OPERATING ENVELOPES FOR CENTRIFUGAL PUMPS

In today’s global economy, a necessity exists in Problem Statement

process facilities to reduce costs and upgrading A Centrifugal pump is used to transfer water from
component specifications do not necessarily a horizontal vessel to a storage tank as shown in
provide fail proof solutions. Of all the equipment Figure 2.
used for both commercial and industrial
applications, centrifugal pumps are a common
feature. They are characterized by their high
efficiency with low power consumption.
To cite a few applications, centrifugal pumps are
used in buildings for water supply, as a booster
and for domestic water supplies, pumping of
sewage and slurries. They are also used in fire Figure 2. Pump System Schematic
protection systems and for heating and cooling The centrifugal pump has the following Head (H)
applications. In addition, they are popular in the vs. Flow (Q) characteristics as shown in Figure 2
beverage, dairy, food and oil & gas, and operates with a flow control valve at its
petrochemical & chemical industries. discharge line. The flow rate required to be
Improper operation of centrifugal pumps, often maintained is 189.2 m3/h with the receiving tank
result in mechanical integrity failures such as, pressure at 1.02 bara. The pump suction
high temperatures, low flow cavitation, Low conditions are as follows,
bearing and seal life, reduced impeller life, Table 1. Pump Suction Conditions
suction and discharge recirculation. Parameter Value Unit
To circumvent such operational failures, Pump 0
Liquid Temperature [T] 20.0 C
Operating Envelopes are made for a given pump
specification to enable engineers and operators Liquid Density [] 997.8 kg/m3
to make decisions regarding its operability. The Dynamic Viscosity [μ] 1.00 cP
below picture shows the various possible failures
for the range of operation. 0.001 kg/m.s

Vapour Pressure [Pv] 0.023 bara

Critical Pressure [Pc] 217.7 bara

Pump Operating Flow [Q] 189.2 m3/h

0.05256 m3/s

Pump NPSHR 0.69 bara

7.05 m

Figure 1. Pump Operating Failures [2] Discharge Head [H] 6.29 bar

The following tutorial provides a step by step 64.24 m

procedure to predict the allowable operating
The piping details of the pump system piping
range or “Operating Envelope” for a centrifugal
connections are as follows,
pump’s range of operation.
Page 1 of 9
 Table 2. Pump System Piping Details The pump curves of the centrifugal pump are,
Parameter Value Unit Table 4. Pump Curves

Suction Pipe Size [ND] 8.625 in Flow Head Head

Suction Pipe [WT] 8.18 mm [m3/h] [bar] [m]

Suction Pipe ID [ID] 0.2027 m 0.0 9.51 97.1

Suction Pipe CS Area [A] 0.03227 m2 93 8.76 89.5

100 8.68 88.7

Suction Pipe Roughness [ε] 45.2 μm
120 8.41 85.9
Suction ε/D 0.00022 -
140 8.01 81.8
Suction Pipe Length 5 m
160 7.45 76.1
Discharge Pipe Size [ND] 6.625 in
172 7.03 71.8
Discharge Pipe [WT] 7.11 mm
180 6.70 68.5
Discharge Pipe ID [ID] 0.1541 m
200 5.73 58.5
Discharge Pipe CS Area [A] 0.01864 m2
220 4.49 45.8
Discharge Pipe Roughness [ε] 45.2 μm
232 3.61 36.9
Discharge ε/D 0.00029 -
The equipment operating pressures and elevation
Discharge Pipe Length 500 m from pump centerline details are as follows,
The pump system has piping components/fittings Table 5. Pump Suction Conditions
like block valves, check valves, flow elements, Y- Parameter Value Unit
strainers, bends, elbows, concentric reducers,
etc. which add a dynamic pressure loss to the Suction Vessel Gauge Pressure 0.02 barg
pump flow. These minor pressure/head losses 1.04 bara
can be estimated by using the relationship,
Suction Vessel Elevation 2.00 m
𝐾𝑉 2
ℎ𝐿 = (1)
2𝑔 Suction Vessel Low Low LL 0.25 m
Where, K is the Loss Coefficient and can be Discharge Vessel Pressure 1.02 bara
estimated from sources such as Crane’s
Handbook. In this tutorial, the total minor head Discharge Vessel Elevation 1.00 m

loss is assumed to be 0.5 bar. Additionally as the Discharge Vessel High High LL 4.00 m
liquid exits the discharge piping, the exit losses is
taken to be 5 m (~0.04 bar) based on Eq. 1 The control valve details to be checked for are,
Table 6. Control Valve Details
Table 3. Minor Losses
Control Valve Cv@100% Opening FL
Parameter Value Unit
4” Size 236 0.82
PMinor Losses 0.5 bar
6” Size 433 0.84
PExit Losses 0.04 bar

Page 2 of 9
Pump System Pressure Losses 0.0163×5×997.8×1.632
∆𝑃 = = 0.0053 𝑏𝑎𝑟 (5)
2×0.202715
The pump operating point is the point where the
system resistance curve intersects the pump The results for suction side pressure drop can be
performance curve. This can be understood as, summarized as,
for a given flow rate, the resultant pump head Table 7. Suction Side Frictional Losses
arrived at, after accounting for all the static and
Parameter Value Unit
dynamic losses (represented as system
resistance) from the generated pump Total Reynolds Number [Re] 329,372 -
Dynamic Head (TDH).
Flow Behaviour Turbulent -
The factors contributing to system resistance is,
Friction Factor Equation Swamee-Jain -
1. Suction Piping & Fittings Frictional Loss
Friction Factor 0.0163 -
2. Discharge Piping & Fittings Frictional Loss
Suction ε/D 0.00022 -
3. Control Valve Pressure drop
Pressure Drop [ΔP] 0.0053 bar
4. Total Static Head
Suction & Discharge Frictional Pressure Losses The calculations are similarly performed for
discharge side with a discharge side velocity of
To estimate the total static head, the worst case
scenario of Low Low Liquid Level (LLLL) in the 2.82 m/s and the results are summarized below,
suction vessel & High High Liquid Level (HHLL) in Table 8. Discharge Side Frictional Losses
the receiving tank is considered. This is taken so Parameter Value Unit
to ensure that during LLLL, the pump does not
suffer from cavitation due to Net Positive Suction Reynolds Number [Re] 433,408 -
Head (NPSH) deficiency, i.e., even during LLLL, Flow Behaviour Turbulent -
NPSHA > NPSHR. To estimate NPSHA, the suction
line frictional pressure drop is calculated using Friction Factor Equation Swamee-Jain -
the Darcy-Weisbach equation (Appendix B). The Friction Factor 0.0165 -
Suction Velocity (VS) is estimated as,
𝑄 0.05256 Suction ε/D 0.00029 -
𝑉𝑠 = = ≈ 1.63 𝑚/𝑠 (2)
𝐴𝑠 0.03227
Pressure Drop [ΔP] 2.123 bar
The Suction Side Reynolds Number is,
0.202715×1.63×997.8
Total Static Head
𝑅𝑒 = = 329,372 (3)
0.001 The total static head is computed as,
As the calculated suction side Reynolds number is [(ℎ𝐻𝐻𝐿𝐿 +ℎ𝑒𝑙𝑒𝑣 )𝑇𝑎𝑛𝑘 −(ℎ𝐿𝐿𝐿𝐿 +ℎ𝑒𝑙𝑒𝑣 )𝑉𝑒𝑠𝑠𝑒𝑙 ]×𝜌×𝑔
∆𝑃𝑆 = (6)
much higher than 4000, the flow is well into the 105

turbulent region. The friction factor can now be [(4+1)−(0.25+2)]×997.8×9.81

∆𝑃𝑆 = ≈ 0.27 𝑏𝑎𝑟 (7)
105
calculated using Colebrook equation but owing to
its implicit nature, the friction factor is calculated Therefore the pump’s NPSHA is calculated as,
using Swamee-Jain correlation (Appendix B), 𝑁𝑃𝑆𝐻𝐴 = 𝑃𝑆 + 𝑃𝑠𝑢𝑟𝑓𝑎𝑐𝑒 − 𝑃𝑣 − 𝑃𝐹𝑟𝑖𝑐 (8)
0.25
𝑓= 0.00022 5.74 = 0.0163 (4) 𝑁𝑃𝑆𝐻𝐴 = 0.27 + 1.04 − 0.023 − 0.0053 (9)
log10 [ + ]
3.7 3293720.9

Therefore the frictional pressure drop becomes,

𝑁𝑃𝑆𝐻𝐴 = 1.23 𝑏𝑎𝑟𝑎 (10)

Page 3 of 9
 −1⁄
The calculated NPSHA is 1.23 bara and is higher 0.726 236 2 2
than NPSHR of 0.69 bara. The pump suction 𝐹𝑃 = [1 + ( 42 ) ] = 0.9216 (26)
890
pressure is calculated as, −1⁄
0.28+0.937 236 2 1 2
(ℎ𝐿𝐿𝐿𝐿 +ℎ𝑒𝑙𝑒𝑣 )𝑉𝑒𝑠𝑠𝑒𝑙 ×𝜌×𝑔 𝐹𝐿𝑃 = [ ( 42 ) + 0.822 ] = 0.8197 (27)
𝑃𝑠 = 𝑃𝑣𝑒𝑠𝑠𝑒𝑙 + − 𝑃𝐹 (11) 890
105
The inlet pressure in psig for choke flow equation
(0.25+2)×997.8×9.81
𝑃𝑠 = 1.04 + − 0.0053 (12) is calculated as by considering pressure drop
105
between Pump Discharge and Control valve Inlet
𝑃𝑠 ≈ 1.25 𝑏𝑎𝑟𝑎 (12.78 𝑚) (13)
is very small as,
The Pump Discharge Pressure for the flow rate of
𝑃1 = (7.54 × 14.7) × 14.7 = 96.1 𝑝𝑠𝑖𝑔 (28)
189.2 m3/h & corresponding pump head of 6.29
bara is calculated as, 𝑃2 = 96.1 − (3.59 × 14.7) = 43.4 𝑝𝑠𝑖𝑔 (29)

𝑃𝑑 = 𝑃 + 𝑃𝑠 = 6.29 + 1.25 (14) ∆𝑃𝑠𝑖𝑧𝑖𝑛𝑔 = 96.1 − 43.4 = 53 𝑝𝑠𝑖 (30)

𝑃 0.34
𝑃𝑑 = 7.54 𝑏𝑎𝑟𝑎 (15) 𝐹𝐹 = 0.96 − 0.28√𝑃𝑣 = 0.96 − 0.28√3200 (31)
𝑐
Control Valve Pressure Drop
𝐹𝐹 = 0.957 (32)
The control valve pressure drop is calculated as,
0.8197 2
∆𝑃𝑐ℎ𝑜𝑘𝑒𝑑 = [ ] [(96.1 + 14.7) − 0.957 × 0.34] (33)
∆𝑃𝐶𝑉 = 𝑃𝑑 − 𝑃𝐹+𝑠+𝑚𝑖𝑛+𝑒𝑛𝑡 − 𝑃𝑣𝑒𝑠𝑠𝑒𝑙 (16) 0.9717

∆𝑃𝑐ℎ𝑜𝑘𝑒𝑑 = 87 𝑝𝑠𝑖 (34)

𝑃𝐹+𝑠+𝑚𝑖𝑛+𝑒𝑛𝑡 = 𝑃𝐹 + 𝑃𝑠 + ∆𝑃𝑚 + ∆𝑃𝑒 (17)
Since Psizing  PChoked, then Psizing = 53 psi
∆𝑃𝐶𝑉 = 7.54 − 2.123 − 0.27 − 0.5 − 0.04 −
189.2×4.4028675
1.02 ≈ 3.59 𝑏𝑎𝑟𝑎 (18) 𝐶𝑣 = = 124.3 𝑔𝑝𝑚/√𝑝𝑠𝑖 (35)
53
1×0.9216√
0.9978
Total Dynamic Losses
The total dynamic losses is computed as, Reinserting the calculated Cv value of 124.3, the
value of Fp, FLP and new Cv is re-computed
𝑇𝐷𝐻 = 𝑃𝐹 + 𝑃𝑠 + ∆𝑃𝑚 + ∆𝑃𝑒 + ∆𝑃𝐶𝑉 (19) iteratively,
𝑇𝐷𝐻 = 0.005 + 2.12 + 0.27 + 0.5 + 0.04 + Table 9. FP, FLP & Cv Iterations
3.59 ≈ 6.52 𝑏𝑎𝑟𝑎 (20)
Iteration FP FLP Cv
Control Valve Cv Required
1 0.92 0.82 124.3
The required Cv of the control valve is calculated
by estimating the valve coefficients first followed 2 0.98 0.82 117.4
by checking if choked flow exists (Appendix A), 3 0.98 0.82 117.1
𝑑 4 4 4
𝐾𝐵1 = 1 − (𝐷 ) = 1 − [7.981] = 0.937 (21) 4 0.98 0.82 117.1
1

𝑑 4 4 4
5 0.98 0.82 117.1
𝐾𝐵2 = 1 − (𝐷 ) 1 − [6.065] = 0.811 (22)
2

42
2 6 0.98 0.82 117.1
𝐾1 = 0.5 × [1 − (7.9812 )] = 0.28 (23)
2 7 0.98 0.82 117.1
42
𝐾2 = 1.0 × [1 − (6.0652 )] = 0.319 (24)
8 0.98 0.82 117.1
∑ 𝐾 = 0.28 + 0.319 + 0.937 − 0.811 = 0.726 (25)
9 0.98 0.82 117.1
The selected control valve is 4” valve with Cv of
236 and FL of 0.82 from Table 6. 10 0.9789 0.8198 117.1

Page 4 of 9
Therefore for a flow of 189.2 m3/h, the Cv control valve. If there are no fittings, Fp is taken
required is 117.1 gpmpsi. For the 4” valve to be 1.0. The term K is the algebraic sum of the
selected, the % Cv becomes velocity head loss coefficients of all the fittings
117.1 that are attached to the control valve & is
% 𝐶𝑣 = = 49.6 % (36)
236 estimated as,
Pump Operating Envelope ∑ 𝐾 = 𝐾1 + 𝐾2 + 𝐾𝐵1 − 𝐾𝐵2 (38)
To generate the pump operating envelope, the
Where,
above set of calculations is performed for various
% Cv between 20% to 80% to estimate the total K1 = Upstream fitting Resistance Coefficient [-]
dynamic head at various flows. Below is the K2 = Downstream fitting Resistance Coefficient [-]
pump performance curve that includes the
KB1 = Inlet Bernoulli Coefficient [-]
system resistance curves for various % Cv & flow
rates. KB2 = Outlet Bernoulli Coefficient [-]
Where,
𝑑 4
𝐾𝐵1 = 1 − (𝐷 ) (39)
1
𝑑 4
𝐾𝐵2 = 1 − (𝐷 ) (40)
2
Where,
D1 = Pipe Inlet Diameter [in]
D2 = Pipe Outlet Diameter [in]
If the upstream and downstream piping are of
Figure 3. Pump Operating Envelope equal size, then, KB1 = KB2, and therefore, are
Appendix A – Liquid Control Valve Sizing [1] dropped from the ΣK equation. If the
Based on ANSI/ISA S75.01.01 Standards, to size downstream pipe size is similar to upstream pipe
the control valve, the following Liquid control size, i.e., D1 = D2, then KB1 = KB2. The most
valve sizing set of equations can be used, commonly used fitting in control valve
installations is the short-length concentric
Step 1: Calculate Piping Geometry (Fp) & Liquid
reducer. The equations for these fittings are,
Pressure Recovery Factor (FLP)
2
−1⁄ 𝑑2
∑𝐾 𝐶 2 2 𝐾1 = 0.5 × [1 − (𝐷2 )] , for inlet reducer (41)
𝐹𝑃 = [1 +
𝑁2
(𝑑𝑉2) ] (37) 1
2
𝑑2
𝐾2 = 1.0 × [1 − (𝐷2 )] , for outlet reducer (42)
2
Where,
If the concentric reducers installed on either side
Fp = Piping geometric Factor [-]
of the control valve are identical, then
N1 = Constant [Value = 1.0] 𝑑2
2
∑ 𝐾 = 𝐾1 + 𝐾2 = 1.5 × [1 − ( 2 )] (43)
𝐷
N2 = Constant [Value = 890]
If the concentric reducers installed on either side
Cv = Valve Coefficient [GPM/psi]
of are identical, then, K = K1 + K2. The liquid
d = Control Valve Size [inch] Pressure Recovery Factor (FLP) is calculated as,
−1⁄
The value of Fp is dependent on the fittings such 𝐾1 +𝐾𝐵1 𝐶𝑣 2 1 2
as reducers, elbows or tees that are directly
𝐹𝐿𝑃 = [ (𝑑2) + 𝐹2 ] (44)
𝑁2 𝐿
attached to the inlet & outlet connections of the
Page 5 of 9
Step 2: Calculate Pressure Drop Required for to estimate the final value of Cv. If the FL value
Sizing (PSizing) were to change between iterations, these values
would need to be updated, and Cv re-calculated.
To estimate the pressure drop required for sizing,
Psizing, first the liquid critical pressure ratio (FF) is Appendix B – Line Sizing [3]
calculated. Therefore, Pressure loss in piping without any size changes
𝐹𝐹 = 0.96 − 0.28√𝑃𝑣
𝑃
(45) or fittings occurs due to friction between the fluid
𝑐 and the pipe walls. To estimate the piping
Where, pressure loss, the Darcy-Weisbach correlation is
FF = Liquid Critical Pressure Ratio [-] used as follows,

Pv = Vapour Pressure [psia] 𝑓×𝐿×𝜌×𝑉 2

∆𝑃 = (49)
2𝐷
Pc = Critical Pressure [psia]
Where,
Using the value of FF, Pchoked is calculated as,
P = Pressure drop [bar]
𝐹𝐿𝑃 2
∆𝑃𝑐ℎ𝑜𝑘𝑒𝑑 = [ 𝐹 ] [𝑃1 − 𝐹𝐹 𝑃𝑣 ] (46) f =Darcy Friction Factor [-]
𝑃

If PValve  PChoked, then P = PSizing L = Pipe Length [m]

Else, Repeat calculations for next size.

 = Fluid Density [kg/m3]
V = Fluid Velocity [m/s]
Step 3: Calculate Required Control Valve Cv
D = Pipe Inner Diameter, ID [m]
The required control valve Cv is calculated as,
The Darcy friction factor may be determined by
𝑄
𝐶𝑣 = (47) either using the appropriate friction factor
∆𝑃𝑆𝑖𝑧𝑖𝑛𝑔
𝑁1 𝐹𝑝 √ 𝜌1 correlation, or by reading from a Moody Chart
[ ⁄𝜌0 ]
which is a function of the Reynolds number (Re).
Or, estimating in terms of pressure drop gives us, 𝐷𝑉𝜌
𝑅𝑒 = (50)
µ
2
𝑄 𝜌
∆𝑃 = [
𝐶𝑣 𝑁1 𝐹𝑝
] [ 1⁄𝜌0 ] (48) Where,
µ = Dynamic Viscosity [kg.m/s]
Where,
The Darcy Friction Factor [f] depends on the
Q = Flowrate [gpm] Reynolds number follows the following criteria,
1/0 = Specific Gravity of Fluid [-] If Re <= 2100 - Laminar Flow Equation
Psizing = pressure drop [psig] If Re <= 4000 - Churchill Equation
Upon calculating the required Cv, it is required to If Re > 4000 - Colebrook White Equation
check if the calculated Cv is within the Cv limit of The Laminar Flow equation also referred to as the
the selected control valve. If not, the next size of Hagen Poiseuille’s equation is,
control valve is chosen and the calculations are
64
repeated. To arrive at accurate predictions for Cv 𝑓= (51)
Re
of the selected size, the calculations are repeated
The Churchill equation combines both the
by re-inserting the calculated Cv & control valve
expressions for friction factor in both laminar and
size (d) value into the Fp equation, i.e., Eq. 1 to
turbulent flow regimes. It is accurate to within
calculate the new value of Fp & further continued
the error of the data used to construct the
Page 6 of 9
Moody diagram. This model also provides an with the least amount of vibration and radial
estimate for the intermediate (transition) region; forces acting on the shaft. Pump system
however this should be used with caution. resistance curve should be calculated
accurately because the pump operates where
The Churchill equation shows very good
the performance curve intersects the system
agreement with the Darcy equation for laminar
curve.
flow, accuracy through the transitional flow
3. It is not always possible to operate the pump
regime is unknown, in the turbulent regime a
at BEP for the conditions required and hence a
difference of around 0.5-2% is observed between
flow variation of ±10% of BEP is allowed.
the Churchill equation and the Colebrook
4. Minimum stable continuous flow (MSCF) is the
equation. For Reynolds number up to ~4000,
minimum flow below which the pump is not
1⁄
8 12 1 12 allowed to operate. Although API 610
𝑓 = 8 [( ) + (𝐴+𝐵)1.5] (52) recommends that the rated region is located
𝑅𝑒

16 between 80% to 110% of BEP, the preferred

1 region of flow is between 70% to 120% of BEP.
𝐴 = [2.457𝑙𝑛 ( 7 0.9 𝜀
)] (53)
( ) +0.27
𝑅𝑒 𝐷 5. Clause 6.1.12 of API 610 11th edition states
37,530 16 “Setting limits for preferred operating region
𝐵 = [( )] (54) and the location of rated flow is not intended
𝑅𝑒
to lead to the development of additional sizes
The Colebrook equation was developed taking
of small pumps or preclude the use of high-
into account experimental results for the flow
specific-speed pumps. Small pumps that are
through both smooth and rough pipe. It is valid
known to operate satisfactorily at flows
only in the turbulent regime for fluid filled pipes.
outside of the specified limits and high specific
Due to the implicit nature of this equation it must
speed pumps that may have a narrower
be solved iteratively. A result of suitable accuracy
preferred operating region than specified
for almost all industrial applications will be should be offered…” Therefore the Allowable
achieved in less than 10 iterations. For Reynolds Operating Region is set by the manufacturer as
number up to ~4000, the allowable region to operate with stability
1 𝜀 ⁄ 𝐷𝐻 2.51 whilst conforming to predefined API 610
= −2 log10 [ + ] (55)
vibration limits.
√𝑓 3.7 𝑅𝑒√𝑓

Owing to the implicit nature of the Colebrook 6. Pumps that are expected to operate less
equation which requires iterations, an alternate frequently can be chosen such that they
correlation may be used to estimate the Darcy operate at lower speeds at the cost of
friction factor, i.e., Swamee-Jain Correlation efficiency. Since the pump is selected to
operate intermittently, a slightly lower
which is calculated for Re > ~4000 as,
efficiency pump is acceptable compared to a
0.25
𝑓= ε⁄D 5.74 (56) higher speed pump. This will also ensure a
log10 [ + ] longer operating life cycle.
3.7 Re0.9

Notes on Centrifugal Pump Operating Curves References

1. For a given set of hydraulic conditions, a pump 1. Control Valve Handbook, 5th Edition, Emerson
is designed to operate for one set of flow and 2. https://www.tapflopumps.co.uk/understandin
head. Deviation from this operating point is g-centrifugal-pumps
allowed only to some degree.
3. https://neutrium.net/fluid_flow/pressure-loss-
2. Pump selection closer to the Best Efficiency in-pipe/
Point (BEP) will yield a more efficient pump
Page 7 of 9
Appendix C – Pump Performance Curves Estimates

Page 8 of 9
Appendix D – Liquid Control Valve Sizing

Page 9 of 9