Академический Документы
Профессиональный Документы
Культура Документы
3. Hydrodynamics
3.1 General considerations �����������������������������������������������������������������������������001
3. Hydrodynamics
3.1 General considerations
Mechanics is the teaching of strengths and effects on bodies which are ex-posed to these
strengths. Mechanics is part of natural science and builds up on theoretical laws and
experiences by scheduled observations of nature’s phenomena. The bodies which play a
roll in the analysis of mechanics (e. g. the effects on a motor vehicle at a crash test), have,
in this case, solid structures.
Hydromechanics considers the effects and laws of flowing liquid media. The name
hydromechanics is derived from the Greek language and means exactly "the teaching
of the mechanics of water". Since water an outstanding meaning and powerful place
in engineering, and because of the fact that many results of experiments are based on
measuring with water, the term hydromechanics is well earned.
Hydromechanics contains two head areas: Hydrostatics and hydrodynamics. Hydrostatics
deals mainly with static strengths of a liquid which acts on a vessel or, foe example,
the wall of a pool. The teaching of hydrodynamics describes the theoretical bases and
equations which are needed to be able to calculate the strengths in a fluent body or
medium.
This includes the liquid flow in pipes fittings and valves. The flow in gutters or open channels
and even rivers can be also described and calculated with the help of hydrodynamics.
This book concentrates on linear and/or one-dimensional flow conditions which appear,
for example, in pipes with a constant cross-sectional area.
Added to this, is the fact that the fluid medium in a valve body, is usually forced into
multiple direction changes which causes additional friction losses and increases the total
resistance.
It is normal to consider liquid flow at first, as an ideal model which doesn’t cause any friction.
A customization of the model is then carried out to examine a more realistic relationships
by the introduction of hydraulic friction coefficients. There are two basic relationships on
which all calculations of resistances build up in hydraulic systems: The flow or continuity
equation and Bernoulli’s equation named after the Swiss physicist Daniel Bernoulli.
The resulting flow rate Q at a velocity w can be calculated according to equation 3-1:
Q = A⋅v (3-1)
If the physical unit is m2 for the cross-section and m/s for the fluid speed, the flow volume
unit becomes m3/s. When the pipe cross-section is changed at an arbitrary value, the
flowing speed of the liquid must change too, since the liquid is incompressible. The mean
velocity of the fluid at constant flow can be calculated at any location, if the accompanying
cross-sections are known (Figure 3.3.-2).
Cross section A1
Cross section A2
Velocity v2
Velocity v1
At a reduced cross-sectional area Figure 3.3.-2) the fluid velocity must inevitably increase,
since the flow rate is constant at any point of the piping system. Hence the continuity
equation is:
A1 ⋅ v1 = A 2 ⋅ v 2 = constant (3-2)
Analogous to the law of Ohm the same flow appears in every cross-section of a closed
piping system.
v2
p2
h1
p3
h2 v3
h3
This fundamental knowledge of the preservation of energy was formulated by the German
Robert Mayer for the first time. Daniel Bernoulli showed that these relationships apply
for a flowing liquid as well. Therefore one speaks about Bernoulli’s equation as far as
the energy content of liquid flow is concerned. (Figure 3.4.-1) shows the relationship
schematically.
A liquid with the density ρ fills out the pipe completely and streams from above to
below. If looks one, for example, at the energy of a liquid particle in position 1, three
different energy forms can be distinguished:
Static pressure energy (p1) which is inherent in the liquid and can be measured by
means of a pressure gauge in the respective position,
Speed (velocity) energy (v1) which depends on the mass of the particle and the flowing
speed of the liquid,
Potential energy (h1) which results from the geodesic height h1. The reference line by
which the respective height is measured is fixed in this example.
Assuming a friction free flow, the energy content remains unchanged in positions 2 and
3. The potential energy has, for example, reduced by the amount (h1-h3). On the other
hand the velocity energy at the same point has considerably increased, so that the total
energy remains unchanged.
If one expresses the different energy contents by equations, the following relationships
apply:
If the unit of pressure height hx is in meter (m) and the density of the liquid is in kilogram
per cubic meter (kg/m3):
p x = 10 m ⋅ 1000 kg m3 ⋅ 9.81 m s2 ≅ 100,000 Pa
The pressure unit Pascal (Pa) will be used in all theoretical considerations. For actual
process values, however, and for firm examples, the unit (bar) is chosen. The conversion
is as follows: 1.0 bar = 1∙105 Pa or 100 kPa (kilopascal).
For the following considerations it is even simpler to assume water as medium and use
for the pressure heights, i.e. energy, the unit (m). By rearrangement of the equation 3-3
we get the equivalent pressure height hx which corresponds to the static pressure px:
px
hx = (3-4)
ρ ⋅g
If one assumes a mass unit and looks at the kinetic energy of 1.0 kg of the liquid, the
kinetic energy becomes:
v2 (3-6)
2⋅g
After using the common unit for the fluid velocity (m/s) the kinetic energy becomes also a
pressure height. Hence the total energy of a liquid particle at any place within the piping
system is:
px v2
E= + hx + = const.
ρ ⋅g 2⋅g (3-7)
Unfortunately the conditions aren’t ideal in practice, i. e. in reality the amount of total
energy is not constant but reduces itself by the amount of the friction losses. This means,
that the total energy content of the liquid at position 2 of Figure 3.4.-1 is less than in
position 1 and the medium at position 3 has a lower energy content than in position 2.
Equation 3-7 must therefore be corrected under consideration of the unavoidable friction
losses hv and develops to be:
px v2
E= + hx + + hv = const. (3-8)
ρ 2⋅g
When summarizing this knowledge, one can make the following statements:
The different energy forms are equal and can be expressed in pressure heights.
Static pressure energy can simply be changed into velocity energy. In accordance with
the continuity equation the fluid velocity rises and the static pressure falls through a
narrowing pipe or valve.
Velocity energy can, be changed into static pressure energy again, if the speed is
slowly reduced by a progressive extension of the pipe or valve cross-section.
All energy conversions are at all times accompanied by losses. If the fluid velocity is
increased at the expense of the static pressure by a pipe reduction and afterwards
reduced by a pipe extension again, a lasting loss of static pressure appears.
All control valves make use of this effect, in order to reach a reduction of the static pressure
or a regulation of flow through the valve. The energy difference in front of and behind
the control valve arises from internal friction (turbulence) of the medium and is always
associated with noise generation. The acoustic power is proportional to the dissipation of
mechanical energy in the throttling valve.
An open container or vessel with a filling height h, measured from the surface of the liquid
to the end of the short outlet pipe is being emptied. Which short-term flow occurs, if the
inside diameter of the outlet pipe is 100 mm and the filling height h is 2.0 m?
The determination of the cross-section area turns out as follows:
The fluid velocity at the vessel outlet arises from the equation for the “free fall”, which is
applicable for the flow out of an open vessel:
v = 2⋅g⋅h
(3-9)
v = 2 ⋅ 9.81⋅ 2 = 6.26 m s
Without certain corrections, which will be explained later, the actual flow Q becomes:
Velocity v1
Velocity v2
Pressure p2
Pressure p1
Figure 3.6.-1: Flow from a nozzle (schematic)
The pressure outside the nozzle corresponds to atmospheric pressure = 1.0 bar. How
much is the outlet velocity and the flow rate in liters per minute (l/min)? Also in this
example static pressure energy and velocity energy are first converted into heights. The
pressure height hx becomes, in accordance with equation 3-3 for water with a density of
1,000 kg/m3,
px 350000
hx = = ≅ 35 m
ρ ⋅ g 1000 ⋅ 9.81
The velocity energy in the tube still cannot be calculated since the flow rate is unknown,
but starting out from equation 3-7 one can, ignoring the friction losses, assume a pressure
height:
v 22 v12
= hx +
2⋅g 2⋅g
The pressure height at the nozzle outlet must be the sum of the energies which are in
the hose. Since the ratio of the cross-sections is very big (A1/A2 = 302/52 = 36), the term
w12/2g is ignored since it represents only a very low pressure height. The velocity w2 is
thus:
v 2 = 2 ⋅ 9.81⋅ 35 = 26.2 m s
When considering the continuity equation (3-2) the velocity in the hose becomes:
2
A2 D
=v1 v=
2 v2 2
A1 D1
2
5
v1 =⋅
v2 = 26.2 ⋅ 0.028 =
0.73 m s
30
0.0052 ⋅ π ⋅ 26.2
Q = A⋅v = = 0.00051m3 s = 30.9 l min
4
I = m⋅v (3-10)
d
R v
Under the assumption of a stationary flow, the impulse acts as a backward push R against
the direction v of the flowing medium (Figure 3.7.-1). Typical is the reaction force which
can be seen in the handling of the heavy hose of a fire brigade when they direct the hose
at a fire.
Mass is the product of volume and density. Mass per time unit pouring from the vessel
outlet is the product of the cross-sectional area and the velocity of the fluid:
2
= d ⋅π ⋅ v ⋅ ρ
m
4
The outlet velocity can be calculated according to equation 3-9:
v = 2⋅g⋅h
When combining both equations the reaction force (backward push) can be calculated:
d2 ⋅ π d2 ⋅ π
FI
= ⋅ ρ ⋅ v ⋅ 2 ⋅ g=
⋅h ⋅ ρ ⋅ v2 (3-11)
4 4
FI
d v
d2 ⋅ π
FI
= ⋅ ρ ⋅ v2 (3-12)
4
Plate
F
y τ vmax
dy
Fluid
dv
v0
Solid boundary
F= A ⋅ τ (3-13)
The shear stress τ is defined as the friction force related to the area A.
dv
τ= η ⋅ (3-14)
dy
Dynamic viscosity η
The expression dv/dy is the quotient of velocity v and distance y and is described as the
dynamic viscosity η. The recommended SI-unit is Ns/m2.
The SI physical unit of dynamic viscosity is the pascal-second (Pa·s), which is identical
to kg·m−1·s−1.
If a fluid with a viscosity of one Pa·s is placed between two plates, and one plate is
pushed sideways with a shear stress of one pascal, it moves a distance equal to the
thickness of the layer between the plates in one second.
The cgs physical unit for dynamic viscosity is the poise (P), named after Jean Louis
Marie Poiseuille. It is more commonly expressed, particularly in ASTM standards, as
centipoise (cP). Water at 20 °C has a viscosity of 1.0020 cP.
More common, however, are the units Poise (P) or Centipoise (cP).
N⋅ s mN ⋅ s
1Poise =
(P) 0.1Pa=
⋅ s 0.1 1cP
= 1mPa=
⋅s 1
m2 m2
η
ν=
ρ (3-15)
cm2 m2
1St 1= 10−4
=
s s
mm2 m2
1cSt 1= 10-6
=
s s
Non-standard units
Englergrad E (named after the inventor Carl Oswald Viktor Engler) is a unit of kinematic
viscosity and defines the ratio of outflow times of the test fluid compared to that of distilled
water. The volume of the receptacle is in this case 200 cm3, the temperature shall be 20
°C. A conversion into the recommended SI-unit is carried out as follows:
6.31
ν= ( 7.37 ⋅ E ) − ⋅ 10−6 m2 s (3-16)
E
185
=ν 0.266 ⋅ t − ⋅ 10−6 m2 s
t (3-17)
The equation above is limited, however, to the following slowing times: Minimum 32
seconds, maximum 100 seconds. For slowing times above 100 seconds the following
conversion equation applies:
135 −6 2
ν 0.22 ⋅ t −
= ⋅ 10 m s (3-18)
t
The following explanations will help to illustrate the properties of a liquid or a gas
with regard to their viscosity:
The viscosity is usually not a constant value. It depends rather on the temperature and
sometimes also on the pressure of the fluid.
The temperature effect has a larger influence on the viscosity than the static pressure
and is different for liquids and gases. The viscosity of liquids decreases at rising
temperatures (e.g. oil). For gases and vapors however, the viscosity increases at
higher temperatures.
The normally foreseeable effects on viscosity loses validity for certain media. Liquids
which behave according to equation (3-14) are described as Newtonian liquids. For
these media the shear stress remains constant, provided temperature and pressure do
not change. Most fluids used in the process automation industries (water, air, natural
gas etc.) belong to this group. Non-Newtonian fluids are liquids with solid particles
(slurries), slime, suspensions and highly viscous substances such as tar, special oils
or varnishes.
Distinctive characteristics of a non-Newtonian fluid are either a variable viscosity or a
behavior which does not conform with equation 3-13. A typical example for a variable
viscosity is a so-called "thixotropic" varnish whose viscosity depends on the painting
speed with a brush. This effect is intended to avoid "tears" or "noses" when painting a
vertical body like a frame of a window. For such fluids a sufficiently low viscosity arises
in the context of a normal painting speed.
The high basic viscosity of such a varnish avoids, the slow vertical flow and the ugly
„tears“. Also special motor oils as used in motor vehicles today, fall into this category.
By adding certain components (additives) the oil has almost a constant viscosity
independent of its temperature. A mixture of liquids and solid substances is another
example for a non-Newtonian liquid which behaves differently compared with a pure
liquid.
* Pitch is the name for any of a number of highly viscous liquids which appear solid. Pitch
can be made from petroleum products or plants. Petroleum-derived pitch is also called
bitumen. Pitch produced from plants is also known as resin. Products made from plant
resin are also known as rosin.
Dynamic Viscosity Unit Convertor:
See the Homepage: http://www.uniteasy.com/en/unitsCon/Dynamic_viscosity.htm
The greatest problem of the experiment is to reach a geometric similarity with respect to
the surfaces, since the original (e. g. a ship) is already relatively smooth on the outside
skin. Therefore a scaled reduction of the roughness is not always possible. The absolute
roughness k of a surface is usually defined as "middle peak-to-valley height". To make
a comparison between original and model possible, one makes use of the relative
roughness which results from the ratio of the absolute roughness k and a characteristic
length L. From this follows:
Two surfaces are geometrically similar if they show the same relative roughness
k/L.
If one examines the flow pattern of a liquid in detail, a differentiation between pressure
forces, inertial forces and friction forces is required. Utilizing these different forces and
corresponding equations we get a dimensionless expression defined as Reynolds number
Re. The following relation applies to a perfect circular pipe with a constant cross-section:
v ⋅ lq
Re = (3-21)
ν
In equation 3-21 v is the speed of the flow in (m/s), lq the length of the lateral dimension
(pipe diameter) in (m) and the kinematic viscosity ν in (m2/s). Since this book deals in the
first place with control valves, the flow behavior in valve bodies, fittings and pipelines are
of particular interest. The general rule is here:
Two hydrodynamic flow patterns are similar, if the contours and boundaries
exposed to the flow stream are geometrically similar and their Reynolds numbers
are the same.
Injection of
red ink
Transparent glas pipe
If the fluid velocity is steadily increased, sooner or later sudden detachments and
turbulence of the medium occurs due to an increased skin friction at the pipe wall.
Injection of
red ink
Transparent glas pipe
Transverse motions now overcome the main flow direction, as represented schematically
in Figure 3.10.-2. A very small disturbance often causes a sudden change of the flow
pattern.
To make optimal use of pipes, fittings and control valves, turbulent flow predominates
in industrial applications. Laminar flow conditions occur only with very tough media and
control valves with very low throttling areas and small pressure differentials. The flow
pattern of pipes with a perfectly circular cross-section becomes predictable by introduction
of the critical Reynolds number Recrit = 2320. A flow below Recrit is (in pipes) always
laminar.
Also above this number a flow can still be laminar if no eddy causing tremors or other
disturbances appear. With control valves a transitional range can be frequently found
which is indicated by an overlapping of laminar flow and turbulence.
hv (3-22)
Jg =
l
vmax
vmean
Figure 3.11.-1: Velocity profile at laminar flow
vmax
vmean
Figure 3.11.-2: Velocity profile at turbulent flow
The velocity profile in a pipe is not unique. In the immediate proximity of the pipe wall
the velocity is almost zero, while the liquid reaches a maximum in the center of the pipe.
In addition, the flow profile depends on the Reynolds number. As mentioned before the
laminar flow prevails at very low velocities. The flow profile for laminar flow conditions is
a parabola (Figure 3.11.-1). The highest fluid velocity is in the middle of the pipe while it
steadily decreases towards the outside. The mean speed of a parabolic velocity profile is
approx. 50% of the maximum value.
At turbulent flow conditions the velocity distribution is more proportionate than for laminar
flow as shown in Figure 3.11.-2. The maximum speed occurs - as for laminar flow - in the
center of the pipe. The mean speed, however, reaches fundamentally higher values than
at laminar flow with the exception of a very thin boundary layer near the wall, so that the
resultant mean speed reaches about 80 % to 87 % of the maximum speed in the center
of pipe.
The flow is, in the boundary layer, laminar. The friction losses in the boundary layer can
be calculated by means of equation (3-25). The friction losses outside the boundary layer
are of another nature and arise mainly from shock losses in the turbulent fluid.
1 (3-23)
v mean= ⋅ v max
2
Without examining the derivation of the following equation more clearly, the hydraulic
gradient Jg becomes, in such a case:
32 ⋅ν ⋅ v
Jg = (3-24)
g ⋅ d2
One can express the head loss by inclusion of the Reynolds number in a different fashion:
64 l v 2
hv = ⋅ ⋅ (3-26)
Re d 2 ⋅ g
The expression 64/Re is called the pipe friction coefficient λ.
As already mentioned laminar flow hasn't a great meaning in process industries, since it
doesn’t often appear. Examples for laminar flows are: Highly viscous fluids, suspensions
and flow through very narrow cross-sections. Due to the thick boundary layer with laminar
flow, the small grooves or scores on the pipe walls are covered so that the pipe roughness
doesn't have any influence on the head loss. This effect is contradictory to turbulent
flow where only a very thin boundary layer exists and near the wall there is already a
considerable velocity and head loss.
l v2 (3-27)
hv = λ ⋅ ⋅
d 2⋅g
The friction coefficient λ is, however, not a constant figure but depends, on the one hand
on the Reynolds number Re, and on the other hand on the relative roughness k/d of the
pipe. With very smooth pipes where roughness can be practically neglected, there are
distinguishable flow patterns. For laminar flow holds:
64 (3-28)
λ=
Re
In the turbulent range two different equations apply depending on the actual Reynolds
number. If Recrit ≤ Re ≤ 105 applies:
0.316
λ= 4
(3-29)
Re
For rough pipes there are many possible conditions. At very low Reynolds numbers the
boundary layer is bigger than the roughness peaks. The flow pattern is therefore as in a
smooth pipe and the friction coefficient λ can be relatively simply determined. As already
mentioned the pipe roughness plays no role in laminar flow so here the equation (3-28)
is applied. The other extreme is a completely rough pipe in which the friction coefficient λ
becomes a constant of approx. 0.02. Values in between usually range from 0.02 to 0.04
but are very difficult to determine. They are therefore normally estimated.
An exact determination of the total head loss requires a summation of all single resistances.
One can calculate the head loss in straight pipes according to the equation of chapter
3.13. Special resistances are mostly determined from corresponding tables which show
the empirically determined friction coefficients (Zeta) for the most common standard parts
like T-pieces, elbows or valves. Generally, the head loss of a component can be calculated
according to equation (3-33) when the individual friction coefficient ζ is known:
v2
hv= ζ ⋅ (3-33)
2⋅g
The frequently required pipe components for matching the pipe diameter to the size of the
corresponding valve inlet and/or outlet (e. g. diffusers) represent a special problem for the
determination of the head loss.
One must differentiate here between inlet and outlet head losses. The corresponding
values ζ1 and ζ2 are empirically determined by tests. The total head loss arises from the
algebraic sum of all resistance coefficients of the fittings or customization pieces:
Σζ = ζ 1 + ζ 2 + ζ B1 − ζ B2 (3-34)
If the pipe diameter at the valve inlet and outlet are the same, then the Bernoulli
coefficients B1 and B2 can be ignored, since they fall out from equation (3-34). When
utilizing customary reducers the pipe friction coefficients ζ1 and ζ2 can be determined
with the help of the following equations.
(a) Pipe reduction at the valve inlet only:
2
DN 2
ς 1 = 0.5 ⋅ 1 − (3-35)
D
2
DN 2
ζ 2 =1⋅ 1 − (3-36)
D
(c) Pipe reduction at the valve inlet and pipe extensions at the valve outlet for the same
nominal pipe size before and behind the control valve:
2
DN 2
ζ 1 + ζ 2 = 1.5 ⋅ 1 − (3-37)
D
If the nominal pipe sizes are different before and behind the control valve, then the
Bernoulli coefficients must be calculated separately and included in equation (3-34):
4
DN
ζ=
B 1.0 − (3-38)
D
Sudden pipe extensions or reductions, as schematically shown in Figure 3.14.-1 and
Figure 3.14.-2, should be avoided to keep turbulences and head losses to a minimum.
Diffusers and extenders should have flat openings and reduced angles instead.
v1 v2
v1 v2
The head loss in pipe diffusers or extenders depends further on the ratio of the cross-
sections of valve and pipe respectively and the ratio valve size/pipe size is also authoritative
for the resistance coefficient ζ of the fittings. An exact calculation of the head loss requires,
therefore, the knowledge of the typical resistance coefficients which can be taken from
corresponding tables. As an example the typical resistance coefficients at sudden pipe
reductions for different ratios of cross-sections A1/A2 are shown in Table 3.14.-3 below.
Continuity equation
In a pipeline of variable cross-section the volume flow rate Q is constant throughout all
cross-sections over time t.
A1 ⋅ v1 = A 2 ⋅ v 2 or A ⋅ v = cons tan t
By smaller cross-sections the liquid streams faster and the other way round.
Example:
A1
A2
v1 Q1 Q2 v2
v1 = 1.02 m/s
v2 = 2.49 m/s
Nowadays mostly the Kv or Cv value is used instead of the resistance coefficient ζ. The
relationship between these parameters will be explained later.
px v2
E= + hx + + hv = const.
ρ 2⋅g
This equation says, that the total energy consisting of pressure height + energy of situation
(geodesic height) + velocity energy + the total head loss caused by inner friction always
remains constant. Since the potential energy (geodesic height) can be mostly neglected
in this consideration, the following happens in a pressure reduction process by throttling:
The static pressure energy is changed into velocity energy by a deliberate cross-section
pipe reduction at which it reaches a maximum in the narrowest point (vena contracta).
The accelerated flow meets, behind the throttling area, fundamentally slower liquid
particles or, in an extreme case, the solid walls of the valve body.
This effect causes pressure losses through intense turbulence and an exchange of
energy. The liquid heats. This fact is easy to prove in a pump circuit with a control
valve as throttling element - as e.g. in the determination of the flow coefficient.
Depending on the construction of the control valve, a certain part of the velocity energy
is again converted into static pressure energy. This effect is usually not desired and
can cause cavitation on. The events at the throttling process of a liquid are represented
schematically in Figure 3.16.-1.
The liquid enters the valve with a pressure of p1. A small pressure drop appears already
at the inlet part of the valve body through deflection of the medium. The medium is
then strongly accelerated in the restricted seat area and this is accompanied by a
sharp decline of the static pressure. Behind the throttling area a certain pressure
recovery occurs, depending on the amount of deceleration. Finally, a small pressure
drop at the valve outlet occurs caused by repeated deflections of the fluid. The actual
measured differential pressure is the total pressure loss between inlet and outlet of
the valve.
W ⋅ ( p1 − p2 )
Pth ≅ (3-39)
36 ⋅ ρ
A concrete example will explain this principle idea of the converted power in a control
valve. A throttling valve is regulating the feed water supply in a closed loop on a boiler.
Given are the following process data:
Flow rate W : 120,000 kg/h
Inlet pressure p1 : 80 bar
Outlet pressure p2 : 2 bar
Valve size : DN 100
Density ρ : 1000 kg/m3
Converted power P : Approx. 260 kW
As can be seen from the example above, almost the complete power consumption of the
pump is used up by the valve itself during the start-up of the boiler. The extremely high
energy density within a relatively small control valve is remarkable and requires special
measures to keep wear and tear caused by cavitation and/or erosion to a minimum.
Figure 3.18.-2: Mechanical construction of Centrifugal Pump (Foto KSB Type Etanorm RSY)
Pump fundamentals:
System Curves
Static Head
Dynamic Head
Pipe Friction
Fitting Losses
System Head = Static Head + Dynamic Head
NPSH
Net Positive Suction Head (NPSH)
NPSH required (NPSHr)
NPSH available (NPSHa)
NPSH is a measure of the energy (pressure) in a liquid above the vapor pressure
If the pressure drops below the vapor pressure the liquid boils
That condition is called cavitation
All pumps require the NPSHa to be > 0
How much, is called the NPSHr
Pump Curves
When the pump head discharge curve and the system head curve are plotted on the same
axes (as shown in Figure 3.18.-3), only one point lies on both the pump characteristic
curve and the system head curve. This intersection defines the pump operating point,
which represents the discharge that will pass through the pump and the head that the
pump will add. This head is equal to the head needed to overcome the static head and
other losses in the system.
Pu m p c h a r a c t
eristic
C u rv
e
Head Losses
Cu rve
System
Static Lift
Flow rate Q
Figure 3.18.-3: System operating point
In addition to the pump head-discharge curve, other curves representing pump behavior
describe power, water horsepower, and efficiency (see Figure 3.18.-4). Since utilities want
to minimize the amount of energy necessary for system operation, the engineer should
select pumps that run as efficiently as possible.
Another issue when designing a pump is the net positive suction head (NPSH1) required.
NPSH is the head that is present at the suction side of the pump. Each pump requires that
the available NPSH exceed the required NPSH to ensure that local pressures within the
pump do not drop below the vapor pressure of the fluid, causing cavitation. Cavitation is
essentially a boiling of the liquid within the pump, and it can cause tremendous damage.
The NPSH required is unique for each pump model, and is a function of flow rate.
Net positive suction head is the term that is usually used to describe the absolute pressure
of a fluid at the inlet to a pump minus the vapour pressure of the liquid. The resultant value
is known as the Net Positive Suction Head available.
The term is normally shortened to the acronym NPSHa, the ‘a’ denotes ‘available’. A
similar term is used by pump manufactures to describe the energy losses that occur
within many pumps as the fluid volume is allowed to expand within the pump body.
This energy loss is expressed as a head of fluid and is described as NPSHr (Net Positive
Suction Head requirement) the ‘r’ suffix is used to denote the value is a requirement.
Different pumps will have different NPSH requirements dependant on the impellor design,
impellor diameter, inlet type, flow rate, pump speed and other factors.
A pump performance curve will usually include a NPSH requirement graph expressed
in metres or feet head so that the NPSHr for the operating condition can be established.
Eff
Pump Efficiency, %
ic i
en
cy
cu
r
ve
Pu m
p dis BEP
cha
rge (Best Efficiency Point)
h ea
Discharge head
)
Head
on
e Sucti
NPSH
tiv
(Net Posi
NPSH
Flow rate Q
Figure 3.18.-4: Pump efficiency curve
Discharge
Suction
Engine
Pump
Figure 3.18.-5: Pump Operating
curve will usually include a NPSH requirement graph expressed in metres or feet head so
that the NPSHr for the operating condition can be established.
Discharge
Positive
head
Negative
head
NPSHr energy loss
occurs within the
pump inlet system
Energy loss occurs
due to friction in the
pipework or within the
fluid.
Figure 3.18.-6: Schematic for NPSHa
Fundamentals
Gage Pressure (psig) Pressure above surrounding atmospheric pressure.
Atmospheric pressure at sea level is 14.7 psig
Absolute Pressure (psia) Pressure above an absolute vacuum.
NPSHa calculation
The elements used to calculate NPSHa are all expressed in absolute head units.
The NPSHa is calculated from:
Fluid surface pressure + positive head – pipework friction loss – fluid vapor pressure
or
Fluid surface pressure - negative head – pipework friction loss – fluid vapor pressure.
Pipework Pipework
friction friction
Fluid Fluid
surface surface
pressure Fluid pressure
vapour
pressure
Fluid
Zero NPSHa Zero vapour
pressure NPSHr pressure pressure NPSHr
absolute absolute NPSHa
NPSHa is sufficient to avoid cavitation NPSHa is too low cavitation will occur
In a system where the fluid needs to be lifted to the pump inlet , the negative head reduces
the motive force to move the fluid to the pump.
In these instances it is essential to size the supply pipe work and isolating valves
generously so that high frictional losses do not reduce the NPSHa below the NPSHr .
Negativ Negativ
Head Head
NPSHa is sufficient to avoid cavitation NPSHa is too low cavitation will occur
Abstract:
At first sight, parallel operation of pumps does not seem to pose any
problems. Under real conditions, however, parallel pumping proves
to be rather more complex than typically portrayed in the relevant
literature.
So as to ensure that pumps and systems produce the required
flow rates and pressures detailed information is needed about the
real curve of both the pumps and the system. The duty point of the
individual pumps operated in parallel will shift and this will affect both
efficiency and NPSH performance. Only if these changes are taken
into account will a parallel pumping system function properly.
Characteristic cu
H0 rve Pump
I + Pump
Chara II
cteri
stic c Operation Point
urve
Pum parallel
p I or
H Pu
mp
II Operation Point
single
System curve HA
Head H
Qsingle
Qparallel
Qp a ra lle l Qp a ra lle
=l QI + QII
Q
=I Q=
II
2
Flow rate Q
Figure 3.19.1.-2: Parallel operation of two identical centrifugal pumps with stable characteristic curve
3 pumps
in para
2 pum llel
ps in
paral
lel
Head H
Pu
mp
I
Discharge
Suction
Main
Discharge
Suction
Booste
Series Operation
This arrangement is frequently used where a larger pump cannot operate with the NPSH
being made available from the system. A smaller pump is therefore installed upstream of
the larger one to boost the Suction Pressure to the larger pump.(see Figure 3.19.2.-2
and Figure 3.19.2.-3)
It should be noted that in the series operation, the only essential similarity between the
two pumps is that they must both be able to operate at the same flow rate. The pumps
can deliver totally different levels of head as long as they operate at the same capacity.
The ultimate example of series operation is the multistage pump where the first impeller
pumps into the second and then the third, etc. This results in a high pressure pump with
all the impellers operating at the same capacity.
In some pumping arrangements we can have multiple pumps that operate on two systems
that are closely combined, but are not in a true series operation and cannot be considered
as such.
One such example would be when one pump is supplying the flow and pressure for one
system, while the second pump is bleeding off the first system to deliver a lesser flow to
another system. While one pump is indeed discharging into the suction of the second
pump, it is doing so at a different flow rate. Consequently, they cannot be treated as
though they were operating in series.
Head H
Pump I + I
I
η
ien cy
fic
Pump I Ef
H dynamic
rve
cu NPSH
tem
Sys
~70% 100%
Flow rate Q
Falling Flow rate with
outage of one pump
Figure 3.19.2.-2: Pumps series operation with geodetic heights = 0
Head H
Pump I + II
rve
cu
H dynamic
m
ste
Pump I Sy
η
y
H geodesic
nc
ie
fic
NPSH
Ef
~40% 100%
Flow rate Q
Falling Flow rate with
outage of one pump
Head H
H0 II Pum
p II
H0 I Pum Pum
pI p I+I
I
System characteristic
Qmin
Series Pumping
Heads H add at the same flow rate Q
Second stage pump must be rated for discharge pressure
Start up and shutdown procedures are critical
Head H
2 Pumps in series
Single Pump
Flow rate Q
Parallel Pumping
2 Pumps parallel
Single Pump
Flow rate Q