

Qa and Ha C

Primary pump

FIG. 8.34s 11 The efficiency of utilization of this design is low, because the water that is bypassing the coils is being circulated unnecessarily.

fluid. The available head measured at the pump suction is called the net positive suction head (NPSH). A pump at sea level that is pumping 60°F water, which has a vapor pressure of hvp = 0.6 ft, and that is operating under a barometric pressure of 33.9 ft, has an available NPSH

(NPSHA) of 33.9 − 0.6 = 33.3 ft. If the impeller centerline is 3.4 ft below the surface elevation of the water being pumped and if the friction losses in the intake piping is 5.3 ft W.C, the NPSHA is 33.9 − 0.6 + 3.4 − 5.3 = 31.4 ft. As shown in Figure 8.34x, the NPSHA increases with barometric

1.00

TABLE 8.34u Variable-Speed Drive Comparison System efficiency index (SEI)

0.80

0.60

Drive Type

0.40

0.20 Minimum flow 0

0

20

40 60 System flow, %

Design flow 80

100

FIG. 8.34t Overall system efficiency index (SEI) curves describe the total 1 pumping efficiency as a function of load.

© 2006 by Béla Lipták

Efficiency (at 70% Speed)

Turndown

Sizes (hp)

Component Requiring Replacement (Frequency in Yrs.)

Wound rotor regenerative

High (85%)

2:1*

25–500+

Brush (3–4)

Direct current

High (80%)

Unlimited

1–500+

Brush (1–2)

Variable frequency

High (78%)

3:1

20–500+

—

Wound rotor

Med. (60%)

2:1*

25–500+

Brush (3–4)

Eddy-current clutch

Med. (58%)

5:1

20–500+

—

Fluid coupling

Med. (57%)

3:1

20–500+

—

Variable voltage

Low (52%)

Limited

10–100+

—

Mechanical

Low (50%)

6:1

1–100

*Unstable below 50%.

Belt or chain (1–3)

8.34 Pump Controls

20 Pump head-capacity curves 18 1750 rpm

16

Region of best efficiency

Total head

14 12 Best efficiency area 10

1150 rpm

8 6

850 rpm

4

550 rpm

2 0

0

20

60 Capacity

40

80

100

120

FIG. 8.34v 1 Several speed combinations are available in two-speed pumps.

pressure and with static head, and it decreases as vapor pressure, friction, or entrance losses rise. Figure 8.34x also illustrates the difference between available and required NPSH (NPSHA and NPSHR). Available NPSHA is the characteristic of the process and represents the difference between the existing absolute suction head and the vapor pressure at the process temperature. The required NPSHR, on the other hand, is a function of the pump design (Figure 8.34f). It represents the minimum margin between suction head and vapor pressure at a particular capacity that

70

Pressure

60

2095

is required for pump operation. If this minimum NPSHR is not available, the pump will fail to generate the required suction lift and the flow will stop. The NPSHR curve in Figure 8.34f describes the required amount of static head at the pump inlet required to avoid a discharge pressure from dropping more than 3% over the zero-cavitation condition (source of definition: ANSI/HI 9.6.1-1998, Paragraph 9.6.1.1). NPSHR is defined in absolute pressure, and typically for water at 0°C unless otherwise stated. The available NPSH (the NPSHA) increases as 1) the barometric pressure increases, 2) the static pressure of the liquid at the entrance of the impeller, and 3) all other suctionside pressures increase. The NPSHA decreases as 1) the vapor pressure of the liquid increases, 2) friction or entrance losses rise, and 3) all other suction-side pressures decrease. NPSH and Cavitation Traditionally, the NPSHR curve has been taken to define the onset of cavitation, and designers have concentrated on ensuring that the NPSHR is met under all operating conditions (Figure 8.34x). However, the accuracy of the NPSHR curve in defining the point when cavitation becomes significant is being questioned. Indications are that cavitation actually begins sooner than previously believed. Allan Burdis (chairman of the Hydraulic Institute’s NPSH Margin Committee) writes: “If you want to operate cavitation-free, you need NPSH margin ratios (NPSHA/ NPSHR) of 4 to 5.” Some evidence suggests that cavitation damage at NPSHA = NPSHR is actually less than the damage when the NPSHA is higher. The Hydraulic Institute’s standard report reads: “There are studies that show the maximum cavitation damage can actually occur at NPSHA values that are twice the NPSHR or more for very high suction energy pumps.” Therefore, we might conclude that the science of cavitation prediction on the basis of NPSHA is still evolving, and if one wants to be positive, actual cavitation testing can be necessary.

50

Water Hammer

40

If a valve opening is suddenly reduced in a moving water column, this causes a pressure wave to travel in the opposite direction to the flow. When this pressure wave reaches a solid surface (elbow, tee, and so on), it is reflected and travels back to the valve. If, in the meantime, the valve has closed, a series of shocks, sounding like hammer blows, results. An example can illustrate this phenomenon. Assume that 60°F (20°C) water is flowing at a velocity of 10 ft/s in a 3 in. Schedule 40 pipe, and a valve located 200 ft downstream is suddenly closed. The pressure rise and the minimum acceptable time for valve closure can be calculated. If the valve closes faster than this time limit, water hammer will result. In rigid pipe, the pressure rise (∆P) is 3 the product of water density (ρ in units of slugs/ft ), the velocity of sound (c in units of ft/s), and the change in water

30 20 10 0 0

4

8

12

16

20

Flow Cavitation region

Characteristic

FIG. 8.34w The dashed segments of the characteristic pump curve show the zones where the regions of cavitation might exist.

© 2006 by Béla Lipták

2096

Control and Optimization of Unit Operations

NPSHA = hb − hs − hvp − (hf + hi)

hb

Allowable NPSH (depends on impeller)

Static head (+ hs)

Barometric pressure on liquid surface (hb)

A

Head, in feet of liquid

hb − hvp

B NPSH = hb + (hs − hvp) − (hf + hi)

ed uir req p m SH NP y pu b

hs

Pipe friction and entrance losses (hf + hi) Pump Allowable suction lift (−hs)

Liquid surface Vapor pressure (hvp)

hf + hi Av a

ila ble

NP

SH

Capacity

FIG. 8.34x The available net positive suction head (NPSHA) increases with barometric pressure and static head, and decreases as vapor pressure, friction or entrance losses rise. (Adapted from Reference 3.)

12

velocity (∆V in units of ft/s). Therefore, the pressure rise can be calculated as follows: ∆P = − ρc ∆V = − (1.937)(4860)(−10) = 94,138 lbf/ft 2 = 653.8 PSI

valves; and (3) when water flows are split or combined, by using vacuum breakers to admit air and thereby cushion the shock resulting from the sudden opening or closing of the second split stream.

8.34(6) Pump Stations

In order to prevent water hammer, the valve closure time (t) must exceed the ratio of two pipe lengths (2L) divided by the speed of sound: t = 2 L /c = (2)(200)/4860 = 0.0823 seconds

8.34(7)

Therefore, in this example, the valve closure should take more than 0.0823 sec. The possible methods of preventing water hammer include (1) designing the system with low velocities, (2) using valves with slow closure rates, and (3) providing slowclosing bypasses around fast-closing valves, such as check 15 valves. When water hammer is already present and the cause of it cannot be corrected, its symptoms can be treated (1) by adding air chambers, accumulators, or surge tanks; (2) by using surge suppressors, such as positively controlled relief

© 2006 by Béla Lipták

When either the flow or the pressure requirements of the process are such that a single pump cannot meet them, pump stations consisting of two or more individual pumps have to be used. Multiple pumps operate in parallel and are used if the process flow rangeability exceeds the throttling capability of a single pump. Booster pumps are installed in series and are used to increase the total discharge pressure of the station. Multiple Pumps in Parallel Individual centrifugal pumps have a rangeability of about 4:1, which can be obtained by either speed control or by discharge throttling. Pump turndown can be increased by 1) bypassing the unwanted flow, 2) turning the pump on and off, and 3) using multiple pumps. When two or more pumps operate in parallel, the combined head-capacity curve is obtained by adding up their individual capacities at each discharge head, as illustrated in

8.34 Pump Controls

P4 P3

a

a b c

P2

b c

P1 Pump head-capacity curves

On @ 2, Off @ 4

Combined head-capacity curve for pumps A and B operating in parallel (Q = QA + QB)

Capacity

PSL a b Head

Check valves

Head-capacity curve, pump A Head-capacity curve, pump B Head

Modified head = P4 − P1 Actual head = P3 − P2

Check valves

a b

System head curve

(5)

(3)

(2)

d e

PSL dead-band

(4)

Two-pump operation Two One pumps pump

Maximum points of operation

Head-capacity curve, pump A Head-capacity curve, pump B Pump head-capacity curves

(3)

Two pumps

100% design head System and pump head

System and pump head

100% system design head

c

Combined headcapacity curve for pumps A and B operating in series (H = HA + HB)

Capacity

FSH dead-band

(1) (4)

d e

Bypasses for single-pump operation

Two-pump operation One-pump operation

c

2097

One pump

(1) (5) One-pump operation

(2)

Maximum points of operation

System head curve

50% design head

Independent head 50% design flow

100% design flow

System flow Two pumps, each with a capacity of 50% design flow at 100% design head

Independent head

100% design flow

System flow Two pumps, each with a capacity of 100% design flow at 50% design head

FIG. 8.34y Pump turndown and rangeability can be increased by operating two or more pumps in parallel.

FIG. 8.34z Multiple pumps in series are effective when the system head curve is steep. The two pumps illustrated by the lower graph are each 1,2 capable of generating 100% design flow at 50% design head.

Figure 8.34y. The total capacity of the pump station is found at the intersection of the combined head-capacity curve with the system head curve. This point also gives the head at which each of the pumps is operating. If the selection is to be very accurate, the head-capacity curves should be modified by substituting the station losses (the friction losses at the suction and discharge of the individual pumps) so that the resulting “modified head” curve will represent the pump plus its valving and fittings. When constant-speed pumps are used in parallel, the added increments of pumping can be started and stopped automatically on the basis of flow. As will be discussed later, a dead band is provided in these controls so that if a new pump is started at flow “x,” the flow will have to drop to, say, “x–5%” before that increment is stopped.

capacity curve is obtained by adding the booster curve to the modified head of the parallel pumps at each capacity point. Series pumping is most effective when the system head curve is steep, such as in Figure 8.34z. With such mostly friction loads, series pumping can substantially reduce the overpressure at low loads. Therefore, booster pumps or twospeed pumps can both be considered for the same kind of steep system curves. Multiple pumps in series are preferred from an operating cost point of view, but the capital cost investment of a single two-speed pump is lower. When constant-speed pumps are used, the booster pump can be started and stopped automatically on the basis of pressure. In this case, an adjustable dead band is provided in the pressure switch. As an example, the normal operating point of the system can be point (1) in Figure 8.34z. As the load increases, the pump discharge head drops and when it reaches point (2) (the set point of the PSL), the booster pump is automatically started. As soon as the booster is on, the system operates at a point to the right of point (3) until the load drops off again.

Booster Pumps When two or more pumps operate in series the total head-capacity curve is obtained by summing up the pump heads at each capacity. When a booster pump is added to a main fed by several parallel pumps, the total head-

© 2006 by Béla Lipták

Control and Optimization of Unit Operations

The booster pump stays on as the load drops below point (3) until the PSL turns it off at point (4). At this point, the system is automatically returned to the single pump operation at point (5). The dead band in the PSL prevents the on/off cycling of the booster pump at any particular load. The width of the dead band is a compromise: As the band is narrowed, the probability of cycling increases, while the widening of the band results in extending the periods during which the booster is operated unnecessarily. If the pumps are identical, their running times can be equalized by alternating them, so that the pump with the higher running time will be the one that is stopped first.

Discharge pressure PSIG (kPa)

2098

120 (828) 100 (690) 80 (552) 60 (414) 40 (276)

Pump curve

20 (138) 20 40 60 80 100 120 140 160 (1.3) (2.5) (3.8) (5.0) (6.3) (7.6) (8.8) (10) Flow, GPM (l/s)

POSITIVE DISPLACEMENT PUMPS

Percent of maximum flow

Reciprocating pumps, such as the piston and diaphragm types, deliver a fixed volume of fluid per stroke. The control of these pumps is based on changing the stroke length, changing the stroke speed, or varying the interval between strokes. In all cases, the discharge from these pumps is a pulsed flow, and for this reason they are not suited to control by throttling valves. In practice, the volume delivered per stroke is less than the full stroke displacement of the piston or diaphragm. This hysteresis is a result of high discharge pressures or high viscosity of the fluid pumped. Under these conditions, the check valves do not seat instantaneously. A calibration chart must, therefore, be drawn for the pump under actual operating conditions. A weight tank or level-calibrated tank is usually the reference standard. Because the discharge is a pulsed flow (Figure 8.34aa), it must be totalized and divided by the time interval to get average flow rate for a particular speed and stroke setting. Metering inaccuracy is approximately ±1% of the actual flow with manual adjustment and ±1.5% with automatic positioning. Methods of stroke and speed adjustment are covered in detail in Section 7.4 in Chapter 7, and other features of metering pumps are discussed in Section 2.14 in Chapter 2 in the first volume of this handbook.

Average flow for triplex pump

Simplex pump Duplex pump Triplex pump 100 94 63 50

Average flow for simplex pump

31

0

90°

180°

270°

Average flow for Duplex pump

360°

Degrees motor shaft rotation

FIG. 8.34aa Flow characteristics of simplex and multiple plunger pumps.

© 2006 by Béla Lipták

FIG. 8.34bb The characteristic curve of a positive displacement (PD) pump operating at constant speed and stroke.

Reciprocating Pumps Reciprocating pumps have a piston or plunger driving element inside a cylinder; the piston moves in and out, in a reciprocating motion (see Figures 7.4n and 7.4o). Liquid is sucked into the pump as the piston moves in, and is forced out as the piston moves out. The liquid flows through oneway valves on the intake to allow flow only into the pump, and on the discharge only to allow it out. The reciprocating piston action results in each stroke displacing a positive, fixed volume of liquid almost regardless of pressure, hence the name. Figure 8.34bb shows a typical PD pump’s flow-to-pressure characteristic. The main feature of this characteristic is that it is essentially constant-flow: The pressure varies greatly while the flow varies very little. Many reciprocating PD pumps allow the stroke length to be adjusted, to change the pumping rate at constant speed, over a 10:1 range. Manufacturers offer higher stroke ranges, and theoretically they could adjust the stroke down to zero, but accuracy tends to decrease at low strokes, because of fixed losses such as valve leakage. PD pumps normally develop higher pressures than centrifugal pumps, and their flow is generally unaffected by discharge pressure, which makes them well-suited to metering applications. They can handle higher viscosity liquids than centrifugal pumps, and they can also handle slurries, but they usually do not handle large entrained solids, as many centrifugal pumps can (Figure 8.34cc). Liquid compressibility can become a consideration at 3 extremely high pressures, so that volume flow (gpm, m /s) reduces slightly more than mass flow (lb/min, kg/s). PD pumps are usually considerably less efficient than centrifugal pumps except in high-head applications, so that large PD pumps are far less common than large centrifugal pumps. Some PD pump assemblies have multiple pumps on a single shaft. This design is normally used to pump reagents

8.34 Pump Controls N = number of pump strokes per minute C = viscosity (centipoise) d = inside diameter of pipe (in)

PSE

Spray dryer

Feed tank Variable speed coupling M

Rotary screw pump

TT TRC

FIG. 8.34cc When PD pumps are transporting slurries, it is advisable to use rupture discs to protect against the development of excessive discharge pressures.

or ingredients that need to be in a specific ratio. Operators adjust the individual pumps so that their volumes per stroke are precisely in the ratio required, then the pumps are driven by a common shaft so that they all run at the same speed and, thus, deliver flow in a specific ratio. Because flow is essentially independent of pressure, PD pumps that operate against a closed discharge (closed isolation valve, blocked pipe, and so on) can develop very high discharge pressures that damage the equipment. For this reason, they often are provided with a pressure relief valve, or in applications with highly viscous liquids and slurries, possibly with a rupture disc to relieve the excessive pressures, back to the pump inlet or the intake source. A PD pump’s high pressure applies equally to the suction side of the pump. Vacuum pressure can damage a PD pump, although this is only common in progressing cavity pumps, where an internal vacuum pressure tends to delaminate the flexible stator seal from the rigid stator support. Vacuum pressure can also damage piping and flexible metal couplings.

Flow Rangeability Speed adjustment is also an effective method of adjusting a PD pump’s output, again generally over a 10:1 range. One can combine both stroke and speed control to achieve rangeability to 100:1, and add on- to offtime (mark-to-space ratio) control to further extend that rangeability. The PD pump’s ability to operate over a wide rangeability may mean that only one pump is needed, which simplifies operation considerably. In process control, a PD pump’s rangeability is important for chemical metering pumps, particularly those used for pH control, where the need for adjusting reagents to 1 part in 100, 1000, or more is not uncommon. Rangeability of a hundred to a thousand is seldom reliably achieved even with a PD pump, while it is impossible with centrifugal pumps having a rangeability of only 4:1 or 5:1. Calibration A PD pump’s check valves do not seat instantaneously, so accurate operation needs to be based on a calibration chart for the pump under actual operating conditions. Usually the reference standard for calibration is a calibration column or weigh tank that measures volume or mass. Calibrate the pump’s flow by dividing the volume (or mass if using mass flow) by the time interval to get the average flow rate for a particular speed and stroke setting. Figure 8.34dd

Nitrogen Vent

LSH FY

FRC

Ratio FY station FR

Calibration zone

× LSL PSV

Pulsation dampener

LSLL

Calibrate Normal path

2

 lvGN   lvC  NPSH = P − Pv ± Ph −  +   525   980Gd 2 

where P = feed tank pressure (psia) Pv = liquid vapor pressure at pump inlet temperature (psia) Ph = head of liquid above or below the pump center line (psid) l = actual length of suction pipe (ft) v = liquid velocity (ft/s) G = liquid specific gravity

FQI

P/D

To process (Under pressure)

2

8.34(8)

S

FT

NPSH and Cavitation It is important to keep the NPSH above 10 psia (69 kPa) or preferably above atmospheric. NPSH can be calculated as follows:

© 2006 by Béla Lipták

2099

QQI

HS

FIG. 8.34dd Ratio and calibration controls for reciprocating pump. When level has reached LHS, the three-way valve returns to the “normal” path and nitrogen enters the tank to initiate discharge. When level drops to LSLL, discharge is terminated by venting off the nitrogen. Counter QQI is running while rising level is between LSL and LSH. Total count, when compared with known calibration volume, gives total error. Hand switch HS initiates calibration cycle by diverting the three-way valve to the “calibrate” path.

2100

0

Control and Optimization of Unit Operations

90

180

270

360

90

180

270

360

3.17 × Average

Simplex

Full stroke Half stroke Suction stroke 180 270 360

180° 90

180° 90

180

270

360

Duplex

0

Average

1.58 × Average Average

1st. head 90

2nd. head 180

270

1st. head 360

90

2nd. head 180

270

360

Triplex

0

1.06 × Average Average 0.90 × Average

2nd. head 1st. head

1st. head

3rd. head

3rd. head

2nd. head

Drive rotation in degrees

FIG. 8.34ee Multiple pistons tend to dampen pressure fluctuations.

shows an automatic calibration facility that can be used for more advanced applications, or to calibrate the pump by hand periodically. If the calibration is done on the suction side of the pump, this usually eliminates errors due to fluid compressibility, because close to the pump intake, the pressure is approximately constant. Pulsating Flow As illustrated in Figure 8.34ee, the reciprocating pump produces a strongly pulsating flow. One method of dampening pulsation is to use multiple cylinders, which, similarly to the way a car engine smoothes its pulsating thrust, smoothes out with multiple cylinders. The other option is to use pulsation dampeners, which are illustrated in Figure 7.4u in Chapter 7, Section 7.4. Dampeners are usually connected into the piping near the pump discharge, and this is effective. However, performance can be improved by minimizing pipe friction losses between the discharge and the dampener, and introducing some friction in the line leading to the process user by a partially throttled valve to provide an acceptably smooth flow. Valves Reciprocating PD pumps have check valves on their intake and discharge, and this adds several requirements that should be considered in the overall design. It is necessary to ensure that the discharge pressure is always higher than the suction pressure. This can be guaranteed by placing a back-pressure regulator on the pump discharge, or possibly through the piping arrangement, as shown in Figures 7.4r and 7.4s in Chapter 7, Section 7.4. This is needed because otherwise the valves may open and allow flow through uncontrolled. It is also recommended to orient the pump vertically, or near-vertically. The check valves are usually seated, at least partially by gravity, so their orientation is important. It is also

© 2006 by Béla Lipták

desirable to strain or filter the pumped liquid to remove particles that may jam the valves, and ensure that the piping is scrupulously cleaned during and after installation. Air-Locking and Cavitation When the purpose of a positive displacement pump is to meter the flow rate, certain precautions are needed. These include the removal of all entrained or dissolved gases, which otherwise can destroy metering accuracy. Figure 7.4t in Chapter 7, Section 7.4, shows how entrained gases can be returned to the supply tank. PD pumps airlock comparatively easily: If a large bubble of gas accumulates between the intake and discharge valves, it can simply expand and contract through the pump strokes, and thereby stop the pumping action completely. This is a particular problem if pumping liquids that give off gas, such as sodium hypochlorite, which gives off chlorine; hydrogen peroxide, which gives off oxygen; or biological sludges, which give off methane, hydrogen sulphide, and other gases. Therefore, one should ensure that the piping inherently intercepts and safely removes gases before they can enter the pump. If the removed gas is poisonous (chlorine and hydrogen sulfide), it must be piped to a safe location. In the case of hydraulic diaphragm pumps, the gas bubble on either side of the diaphragm has the same effect. If the pump acts as though it is airlocked, but there is definitely no gas between the valves, there may be some on the hydraulics side of the diaphragm. PD pumps are unaffected by cavitation, because no sudden collapse of the bubbles formed by cavitation is allowed. The piston strokes under complete control, the cavity (the bubble of vapor that is formed during cavitation) only collapses at the rate allowed by the piston. As a result, the destructive velocities that can be achieved by cavitation in a centrifugal pump do not occur in a PD pump. Chemical Metering Pump Operation A major operational consideration that applies specifically to chemical metering pumps is the distance between the pump and the point where the pumped fluid enters the process. This results in dead time. Assume, for example, that this distance is 100 m and that the liquid’s speed in the pipe is 0.33 m/s. If the pump is transporting a constant dilution water flow, it will take 300 sec from the time when the dilution is changed to the time the new dilution reaches the process. Consequently, dead time limits the best achievable control, and it is always desirable to minimize it. One method to minimize this dead time naturally is to reduce the distance between the point of control and the point of use. This may require a second pipe to carry the dilution liquid. If using this arrangement, consider running the chemical pipe inside the dilution pipe, to provide inherent secondary containment. Another solution is to use feedforward based on a measurement that is in advance of the actual process requirement, by at least as much time as is the dead time.

8.34 Pump Controls

Pacing the dilution or reagent flow to the chemical flow usually requires a separate metering pump, or possibly a flow control valve. If diluting or charging to a number of process users in a specific ratio, one can use fixed valves, such as needle valves, to ensure the correct distribution ratio to individual dilution points.

CONTROL OF PUMPS Capacity control of pumps must recognize the incompressibility of liquids. For this reason, changes in the volumetric flow rate throughout the system occur simultaneously, and density is constant at constant temperature, regardless of pressure. Pump capacity may be affected by (1) a control valve in the discharge of a pump, (2) one-off switching, (3) variation in the speed of the pump, or (4) stroke adjustment of PD pumps. Flow control by on/off switching provides only zero or full flow, whereas the other control methods provide adjustable flows in the system. The applicability of these four methods of capacity control is a function of the pump type, such as centrifugal, rotary, or reciprocating. The possible types of capacity controls for the various pumps are summarized in Table 8.34ff. On/Off Control On/off switching is the most common capacity control in use. It has many disadvantages such as flow surges that often hinder processing, high friction losses, and high electricity peak demand charges. However, on/off control is simple and can be economical, as its consequences do not require redesign to accommodate the limitations of on/off control. Pumps that are controlled only by starting and stopping are said to be constant-speed (CS) pumps. When pumping suspended solids or slurries, when the pump is stopped, a specific disadvantage of on/off control is that the solids may settle out of the liquid and may not go back into suspension when the pump starts again. This can cause plugging. This can also happen if pumping entrained oil and grease: The grease will float and may stick to the top of the pipe. In

TABLE 8.34ff Pump Control Methods Possible Types of Controls Method of Control

On/Off

Throttling

On/off switch

Centrifugal, rotary, or reciprocating

Throttling control valve

Centrifugal or rotary

Speed control

Centrifugal, rotary, or reciprocating

Stroke adjustment

Reciprocating

© 2006 by Béla Lipták

2101

these cases, an option is to keep the pump running: deliver flow through a circulating loop and control capacity with a pressure-controlled bypass back to the feed tank. For example, one can provide intermittent flow to feed a centrifuge by opening an on/off valve by a cycle timer. Such a loop is shown in Figure 8.7j. The pressure-controlled bypass shown in this figure allows the normal pump flow to be maintained in the loop, while the centrifuge feed valve is closed. CS (on/off) pump operation is usually straightforward, except that the pump motor may overheat if it is started and stopped too frequently. Motors are usually rated for a maximum number of starts per hour (sph), because bringing the motor up to speed involves higher currents than keeping it at speed; perhaps ten times higher. Motor heating is propor2 tional to current squared (I R), so the heating during start-up may be 100 times higher than normal. Submersible pumps are usually rated for up to 15 sph, whereas large dry-pit pumps may be limited to 2–4 sph. Control systems must be designed to accommodate starts per hour limitations. On/Off Level Control Figure 8.34gg illustrates the use of level switches for on/off pump control. The interlocks keep the tank level between the settings of LSH and LSL. In this illustration, the two-probe conductivity level switch operates a relay. When conductive liquid reaches the upper LSH probe, the relay closes contacts H and I. At this point the pump starts, and although the level then drops below the LSH probe, the pump keeps running because the holding contact (H) maintains the circuit. When the level drops below the LSL probe, both the load (I) and the holding (H) contacts open, stopping the pump. When the level rises again, no action occurs when the LSL is contacted, because the holding contact is still open. However, when the level reaches the LSH, electrical contact is established, and the relay closes to repeat the pumping cycle. The bottom portion of Figure 8.34gg shows a simple pump starter circuit that is controlled by the three-position hand-off-automatic switch. When the controls are in automatic (contacts 3 and 4 connected), the status of the interlock contact I determines whether the circuit is energized and whether the pump is on. The purpose of the auxiliary motor contact M is to energize the running light R while the pump motor is on. The parallel hot lead to contact 6 allows the operator to check quickly to see if the light has burned out. The amount of interlocking provided for pumping systems is usually greater than that shown in Figure 8.34gg. In addition to the overload (OL) contacts shown in Figure 8.34gg, controls often include safety overrides. These usually detect excessive pressure or vibration, low flow, leaks (submersible pumps), or motor winding temperature, and these overrides usually energize remote alarms. Most pump controls include a reset button that must be pressed after a safety shutdown condition is cleared and before the pump can be restarted. If the same pumps can be

2102

Control and Optimization of Unit Operations

Discharge capacity is 5,000 GPM (315 1/s) per pump

Continuous feed at 75 GPM (4.7 l/s) LSH LSL

1,000 gallon (3.785 l) between settings

Continuous feed at 4,000 GPM (252 l/s)

Intermittent discharge at 100 GPM (6.3 l/s)

M

M

Alternator

M LSHH

SUMP

LSH

LSL

B Holding contact (H)

To pump-down motor starter

FIG. 8.34hh On-off level control of dual pump station.

Electrodes

Load contact (I)

3

LSL LSH

A. C. supply Relay

Conductive liquid

Transformer

H

N On front of panel H–0–A

7

1

2

3

4

10

OL’s M

11

I 8

5

9

R

M Interlock contact for automatic control by LSH and LSL

6 Auxiliary motor contact

FIG. 8.34gg On-off pump-down interlocks often utilize two-probe conductivity switches.

controlled from several locations, interlocks should be provided to resolve conflicting requests. One method is to provide an interlock to determine the location that is in control. This can be a simple local/remote switch or a more complicated system. When many locations are involved, conflicts between requests are resolved by interlocks that establish priorities between control locations on the basis of selecting the lowest, highest, safest choice of pump operation. In such installations with multiple control centers, feedback should be provided, so the operators not only know the actual status of all pumps, but also are aware of any conflicting requests coming from the other operators or computers. Multiple Speeds, Multiple Pumps When two-speed pumps are used, added interlocks are frequently provided. One inter-

© 2006 by Béla Lipták

lock might guarantee that even if the operator starts the pump in high speed, it will operate for 0–30 sec in low before advancing automatically to high. This makes the transition from off to high speed more gradual. Another interlock might guarantee that when the pump is switched from high to low, it will be off the high speed for 0–30 sec before the low speed is engaged, to give time for the pump to slow down. When several pumps are supplied from a common electrical feeder, it may be necessary to ensure that the feeder is not overloaded by the inrush current, when pumps are started simultaneously. If this feature is desired, a 0–30 sec time delay is usually provided between pump starts. When two or more pumps are used, an alternator should be interposed between the level switches and the pump motors. The alternator places the pumps in service in an alternating sequence. Thus, if two pumps are used, each will have half as many starts per hour. However, while one pump is out of service, the other must pump all the flow so it will have the same sph as a single pump. Alternation tends to equalize pump running hours and reduces starts per hour. In Figure 8.34hh, a cooling water return sump is illustrated. In this system, each pump is designed to handle the normal flow by itself, but both pumps operate together if abnormally high flows are required. On/Off Flow Control Figures 8.34y and 8.34ii show a twopump arrangement that responds to varying flow demands measured on the discharge side of the pumps. Pump I normally operates at point (1) (at 80 gpm and 36 ft, or 5 l/s and 10.8 m). When flow demand increases to 120 gpm (7.6 l/s), the head drops to 22 ft at point (2), and FSH starts pump II. The combined characteristic gives 120 gpm at 40 ft (7.6 l/s at 12 m) at point (3). In this control scheme, a wide range of flows is possible without serious loss of discharge pressure. On/Off Pressure Control As was illustrated in Figure 8.34z, a pressure switch may be used to start a spare pump in order to maintain pressure in a critical service when the operating pump fails. In this case, a low pressure switch would actuate the spare pump, which is piped in parallel with the first pump.

8.34 Pump Controls

M

Set at FSH 120 GPM (7.6 1/s)

M

Pump I

Pump II

Pump discharge pressure (in feet or m of head)*

FSH dead-band Combined pump curve 50 (15) 40 (12) 30 (9) 20 (6) 10 (3)

(4) (3)

(1) (5)

0

(2)

20 40 60 80 100 (1.3) (3.8) (6.3) (2.5) (5.0)

Single pump curve

140 (8.8)

180 (11.3)

220 (13.5)

260 (16.4)

Flow, GPM (1/s)

FIG. 8.34ii On-off flow control can be used with parallel pumps. *1.0 ft of water = 2.98kPa.

A second possibility is to boost pressure, as shown in Figure 8.34jj. In this case, if pump I is normally operating at point (1), when the discharge pressure rises to 50 ft (15 m) at point (2), the flow is reduced from 53 to 20 gpm (3.3 to 1.3 l/s). At this point, the pressure switch (PSH) will start pump II and close the bypass valve. The system will now

Pump discharge pressure (in feet or m of head)*

PV

PSH

M Pump I

110 (33) 100 (30) 90 (27) 80 (24) 70 (21) 60 (18) 50 (15) 40 (12) 30 (9) 20 (6) 10 (3)

M Pump II Combined pump curve

operate at point (3) on the combined characteristic curve, delivering 60 gpm at 50 ft (3.8 l/s at 15 m) pressure. Using PLCs The previously described simple control strategies can be easily implemented in hardwired controls. With a programmable logic controller (PLC), it is easy to add many other features beyond what is practical in hardwire, such as to equalize the pump run times better, rather than just alternating when it is necessary to stop or start a pump. For example, it is simple to stop the running pump that has logged the most hours or start the pump that has the lower sph. One can also program the PLC to flush the piping periodically by running both pumps together. If it has been a long time since both pumps ran together, and there is storage capacity available, one can initiate the running of both pumps by delaying the starting of one pump until more demand has accumulated and then run both pumps together. This strategy adds a pump start but is unlikely to result in an sph violation because it only happens if both pumps have not run together for some time. If a pump has not run for several days because demand is low, and there is some liquid available to pump, it is recommended to run the pump regardless of demand, even if only for a few seconds. This strategy helps to keep the bearings lubricated and avoids developing flats on ball- or rollerbearings. It also adds starts, but again this only happens if demand is low, when starts are unlikely to be a consideration. Starts per Hour Traditional SPH methods include the use of reduced-voltage or soft motor starters to increase the allowable number of SPH. Other design solutions include the providing of enough storage capacity in the system to accommodate the SPH limitation. One generally tries to avoid the technique of measuring the time between starts, and actively limit these starts, because this is an intrusive approach that is likely to interfere with process performance. Reduced-voltage starters increase the allowable number of SPH, so they are a successful way of accommodating the sph limitation. However, when this means has been exhausted, the next step is to provide storage capacity. The sph limitation can be met by calculating the pumps’ minimum cycle time (minimum start-to-start time), which is given by MCT = 4V/Q

(2) Single pump curve

0

8.34(9)

(3)

(1)

10 20 30 40 50 60 70 (0.6) (1.9) (3.2) (4.4) (1.3) (2.5) (3.8) Flow, GPM (l/s)

FIG. 8.34jj On-off pressure control of pumps. *loft of water = 2.98kPa.

© 2006 by Béla Lipták

2103

where MCT is minimum cycle time in minutes V is the available storage volume Q is the design flow of the pump Example: A 10,000 gpm pump is rated for up to four SPH. Find the working volume needed to ensure that the rating is not exceeded. Four SPH corresponds to an MCT of 15 min. Therefore, V=

15 ⋅ 10, 000 = 37,500 gallons. 4

Control and Optimization of Unit Operations

Bypass Valves With a bypass valve, the pump delivers essentially a constant flow, and the bypass valve returns whatever the process does not need back to the pump inlet. The process determines by the opening of the flow control valve. For example, if the pump is supplying heating water to a reactor, when more heat is called for, the bypass valve is throttled to close, to increase heat to the reactor. Bypass valves can work with both PD and centrifugal pumps, and they have other advantages, such as 1) When pumping slurries, greases, or mixtures that may separate, solidify, or coagulate, the bypass keeps the flow moving and so tends to keep the liquid homogeneous. 2) When pumping heating water, the water in the piping tends to lose heat. A bypass keeps the flow constant, which tends to ensure a consistent temperature to the process. 3) When pumping liquids with entrained gases, keeping the flow moving tends to avoid accumulation of larger gas bubbles that can airlock the system. 4) The heat gain in the pumped liquid stays constant. This is particularly important if pumping liquids near their vapor pressure. Throttling Valves Throttling restricts the pump’s discharge flow. When pumping incompressible (liquid) flow, throttle the discharge, but when gas (compressible) is transported, throttle the inlet. With a liquid, the problem with inlet throttling is that it causes cavitation, which cannot happen with a gas. With a gas, the advantage of inlet throttling is that the throttled gas expands into the blower at a reduced density (which reduces its power consumption). However, this cannot happen with an incompressible liquid — incompressible also implies inexpandable. So, the arrangement of liquid discharge throttling, and gas inlet throttling, makes the best of both possibilities. Essentially, throttling changes the friction factor of the system curve. It works well with centrifugal pumps by shifting the pump’s operating point, but does not with PD pumps because their characteristic curve is very steep (Figure 8.34bb), and therefore the discharge pressure just increases to force the positive displacement flow through the valve.

© 2006 by Béla Lipták

y nc

nc

y

cie

cie

effi

% 50

70 (2) (483) 60 (414) 50 PSID 50 (345 kPa) (345) control valve 40 (276) drop 30 (207) 20 (138) 10 (69)

%

Pump discharge pressure PSIG (kPa)

The flow from a pump or pumping station can be controlled by throttling either the forward flow or the bypass flow around the pump. Capacity control by valve throttling works essentially by throwing away what is not needed. The unnecessarily introduced pumping energy is wasted by either diverting the flow that is not needed through a bypass, or by restricting the pump discharge. Pumps are usually powered by induction motors, which for a long time have been constant-speed devices. As a result, it was difficult or impractical to adjust pump speed, and alternative ways of controlling the pump’s capacity were required. Modulating valves made CS pump capacity control practical, but they wasted energy. The development of reliable adjustablespeed (AS) induction motor drives has changed this situation, although modulating valves are still used.

Pump curve

effi

Modulating Control

30

2104

(1)

System friction plus static loss curve

15 PSID (103.5 kPa) control valve drop 15 PSID system friction loss 13 PSID (89.7 kPa) Static loss

10 20 30 40 50 60 70 80 (0.6) (1.3) (1.9) (2.5) (3.2) (3.8) (4.4) (5.0) Flow, GPM (l/s)

FIG. 8.34kk Throttling control of centrifugal pump.

In Figure 8.34kk, design point (1) is near the maximum efficiency of the pump. Therefore, when throttling to point (2), the efficiency will drop. For good controllability, the control valve should be sized to pass the design flow with a pressure drop equal to the system dynamic friction losses excluding the control valve but not less than 10 psi (70 kPa) minimum. (For more details on assigning sizing pressure drops to control valves, refer to Section 6.15 in Chapter 6.) The control of flow by varying the pressure drop across the valve is illustrated in Figure 8.34kk. Here, when the flow is throttled from point (1) at 73 gpm (4.5 l/s) to point (2) at 15 gpm (0.95 l/s), the differential pressure across the control valve increases from 15 PSID (100 kPa) to 50 PSID (350 kPa). However, one should be careful not to run the pump at flows low enough to overheat the liquid, or vaporize it, because vaporization can cause the pump to cavitate. One should use a heat balance on the pump to calculate the minimum flow needed through the pump to prevent vaporization. To be conservative, assume that all the motor’s power is converted into heat. If this minimum flow was calculated to be 20 gpm (1.3 l/s), then size the PCV in the bypass to pass 20 gpm with a corresponding set pressure of 63 psi (435 kPa). Figure 8.34ll shows a typical flow control loop with a pressure-controlled kick-back bypass. The rangeability of the control valve is assumed to be 25:1 (see Section 6.7). Thus, if the maximum flow required is 70 gpm (4.4 l/s) through the flow control valve, then the minimum controllable flow would be about 3 gpm (0.2 l/s). If control of lower flows is required, then install a second, smaller flow control valve in parallel with the first, and use it to control these lower flows. Throttling Valves Waste Energy If there is no throttling valve in the system, a constant-speed pump will always operate at the intersection of its characteristic curve and the system

8.34 Pump Controls

burn up the differential pressure between points (1) and (3). This drop represents wasted power:

PCV

Pw = p ⋅ Q

FRC

M

FIG. 8.34ll Throttling control with pressure kickback.

curve. If the actual system curve has less slope than was designed for, the pump will deliver more flow than was intended. For example, in the case of the pump and system curves shown in Figure 8.34mm, the designers expected one system curve (solid line), but the actual system curve (dashed line) turned out to be flatter. Consequently, the actual operating point will be at point (2) instead of point (1), and therefore the actual flow will be higher, and the pressure lower, than the designers intended. If a throttling valve on the pump discharge controls the flow, the pump will operate at point (1), and the valve will

Pump head-capacity curve Point 4

System and pump head

Point 2 Point 1 Design system head curve

If the valve throttles the flow to 50%, as in points (2) to (4), the energy wasted in the form of valve pressure drop increases, because it becomes the difference between points (4) and (5). This illustrates that throttling always wastes energy. In addition, throttling introduces another source of energy wastage, because it almost always also reduces the pump’s efficiency (Figure 8.34k). In Figure 8.34j, the useful pumping pressure is identified as Hs, and the actual pressure of the throttled system is given as Hp. The (Hp − Hs) difference identifies the energy wasted through throttling. However, this is only part of the total waste, because moving the operating point from (2) to (1) also reduces the pump efficiency from 81 to 71%. As a result, not only is power lost to throttling, but power is also lost to reduced pump efficiency. One can use Equation 8.34(10) to quantify the loss, by taking the example in Figure 8.34kk. If working with metric units, when the flow is throttled from point (1) at 4.5 l/s to point (2) at 0.95 l/s, the pressure across the valve increases from 100 –3 3 kPa to 350 kPa. At point (1), the flow is 4.5 l/s (4.5 × 10 m /s) 3 and required pressure is 190 kPa (4.5 × 10 Pa), giving a required power of Pw = 4.5 × 10 −3 ⋅ 190 × 10 3 = 850 w However, the consumed power, including the valve’s pressure drop, is

Overpressure with constant speed pump

Point 3

Actual system head curve

100% design flow

50% design flow

Pw = 4.5 × 10 −3 ⋅ 290 × 10 3 = 1300 w 850 Therefore, the control’s efficiency at point (1) is 1300 = 65% We can now compare that result with point (2), which has a flow of 0.95 l/s and required pressure of only 95 kPa, giving a required power of

Pw = 0.95 × 10 −3 ⋅ 95 × 10 3 = 90 w

Point 5

System flow

FIG. 8.34mm Illustration of the consequence when the assumed and actual system curves are not the same.

© 2006 by Béla Lipták

8.34(10)

where Pw is the wasted power (watts, W) p is the differential pressure across the valve (Pa) 3 Q is the flow through the valve (m /s)

FT

Design head

2105

The consumed power, including the valve’s pressure drop, is Pw = 0.95 × 10 −3 ⋅ 440 × 10 3 = 420 w 90 Therefore, the controls efficiency at point (2) is 440 =20% Figure 8.34kk does not provide the full pump efficiencies, but point (1)’s efficiency is likely to be about 60%, and

2106

Control and Optimization of Unit Operations

point (2)’s about 15%. Including these values gives the overall efficiencies of 40% (65 × 60%) for point (1) and 3% (20 × 15%) for point (2).

80 Large pumps 70

Capacity Control by Speed Adjustment The cost of reliable and efficient adjustable-frequency drives (AFDs) has dropped rapidly in the past decade. As a result, adjustablespeed pumping is preferable today for both centrifugal and PD pumps. The main advantage of adjustable speed pumping is that it is efficient. It is efficient in two ways: 1) Rather than wasting energy (as do valves), speed control avoids introducing unnecessary energy in the first place, and 2) A modulating valve almost always moves the pump’s operating point away from its best efficiency point (BEP). Speed control also moves the pump away from its BEP, but not nearly as much. The basic concept is that a pump’s speed controls its discharge flow: increase speed to deliver more flow, and reduce it to deliver less. Multiple-Pump Controls Distribution Controls From a control quality point of view, distribution controls have already been discussed in Section 2.23

© 2006 by Béla Lipták

Curve (3) Variable speed pumps (Figure 8.34pp)

60 Small pumps 50 Overall efficiency, %

Capacity Control by Stroke Adjustment The flow generated by a PD pump is the product of the volume per stroke and the number of strokes per second. Therefore, one can adjust the PD pump flow by either adjusting the volume per stroke or the strokes per second (the speed). The volume of a piston is its area multiplied by the stroke length. Because the stroke length is readily adjustable, stroke control refers to the pump’s flow adjustment method of modulating its stroke length. The stroke can be adjusted by hand or automatically while the PD pump is operating at constant speed. Hand adjustments are currently more precise, usually rated at about 1% error, compared to 1.5% error when automatic adjustment is used. While manual setting is more accurate, automatic adjustment by closed-loop control will give better performance overall. The range of flow control by stroke adjustment is 0–100%. However, in order to maintain accuracy, the practical range is 10–100% of maximum flow. The flow is related to stroke length through system calibration. In chemical metering pump applications, it is often the case that the automatic controls modulate the pumps’ speeds, while the operators adjust the strokes by hand. The reason for this technique is to use stroke adjustment to compensate for factors that are hard to measure and remain constant over long periods, such as the reagent concentration, that may change only from delivery to delivery. In such configurations, the feedback controls automatically adjust the speed to compensate for factors that change often and can be measured, such as the process flows or the reagent deterioration over time, possibly caused by temperature and so on.

40

30 Curve (2) Constant speed pumps (Figure 8.34oo)

20

Curve (1) Constant volume system (Figure 8.34s)

10

0

0

20

40

60

80

100

System flow, % full load

FIG. 8.34nn Overall efficiency is maximum if two variable-speed pumps are used 4 and is minimum with a constant volume installation.

in Chapter 2 and Section 8.28. Here, their description will be from the perspective of the performance of the pump station. The overall efficiency of the water distribution system, which was shown in Figure 8.34s, is described in Figure 8.34nn by curve 1. A substantial increase in efficiency (reduction in operating cost) can be obtained by replacing the three-way valves in Figure 8.34s with two-way ones and by replacing the single large pump with smaller ones. Figure 8.34oo shows such a system. Here, a small and a large primary pump are provided at the main supply point in the central plant, and a small and a large booster pump are furnished in each of the user buildings. When the load is low, the small pumps are operating; when it is high, the large pumps take their place. The minimum flow requirements of the chiller are guaranteed by a bypass valve, and the chilled water makeup into the recirculating loop of each building is under temperature control (TC). The resulting improvement in overall efficiency is shown by curve 2 in Figure 8.34nn.

8.34 Pump Controls

Two-way control valves on cooling coils Secondary booster pumps in each building, one small and one large constant speed pump

Remote building

Central plant

Coil

Coil

Coil

Coil

Coil C

TC

C

C

C TC

Chiller bypass valve for minimum flow

Chiller

Remote building

Coil

C

C

2107

Return temperature control valves

Primary pumps, one small and one large constant speed pump

FIG. 8.34oo 2 Supply-demand matching can be achieved using constant-speed pumps of different sizes.

The highest overall efficiency can be obtained through the use of variable-volume load-following. Figure 8.34pp illustrates such a system, utilizing variable-speed pumps in two sizes. In this system, all waste is eliminated except what

is generated by the small minimum flow bypass around the chiller, which is guaranteed by a small constant-speed pump. The resulting increase in overall efficiency is illustrated by curve 3 in Figure 8.34nn.

Remote building

Two-way valves on cooling coils

Central plant V

Remote building

Coil

Coil

Coil

Coil

Coil

Coil

V

V

V

FSL C

Chiller bypass pump started by FSL to guarantee minimum flow

Chiller

V

V

Primary pumps. One large variable speed and one small variable speed pump

FIG. 8.34pp 4 Variable-volume water distribution systems provide maximum efficiency.

© 2006 by Béla Lipták

Secondary pumps in each building. One large and one small variable speed pump

2108

Control and Optimization of Unit Operations

Starting and Stopping Pumps When the running adjustable speed pumps are near their maximum speeds and when more flow is needed, the capacity of the pumping station must be increased by starting another pump. Similarly, when all pumps are at a low speed and less flow is needed, a pump must be stopped. It is necessary to know the maximum pump speed a pump can run at and still deliver zero flow (called omega zero, ωo ), the pressure up to which the flow remains zero (shut-off pressure). This is the pressure that a starting pump must overcome before it can start delivering flow (superimposed back-pressure), and the flow delivered by an adjustable-speed centrifugal pump, which is running against the back-pressure, which is superimposed by the other operating pumps. The subject of determining the above values for particular pumps and processes and to use these values in developing a detailed strategy for starting and stopping individual pumps is beyond the scope of this section, but if the reader needs such information, it can be found in Reference 15. When to Start or Stop a Pump Accumulated running hours should be recorded on an elapsed time meter (ETM), because pumps are subject to wear while running, so recording their running hours can help distribute wear among them evenly, and it provides a guide to planned maintenance. The time since the last stopping of the pump, called idle time, also should be recorded. During idle time, pump bearings lose lubrication, and they can be flattened out of round, so long idle periods are undesirable. Recording the idle time helps deal with this. A pump accumulates idle time while stopped, in the same way that it only accumulates running hours while running. In critical applications consider always keeping an extra pump running at the same speed as the other pumps, so that if a pump fails it is only necessary to speed up the running pumps. One of the pumps should be stopped when the speed of the operating pumps has dropped to its low limit for over 30 sec. The operator or the control system should initiate a start immediately when a running pump fails, unless a start is already in progress. A pump should also be started when the running AS pumps have been at their maximum speeds (ωm) for over 30 sec. Similarly, if an operating pump has traveled on its characteristic curve to the point where cavitation starts (Figure 8.34w), an additional pump should be started. Start a pump if its idle time indicates that its bearings would benefit from being rotated. Preferably do this when the pumps are running around the middle of the speed range, so that a normal start or stop is unlikely to occur at the same time. Probably also stop another pump at the same time, to avoid a surge or having too much running capacity. Selecting the Pump to Start or Stop When it is time to start a pump, start the AS pump that has accumulated the longest idle time. If all the AS pumps are running, start the

© 2006 by Béla Lipták

CS pump that has the longest idle time. This strategy avoids accumulating idle time, consistent with not adding any extra starts. When it is time to stop a pump, stop the running CS pump that has accumulated the most running hours. If all CS pumps are stopped, stop the AS pump that has accumulated the most running hours. This strategy tends to equalize the pumps’ running hours, but it is not extremely successful because equalizing hours has the lowest priority— everything else takes precedence. A highly successful strategy is to start the pump with fewest hours, stop the pump with the most hours. However, this can allow equipment to stand idle for long periods, particularly when other equipment is brought back into service after a long downtime. Starting and Stopping When the pumps are of equal size and a pump is being started, accelerate it up to the omega zero speed (ω 0 ) quickly on its AFD’s acceleration ramp. Then, gradually increase the flow through the starting pump, and reduce it through the running adjustable-speed pumps, until they are all running at the same speed and passing a similar flow. Similarly, when stopping, control the pump’s deceleration down to ω 0, while accelerating up the running pumps. When the stopping pump reaches ω 0, stop it quickly. When starting an AS pump, it is necessary to know how many AS pumps are running. Next, determine the starting pump’s ω 0. If necessary, perhaps because as-commissioned pump measurements may not be available when the controls are configured, use the characteristic and system curves to estimate ω 0 against n. Start the pump and accelerate it up to ω 0 quickly. When the starting pump reaches ω 0, start a timer, the transition timer, to bring the starting pump into action. Let the duration of the timer be T sec and the elapsed time since the start be t sec. Depending on the system, the timer’s duration may be faster or slower; it is usually not critical, provided it is slow enough that the pumps can follow the flow changes it requires, and fast enough to ensure it will have timed out before there is any need to start another pump. While the transition timer is running (0 < t < T), calculate the starting pump’s flow increase, and the running pumps’ flow decrease, to pump the instantaneous flow required by the control loop output, Q, throughout the start transition. Then convert these flows to pump speeds, as follows: When the transition timer has timed out (t = T), the starting pump has joined the running pumps. Update the count of the number of running pumps, n. When stopping an AS pump, the principles are the same as was for starting an AS pump, except in reverse: decelerate the stopping pump to ω 0 while accelerating the running pumps, so that total flow is controlled while it also transitions smoothly from the stopping pump to the running pumps.

8.34 Pump Controls

CONCLUSIONS This section covered the basics of pump control, while the next section will concentrate on the optimization of this unit operation. Pumping controls are a prime example of applications where it is essential to fully understand the personality of both the process and the pumping equipment used, before a successful control system can be designed. Another unique characteristic of the pumping process is that over the life of the plant, the operating cost of a pump is much greater (sometimes a hundred times greater) than the first cost of the pumping equipment. It is for this reason that good process controls and optimization can have much higher returns when operating pumping stations than on other unit operations. The goal of a well-designed pumping control system is good supply–demand matching, which will not only lower operating costs, but also reduce maintenance and cycling. The full automation of pumping stations — including automatic start-up and shutdown — not only will reduce operating costs but will also increase operating safety as human errors are eliminated.

References 1.

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

American Society of Heating, Refrigerating, and Air-Conditioning Engineers (ASHRAE), “Centrifugal Pumps,” in ASHRAE Handbook, 2003 Equipment Volume, Atlanta, GA: ASHRAE, 2003, Chapter 39. Tchobanoglous, G., Wastewater Engineering, Metcalf and Eddy, Inc., New York: McGraw-Hill, 1981. Karassik, I., Centrifugal Pumps, F. W. Dodge Corp., 1960. Rishel, J. B., “Water System Head Analysis,” Plant Engineering, October 13, 1977. Conzett, J. C., “Adjustable-Speed Drives,” Bulletin D-7100, Reliance Electric, July 1981. Lipták, B. G., “Save Energy by Optimizing Your Boilers, Chillers, and Pumps,” InTech, March 1981. Langfeldt, M. K., “Economic Consideration of Variable Speed Drives,” ASME Paper 80–PET–81, 1980. Liu, T., “Controlling Pipeline Pumps for Energy Efficiency,” InTech, June 1979. Merritt, R., “What’s Happening with Pumps,” Instruments and Control Systems, September 1980. Schroeder, E. G., “Choose Variable-Speed Drives for Pump and Fan Efficiency,” InTech, September 1980. Rishel, J. B., “Wire to Water Efficiency of Pumping Systems,” Central Chilled Water Conference, 1975, Purdue University. Baumeister, T. (Ed.), Mark’s Standard Handbook for Mechanical Engineers, 8th edition, New York: McGraw-Hill, 1978, p. 3.69. Shinskey, F. G., Energy Conservation Control, New York: Academic Press, 1978. Systecon (Division of CEC), Bulletin No. 10-320-1. Karassik, I., “What are the Characteristics of the Unknown Pump?” Center for Professional Advancement.

© 2006 by Béla Lipták

16.

2109

Bain, S., “Variable-Speed Pumping Requires Careful Study of Conditions,” WEFTEC 65th Conference, Collection Systems Proceedings.

Bibliography Avallone, E.A., and Baumeister, T., Mark’s Standard Handbook for Mechanical Engineers, New York: McGraw-Hill, 1996 Baumann, H. D., “A Case for Butterfly Valves in Throttling Applications,” Instruments and Control Systems, May 1979. Baumann, H. D., “Control Valves vs. Speed Controlled Pump,” Texas A&M Symposium, 1981. Baumann, H. D., “How to Assign Pressure Drop Across Liquid Control Valves,” Proceedings at 29th Annual Symposium on Instrument Engineering for the Process Industry, Texas A&M University, January, 1974. Bose, B. K., Power Electronics and AC Drives, Englewood Cliffs, NJ: Prentice Hall, 1986. Bower, J. R., “The Economics of Variable-Speed Pumping with SpeedChanging Devices,” Pump World. Cheremisinoff, N., Pumps and Pumping Operations, Engelwood Cliffs, NJ: Prentice Hall, 1992 Fischer, A. K., “Using Pumps for Flow Control,” Instruments and Control Systems, March 1983. Gibson, I. H., “Variable-Speed Pumps for Flow Control,” ISA/93 Technical Conference, Chicago, IL, September 19–24, 1993. Gottliebson, M., “Explore the Use of Variable-Speed Water Booster Pumps,” Water and Wastes Engineering, May 1978. Gray, C. B., Electrical Machines and Drive Systems, Longman Scientific and Technical, 1989. Greene, Pumping Machinery, New York: Wiley. Hall, J., “Motor Drives Keep Pace with Changing Technology,” Instrument and Control Systems, September 1982. Hall, J., “Pump Primer,” Instruments and Control Systems, April 1977. Holland, F. A., “Centrifugal Pumps,” Chemical Engineering, July 4, 1966. Janki, C., “What’s New in Motors and Motor Controls?” Instruments and Control Systems, November 1979. Karassik, I.J. et al., Pump Handbook, 3rd Edition, New York: Mc GrawHill, 2000. Leonhard, W., Control of Electrical Drives, Springer-Verlag, 1985. Lipták, B. G., “Costs and Benefits of Boiler, Chiller, and Pump Optimization,” Instrumentation in the Chemical and Petroleum Industries, Vol. 16, Research Triangle Park, NC: Instrument Society of America, 1980. Merritt, R., “Energy-Saving Devices for AC Motors,” Instruments and Control Systems, March 1980. Murphy, J. M. D., and Turnbull, F. G., Power Electronic Control of AC Motors, Pergamon Press, 1988. Papez, J. S., and Allis, L., “Consideration in the Application of VariableFrequency Drive for Pipelines,” ASME Paper 80-PET-78, 1980. Poyton, Metering Pumps, Marcel Dekker. Rishel, J. B., “The Case for Variable-Speed Pumping Systems,” Plant Engineering, November 1974. Rishel, J. B., “Matching Pumps to System Requirements,” Plant Engineering, October 1975. Stewart, R. F., “Applying Adjustable-Speed AC Drives,” Instruments and Control Systems, July 1981. Thurlow, C., “Pumps and the Chemical Plant,” Chemical Engineering, May 24, 1965.