Algorithm For Pump Selection-2003
Algorithm For Pump Selection-2003
S. L. Abhyankar
One would often select a centrifugal pump by default, because most thinking of pumps focuses on
centrifugal pumps. This is so, because centrifugal pumps are made in such majority over other types of
pumps. However an algorithm should rather ensure that there is no default section. This algorithm hence
provides for checking whether the application would warrant a pump other than a centrifugal pump. In
short, it takes into account the limitations of centrifugal pumps. From the algorithm, one would realise
that centrifugal pumps do have many limitations.
An algorithm is better readable as a flowchart. An algorithm has been attempted and is at Annex 1.
Notes below explain the logic of the algorithm.
Pump selection should start with data on hand about some most common parameters of pumping
duty, viz. head and/or pressure, flow-rate required, preferred speed, number of pumps.
There are statutory codes to be followed, especially when selecting pumps for fire fighting. Fire-
fighting pumps (IS-12469) have nominal discharge ratings specified in the code itself. Depending upon
the degree of hazard, one has to select a discharge rating from the nominal ratings given in the code.
For pumping water to the overhead reservoir in a multistorey residential building, number of hours of
inflow of municipal supply, capacity of Ground Level Reservoir (GLR) and total daily requirement of the
residents will together decide the flow-rate. E.g., if there are 50 families in a building, i.e. a population of
250 people at an average of 5 persons per family, then at 200 litres per capita per day, daily requirement
3
becomes 50 m . If number of hours of inflow of municipal supply is 3 hours and the capacity of GLR is
3 3
only 20 m , balance of the total requirement i.e. 30 m will have to be pumped to OHR in the 3 hours of
3
inflow. So required flow-rate of pumping becomes 10 m /h.
In a sewage pumping station, flow-rate for pumps has to vary at different times of the day, depending
upon the rate of inflow of sewage, size of sump, and time within which sewage should be pumped so that
it does not become septic, also taking care that the pump should not suffer too many starts and stops at
too short intervals. Traditionally this is managed by running different number of pumps in parallel. One
may as well have pumps of different flow –rates to run in parallel. Variable Speed Drives (VSD) make a
good, energy- saving alternative for such regulation of flow-rate.
Total head primarily comprises the level difference and/ or pressure difference and the hydraulic
friction. Hydraulic friction depends upon the selection of types and sizes of pipes and fittings.
All these data on the most common parameters of pumping duty may often prompt the default
selection of a centrifugal pump. However, there are many other parameters to be considered to decide
the type of pump.
Wear life, bearing life, preference for a compact design and variety reduction influence selection of
operating speed of the pump. Abrasive and corrosive wear will be more pronounced at high operating
speed. Bearing life will also be less at high operating speed. But pumps designed to run at high
operating speed are compact and space-saving and are better portable. If pumps have to operate even
optional drive from internal combustion engines, or if standby pumps for emergency operation have to be
with internal combustion engines, variety of pumps and carrying of inventory of spares can be less if both
electrically-driven and I. C. engine-driven pumps are of common design.
1.1 If yes, select a positive displacement pump Check if viscosity is shear sensitive?
1.1.1 If yes, select positive displacement pump, where liquid will not suffer shear at the pumping
element. A helical rotor, progressive cavity pump or a diaphragm pump may be considered.
4. Is multi-phase medium to be handled e.g. oil, gas and mud in oil exploration? If yes, consider
piston rod or plunger pumps. Pumps called as sucker rod pumps in the oil Industry are
basically plunger pumps.
5.1 If yes, select a positive displacement pump, but get back to step 2.
Calculate the specific speed from the pumping duty, viz. Q, H and rpm. If pressure is dominant factor
of the duty, convert pressure to equivalent head to find the specific speed.
6.1 Check if running at higher speed acceptable, so that higher speed will make higher specific speed
and will make a centrifugal pump feasible. Then go to 5.6.
6.2 Check if liquid is clear, i.e. if turbidity is <500 ppm, if yes, check whether multistage construction
is acceptable, so that specific speed per stage will again be higher.
6.3 If checks (5.1) and (5.2) are negative, and liquid is clear, check if pulsating flow is acceptable.
6.3.1 If pulsating flow is acceptable, a positive displacement pump can be more energy-efficient
than any impeller pump. But with thin liquids, there will be slip flow, which will cause the
efficiency of pumps to be less than normal.
6.3.2 If pulsating flow is not acceptable, one may consider a regenerative turbine type pump.
But these pumps are not energy-efficient. But the advantage is that they can run at high
running speed and are in turn compact, have non-pulsating flow, and also have a fair
degree of self-priming capability. Check whether this capability has cost-benefit advantage
in the application.
6.4 Further option for selecting centrifugal pumps is to select the pump for higher discharge and bye-
pass the excess flow back to suction. This again will not be energy-efficient.
6.5 Selecting from feasible options should be by Life Cycle Cost (LCC) analysis.
6.6 Specific applications where helical rotor, progressive cavity pumps, metering or dosing pumps,
peristaltic pumps, diaphragm pumps would become selection options as have been mentioned
above. Where the algorithm lends selecting positive displacement pump as a general option,
various types of positive displacement pumps can be considered, viz. piston or plunger pumps,
internal gear pumps, external gear pumps, lobe pumps, vane pumps, flexible vane pumps,
multiple screw pumps, etc. Again, selecting from feasible options should be by Life Cycle Cost
(LCC) analysis.
Most common arrangement of a pump is with its working axis horizontal. Centrifugal pumps with
horizontal working axis often operate with a suction lift, unless the pump is to draw the liquid from a
suction vessel above the pump.
(a) NPSH available may not be adequate. If so, mounting the pump to be submerged in the liquid will
improve NPSH available. Vertical turbine pumps become an option to large mixed flow or axial flow
end suction pumps or axially split casing horizontal pumps. But dismantling vertical turbine pumps for
overhauling becomes quite an exercise. Also it becomes important to ensure structural stability, non-
resonant frequency and limited and controlled vibration in case of vertical turbine pumps.
(b) Mounting more than one pump in a pit would warrant proper design of sump.
(c) Volatile or high temperature liquids with high vapour pressure would warrant pump to be mounted in
a vertical can with enough depth, to take care of NPSH required by the pump in the depth of the can
itself.
(d) Vertical mounting often makes a compact, space-saving installation. Vertical in- line pumps are an
eminent example.
(e) Submersible pumps in tube wells are vertical pumps. Most pumps for drainage and dewatering are
also vertical, since they then are compact and hence portable.
(f) Submerging pumps in the liquid, often called as ‘wet pit’ mounting has become feasible with the
development of submersible motors.
TABLE 1
Canned Motor Pumps Are truly zero leak. Detailed consideration needed, if liquid
does not have lubricity or pumping temperature is high.
Part of discharge is circulated through the motor. This
reduces efficiency.
Magnetically coupled pumps Are truly zero leak. Detailed consideration needed, if liquid
does not have lubricity or pumping temperature is high.
Magnetic induction itself generates heat, causing
transmission losses and reducing overall efficiency.
TABLE 2
7. Shaft-sealing:
See Table 1.
7.1 Traditionally seal – manufacturers supply seals to be assembled by the user. Now seals are
also available as ready, assembled seals, assembled in a cartridge. Seals are also available as split
seals for in situ installation or replacement without the need to dismantle the pump. API-682 details
various plans for using seals. When to use which plan, is detailed in Table 2.
8. Heating Jacket:
If the liquid has high melting point and in turn, a tendency to solidify a ambient, the pump will not
restart after shut down. A heating jacket will help to liquefy the solidified mass blocking the suction.
9. Cooling jacket:
9.1 When pumping temperatures are high, to prevent the shaft seal from malfunctioning and bearings
having short life, cooling will be needed around stuffing box and bearing housing.
9.2 If pumping duty is less than safe minimum flow, pump will experience rise in temperature, vapour
pressure will be higher and pump may suffer cavitation. To control the rise in temperature, cooling
jacket will be necessary for the pump casing also.
The main considerations behind selecting material of construction for different components are to
have
Corrosion, abrasion, cavitation-erosion and liquid temperature will primarily affect the ‘wetted’
components of the pump. In case of centrifugal pumps, ‘wetted’ components include the casing impeller,
impeller nut, casing cover and seal housing, shaft sleeve, gaskets, shaft seal material for whichever shaft
sealing arrangement is selected
Corrosive effect of a liquid is mainly due to alkalinity or acidity. Liquids with pH value between 6.5 and
7 are neutral and hence are most comfortable to handle with commonplace material like Cast Iron. Until
the development of stainless steels, for alkaline corrosion, especially for alkalinity as of sea water copper-
based alloys like Naval Bronze were much in vogue. Many stainless steel materials however have good
corrosion resistance both against alkaline and acidic corrosion.
Many commonly used materials are better known by acronyms or codes. For the sake of accuracy,
one should be conscious of finer distinctions in the codes. For example, 18-8 Cr-Ni stainless steel is
commonly known as SS-304. However, since components like casing and impeller are made by casting
process, the code name is CF8, derived from grade CF8 in specification A-351 of American society of
Testing and materials (ASTM). American Iron and Steel Institute (AISI) designated same material in
wrought form as Grade 304.
Corrosion-resistance and abrasion-resistance do not go together. Abrasion-resistance would require
high degree of hardness. Alternatively if the abrasive particles would impinge on the boundary surfaces
at some angle of impedance, resilient materials would cause the abrasive particles to bounce off and not
cause wear. Abrasion with low or zero angle of impedance becomes erosion. So, for abrasion-
resistance, one should choose hard materials for erosive wear and resilient materials for bouncing
particles, e.g. particles of sand. For components like pump casing resilient materials like rubber are better
provided as linings in a cast casing.
Often components running close to each other, even when made of appropriate corrosion-resistant
materials will suffer galling and electrolytic corrosion because of their chemistry being identical. A
differential hardness or difference in chemistry of the close running surfaces can reduce or eliminate this
galling and corrosion. Simple example to note this is by realising that a piece of glass moves harder over
another surface of glass, than how hard a piece of steel would move on a glass surface.
When flow to be pumped needs regulation, one can get increased flow by running more pumps to run
together in parallel. Switching off some of the pumps can reduce the flow. The flow is however not
directly proportional to the number of pumps running. Study of system characteristics along with pumps’
characteristics is necessary to devise a trouble-free system for pumps running in parallel. It is also
necessary that H-Q characteristics of individual pumps to run in parallel should be continuously rising
towards shut-off. Also shut-off head of pumps should be identical as far as possible.
Most situations of pump selection can provide opportunity to explore options. Most economic
selection would emerge by Life Cycle Cost analysis.
Where
PSC = Pumping System Cost as the sum total of costs of Pump, Driver, Controller, Coupling, Piping,
Valves, etc. Options in sizes of pipes and fittings should be considered as independent options of the
Pumping System.
ICC = Installation and Commissioning Cost inclusive of costs of excavations and building of sump
and pump room and material handling tackle, on-site assembly and testing
OC = Operating Cost over the life. Period for all options should be life of that option, which has
longest life.
MIDC = Maintenance, Inventory and Downtime Cost over the same period as for Operating Cost.
Reference:
Indian Pumps December 2003