You are on page 1of 15

MECH 4305/5401 Fluid Machinery/Turbomachinery

Problem Set #1
2. Centrifugal pump measurements
(a) Non-dimensional characteristics
(i) Head coefficient
CENTRIFUGAL PUMP EXPERIMENTAL RESULTS
Head Coefficient versus Flow Coefficient

0.2

Head Coefficient, g/N D

2000 RPM
2600 RPM
3000 RPM

0.15

0.1

0.05

0.0005

0.001

0.0015

Flow Coefficient, Q/ND

0.002

(ii) Power Coefficient

CENTRIFUGAL PUMP EXPERIMENTAL RESULTS


Power Coefficient versus Flow Coefficient

0.0004

Power Coefficient, Pshaft/N D

0.0005

0.0003

0.0002

2000 RPM
2600 RPM
3000 RPM

0.0001

0.0005

0.001

Flow Coefficient, Q/ND

0.0015

0.002

4305/5401 PS#1/2

(iii) Efficiency
CENTRIFUGAL PUMP EXPERIMENTAL RESULTS
Efficiency versus Flow Coefficient

0.5
2000 RPM
2600 RPM
3000 RPM

Efficiency

0.4

0.3

0.2

0.1

0.0005

0.001

0.0015

Flow Coefficient, Q/ND

0.002

(b) Influence of Reynolds number


Reynolds number affects mainly efficiency. The efficiency plots in part (a) for the three different
values of rotational speed show no systematic trend with speed that could be attributed to Re
differences (one would expect highest efficiency for the highest Re). The differences in the
curves appear to be mainly due to experimental scatter.
Projected variation in based on the ASME PTC correction formula:
1 1

Re2
=

1 2 Re1

n = 0.1 recommended for centrifugal compressors

Re =

water := 1000 kg/m3

D := 0.2286

water := 1.1 10

kg/m s

Let condition 1 correspond to N = 2000 RPM, condition 2 N = 3000 RPM


N1 := 2000

RPM

2
1 := N1
60

1 = 209.44

rads/s

N2 := 3000

RPM

2
2 := N2
60

2 = 314.159

rads/s

4305/5401 PS#1/2

Re1 :=

water 1 D

Re1 = 9.95 10

water
2

Re2 :=

water 2 D

Re2 = 1.492 10

water

2 := 0.37

Taking peak efficiency

at 3000 RPM

Then projected efficiency at 2000 RPM, using n = 0.1 as recommended for centrifugal
compressors:
With

n := 0.1
n

Re2 Re2
1 := 1
+ Re 2
Re1

1 = 0.344

Thus the correction formula predicts that the efficiency should have been 34.4% at 2000 RPM
compared with 37% at 3000 RPM, which does not seem to be borne out by the experiments.
The change in efficiency also seems a little large for only a 50% change in Re. What is the
explanation for this? The formula corrects the efficiency on the basis that all physical effects
that contribute to are sensitive to Re: that is, are due to viscous effects. For the present
pump, the poor peak efficiency implies that the losses are very high (more that 60% of the input
power is lost in one way or another). This makes it sensitive to any changes in the losses: for
example, a 1 % increase in losses results in about a 1*(60/40) = 1.5% reduction in the
observed efficiency. This effect can be seen by applying the correction formula with a much
higher initial :
With

n := 0.1

but starting from


n

2 := 0.8

Re2 Re2
1 := 1
+
2
Re1 Re1

1 = 0.792

Now the predicted change in observed efficiency is less than 1%.


The question then arises: is the low efficiency of the present pump entirely due to viscous
effects? As mentioned in class, this pump is believed to have a high level of leakage from the
outlet of the pump back to inlet, through the clearance between the pump shroud and the
casing. This recirculation means that the mass flow rate through the rotor is a lot higher than
the flow rate actually delivered by the pump. The extra work required to compress this flow
shows up as a reduction in efficiency, but it is probably not an effect that is particularly
sensitive to Reynolds number.
Conclusion: The ASME PTC correction formula may give misleadingly strong changes in
due to Re changes for pumps such as the present one for which non-viscous effects play an
important part in determining the efficiency.

4305/5401 PS#1/2

An alternative form of correction formula has sometimes been used in the past (see Shepherd,
1956), but not seem to be discussed in recent text books or papers. This form is written as
follows:

1 1

Re2
= A + B

1 2
Re1

where

A + B = 1.0

For example, if only half the losses were thought to be sensitive to Re, the formula would take
the following form

1 1

Re2
= 0.5 + 0.5

1 2
Re1

Applying this to the centrifugal pump with

n := 0.1

2 := 0.37

Re2
Re2
1 := .5 + .5 2 .5
+
.5

Re 2
Re1

1
1 = 0.357
Now the predicted drop in efficiency is 1.3%, which seems more plausible.

(c) Performance of geometrically-similar pump


Working fluid: Motor oil

SG := 0.89

oil := SG water

oil = 890

kg/m3

oil := 0.06 N/s m2


Geometry

Doil := 0.05

Speed

Noil := 5000

RPM

2
oil := Noil
60

oil = 523.599 rads/s

The pump is assumed to be operating at its design point: its point of maximum . From the
experimental measurements, at best
CQ := 0.0013

CQ =

CH := 0.12

CH =

Q
3

g H
2

g := 9.81

m/s2

4305/5401 PS#1/2

Negelcting Re effects on C Q and CH, these values this value also applies to the oil pump.
(i) Flow rate delivered by oil pump
Qoil := CQ oil Doil

Qoil = 8.508 10

m3/s

Corresponding mass flow rate, in kg/hr


moil := oil Qoil 3600

moil = 272.612 kg/hr

(ii) Pressure rise delivered by oil pump

H :=

CH
g

oil Doil

H = 8.384

P0 := oil g H

m of oil

P0 = 7.32 10

Pa

(iii) We have already established that the ASME PTC may not provide reliable estimates of efficiency
changes due to Re differences for this family of pumps. Examine the relative Reynolds numbers:
7

Re2 = 1.492 10

Water pump (at 3000 RPM say)


Oil pump

Reoil :=

oil oil Doil

oil

Reoil = 1.942 10

Thus, the Reynolds number is much lower for the oil pump. We would therefore expect it to have a
noticeably lower (by 5%?, 10%?) than the water pump, although we don't have a satisfactory basis
for estimating the change at this point.

4305/5401 PS#1/2

MECH 4305/5401 - Fluid Machinery/ Turbomachinery


Problem Set #1

3. Scaling up an axial-flow fan.


Data for small fan:

N1 := 1750

RPM

Q1 := 4.25 m3/sec

H1 := 153

mm water

The scaled-up, geometrically-similar fan is to deliver the same head at the same efficiency but
with a speed of 1440 RPM. We want to determine the flow rate.
Data for the large fan:

N2 := 1440

RPM

H2 := 153

mm water

Since the two fans are geometrically similar they will have identical non-dimensional performance
curves when plotted against the criterion of similarity, Q/ND3 (neglecting Reynolds number effects).
Furthermore, since the two fans have the same efficiency, they must be operating at the same
non-dimensional operating point.
Q1

Thus, from the flow coefficients:

Q2

=
3

N1 D1
Solving for Q2:

Q2 =

N2 D2

Q1

N2 D2

(1)

N1 D1

To solve for Q2 we therefore need the scale ratio D2 /D1


Since all non-dimensional parameters (both dependent and independent) are matched for the two
machines at the operating points of interest, we can use this fact to obtain the scale ratio. From the
head coefficients:
g H1
2

N1 D1

g H2

(2)

N2 D2

Letting DR = D2/D1, the scale ratio, solve (2) for DR:


DR :=

Then from (1)

Q2 :=

1
H1 N2
Q1
N1

H2 N1

DR = 1.215 (scale ratio)

N2 DR

Q2 = 6.277

m3/sec

This is the volume flow rate which will be obtained from scaled-up fan.

4305/5401 PS#1/3

MECH 4305/5401 - Fluid Machinery/Turbomachinery


Problem Set #1
4. Model-scale testing of a centrifugal pump.
Data for prototype (full-scale machine):
3

QP := 7

HP := 130 m

sec

also

NP := 350

M := 0.7

P := 0.7
g := 9.81

(assumed)

m
sec

:= 1000

kg
3

Constraints on the model:


3

QM := 0.15

sec

WM := 220 kW

(maximum available values)

(a) Speed of the model and the scale ratio


The problem states that the model and prototype will have an efficiency of 70%. There are two
assumptions implicit in this information. Firstly, that differences in Reynolds number between the model
and the prototype do not affect performance, since efficiency is in general a function of Reynolds number.
Thus, it is implied that the Reynolds number can be neglected as a similarity criterion in the present
analysis. Secondly, engineering judgment and experience has been applied to arrive at 70% as a
reasonable value for the efficiency. Since the machine has not yet been tested, the efficiency would not
be known and the quoted value must be an assumption.
Since Reynolds number effects can neglected, there is only one independent similarity parameter. For a
hydraulic machine this would normally be the flow coefficient:
Q
3

N D

Since it is also stated that the efficiencies are the same for the operating points of interest, this implies
that the two machines are operating at the same value of Q/ND3 . If the flow coefficients are matched
between the model and the prototype, all other similarity parameters must also be matched. From the
flow coefficients;
QM
NM DM

Rearranging

Thus

=
3

QP
NP DP

DP3
QM
NM =
NP 3
Q
P
DM
DP3

NM = 7.5
D 3
M

QM

and

QP

NP = 7.5

(1)

4305/5401 PS#1/4

Clearly any diameter ratio (scale ratio) will satisfy the similarity requirement, provided the model speed is
then set according to (1). However, many combinations of scale ratio and speed will result in models
which require more power to drive than the maximum power available. Therefore, we use the fact that al
dependent non-dimensional parameters will also have the same values at the equivalent operating points
This will allow us to determine suitable values of the speed and scale ratio.
Any dependent non-dimensional parameter (head coefficient, power coefficient etc.) will do. Therefore,
there are several options. Since the model power has been specified, the most obvious approach is to
use the power coefficient.

WM

WP

NM DM

NP DP
1

WM
NM =

WP

5
3

DP

NP
DM

(A)

Power to drive prototype:

WP :=

g QP HP

WP = 12753 kW

( = 17,100 HP - it is obvious why the


design needs to be tested at model
scale)

1
3

WM
W NP = 90.435
P

Then

and substituting into (A)

DP
NM = 90.435

DM

(2)

Letting R = DP/DM, and equating (1) and (2)


5

90.435 R
3

= 7.5 R

5
3

90.435
7.5
3

Then

90.435
7.5

R :=

90.435

7.5

R = 6.471

(required scale ratio)

4305/5401 PS#1/4

From (1), the corresponding model speed is


5

NM := 90.435 R

NM = 2032

RPM

The value of R = DP/DM obtained defines the largest model which is possible. Smaller models
could be used. The model speed would then need to be adjusted according to either (1) or (2) to
maintain similarity.
An alternative approach would be to use the head coefficient:
g H
2

N D

The head rise across the model pump, for the maximum available power and flow rate, is

HM :=

M WM

HM = 104.66 m

g QM

Then equating the head coefficients for the model and the prototype, it is easily shown that
the same scale factor and model speed are obtained.

(b) Considering possible effects of Reynolds number, using:


1 P
1 M

ReM
=

ReP

1 + P +

Solving for

M =

ReM
Re
P

ReM
Re
P

From the definition of the Reynolds number


2

Re =

N D

ReM

then

ReP

NM DM
NP DP

Letting RRe = ReM/ReP, then

RRe :=

NM 1

NP 2
R

RRe = 0.13866

4305/5401 PS#1/4

If

n := 0.1

with

P = 0.7
n

M :=

1 + P + RRe

M = 0.63447

RRe

If

n := 0.25

with

P = 0.7
n

M :=

1 + P + RRe

M = 0.50838

RRe

Thus, the Reynolds numbers are sufficiently different that a noticeable difference in efficiency can be
expected between the model and the prototype, with the prototype having the higher efficiency. As a
result, the model pump will draw more power than was assumed in (a) and a larger scale ratio than
6.47 should be used.

4305/5401 PS#1/4

MECH 4305/5401 - Fluid Machinery/Turbomachinery


Problem Set #1
5.

Sizing centrifugal pump:


Q := 200 0.002676

Required operating point:

Q = 0.535

ft3 /sec

(1 Imp. gal = 0.16054 ft3, 1 Imp. gal/min = 0.16054/60 = 0.002676 cfs)


P0 := 100 144

P0 = 14400

lb/ft2

:= 1.94 (slug/ft3)

For water:
(a) "Ideal" pump

The "ideal" pump from the available family is the one that operates at its design point (that is,
its point of best efficiency) at the desired operating point.
The pressure rise characteristic is given in terms of pressure coefficient vs flow coefficient:
CP =

P0
2

(1)

CQ =

N D

(2)
3

N D

Note that in this form, the parameters are truly non-dimensional if the variables are expressed
in compatible units. If you are given no other information, you should assume that compatible
units have been used. In this case, you are told that N is in revs/sec. In the absence of other
information, you would then assume that Q should be expressed in ft3 /sec, etc.
:= 0.795

From the figures, at the design point

CQ := 0.052

CP := 19

Then finding the size and speed of the pump that will operate at the design point while giving the
required performance:

From (2):

N=

(3)
3

CQ D
Substituting into (1):

CP =

P0
2

CQ D

Q
Solving for D:
1

D := CP

D = 0.7216

1
4

P0 CQ

or

ft

D 12 = 8.66

in

4305/5401 PS#1/5

Then from (3):

N :=

N = 27.4

rps

CQ D

or

Finally, required power:

W=

Q g H

W = 9694.2

thus

or

N 60 = 1643.4

550

= 17.6

or

RPM

W :=

P0

ft-lbf/sec
HP

Since present pump is not available with 8.66 in. diameter, we will have to choose the
best size from those available.
(b) Choosing best available pump:
Since the diameter is now constrained, the required pressure rise (with specified Q), will be
obtained at some RPM which will not correspond to the point of peak efficiency.
The most obvious choice for size is the 8 in diameter model. However, we can not automatically
rule out that possibility that the 10 in. diameter model may give a higher efficiency at the desired
operating point and we should check it.
(i) Examining the 8 in. pump:
Q := 0.535 ft3 /sec

D :=

8
12

D = 0.6667 ft

When we want dimensional performance curves, we usually obtain them from the
non-dimensional characteristics by plotting, say, P0 versus Q for various constant values
of N. We could do that here and then estimate the speed at which we get the desired 100
psi pressure rise with Q = 0.535 ft3 /sec. We could then find the corresponding value of
Q/ND3 in order to read the efficiency from the second performance diagram. However,
there is a neater way to approach this. From
CQ =

Q
3

N D

holding Q constant (and knowing D), we can calculate the values of N


corresponding to various values of CQ on the performance diagram
N=

(1)
3

CQ D

4305/5401 PS#1/5

and each value of CQ has a specific corresponding value of


P0

CP =

N D
from which

P0 = CP N D

(2)

Thus, from (1) and (2), we could replot our non-dimensional CP versus CQ characteristic
as a plot of P0 versus N, for fixed Q and D.
From the tabulated performance:
CQ :=

i := 1 .. 8

0.035
0.040
0.045
0.050
0.052
0.055
0.060
0.065

CP :=

i :=

22.3
21.8
20.8
19.5
19.0
17.4
14.8
11.0

0.57
0.673
0.749
0.792
0.795
0.775
0.695
0.56

Then calculating corresponding values of


2

N :=
i

60 Q

Ni 2
CP D
i
60
Pi :=

(RPM)

(psi)

144

CQ D
i

and we might as well calculate the resulting values of the power required
W :=
i

Q Pi 144
i 550

(HP)

N =

Pi =

3095.4
2708.4
2407.5
2166.8
2083.4
1969.8
1805.6
1666.7

355.4
266
200.5
152.3
137.2
112.3
80.3
50.8

W =
i

87.3
55.4
37.5
26.9
24.2
20.3
16.2
12.7

4305/5401 PS#1/5

Plotting:
120

Thus for the 8 in. pump, 100 psi


is obtained at a rotational speed
of about 1910 RPM.

Pi 100

80
1800

1900

2000

Ni

19

The corresponding power required is


seen to be about 18.6 HP

Wi 18

17
1800

1850

1900

1950

2000

Ni

(ii) A 10 in. pump may also be able to produce the required combination of flow rate and
pressure rise. Check this case to see if it results in better efficiency (or alternatively a lower
drive power).
Proceeding as in part (i) with
Q := 0.535 ft3 /sec

D :=

10
12

D = 0.8333 ft
2

N :=
i

60 Q

(RPM)

Ni 2
CP D
i
60
(psi)
Pi :=
144

CQ D
i

W :=
i

Q Pi 144
i 550

(HP)

4305/5401 PS#1/5

N =

Then

Pi =

W =
i

145.6

1585

35.8

108.9

1387

22.7

82.1

1233

15.4

62.4

1109

11

56.2

1067

9.9

46

1009

8.3

32.9

924

6.6

20.8

853

5.2

120

Pi 100

80
1300

Thus for the 10 in. pump, 100 psi is


obtained at a rotational speed of about
1335 RPM

1350

1400

Ni
21

The corresponding power required is


about 20.3 hp
Wi 20

19
1300

1350

1400

Ni

Conclusion: 8 in. pump is preferred since it has lower power required (i.e. better efficiency)
at the desired operating point. The 8 in. will of course also be cheaper than the 10 in.
model.

4305/5401 PS#1/5

You might also like