
2.372 CHAPTER TWO
be as low as 50% of the suction recirculation values shown for continuous operation and
as low as 25% for intermittent operation. For hydrocarbons, the minimum operating flows
can be as low as 60% of the suction recirculation values shown for continuous operation
and as low as 25% for intermittent operation.
51, 52, 53
Temperature Rise Under steady-state conditions, friction and the work of compression
increase the temperature of the liquid as it flows from suction to discharge.A further tem-
perature increase may arise from liquid returned to the pump suction through wearing
rings, a balancing device, or a minimum-flow bypass line that protects the pump when
operating at or near shutoff.
Assuming that all heat generated remains in the liquid, the temperature rise is
(33)
where g/g
o
1 lbf/lbm; but when using SI units, g/g
o
is replaced by 9.80665 m/s
2
( g in the
SI system). T
c
is due to the compression of the liquid and is not a consequence of loss or
dissipation as is the term involving the pump efficiency h (see Section 2.1). As shown in
Reference 1 of Section 2.1, T
c
is 3°F per 1000 psi (0.24°C per MPa) of pump pressure rise
for hydrocarbon fuels. For boiler feedwater at 350°F (177°C), T
c
1.6°F per 1000 psi
(0.129°C per MPa), but it is much smaller for cold water. By consulting tables of properties
for the liquid phase of the fluid being pumped and assuming the compression process
between the actual inlet and discharge pressures to be isentropic, T
c
can be determined.
This is important if Eq. 33 is used to evaluate overall pump efficiency from temperature
rise measurements. T and T
c
are often of the same order of magnitude at BEP, and seri-
ous errors have been made by excluding T
c
from the efficiency computation. At very low,
off-BEP flow rates, T will be high in comparison to T
c
; so, the latter can be safely ignored
in temperature rise calculations at such low-efficiency conditions.
In practice, determination of efficiency from T-measurements is accomplished by the
direct thermodynamic method
54
, rather than by the T
c
-method. Both approaches are
based on the definition of pump efficiency as the ratio of an isentropic rise of total enthalpy
( gH) to the actual rise of total enthalpy (Eq. 1 of Section 2.1), allowances being made
for the usually small external power losses that do not appear in the pumped fluid (such
as bearing drag) and the similarly small effects of heat transferred between pump and
surroundings. In the direct thermodynamic method, the enthalpy rise h is found from the
chain rule,
the coefficients a and C
p
J being average values of the two partial derivatives as found from
tables of thermodynamic properties of the fluid. Values of these partial derivatives are con-
veniently tabulated for water in Reference 54.
General service pumps handling cold liquids may be able to stand a temperature rise as
great as 100°F (56°C). Most modern boiler-feed pumps may safely withstand a temperature
rise of 50°F (28°C). The NPSH required to avoid cavitation or to prevent flashing of hot liq-
uid returned to the pump suction may be the controlling factor. Minimum flow may be dic-
tated by other factors, such as recirculation and unbalanced radial and axial forces on the
impeller. Axial forces can be the controlling factor with single-stage double-suction pumps.
It is especially important to protect even small pumps handling hot liquids from oper-
ation at shutoff. This is usually done by providing a bypass line fitted with a metering ori-
fice to maintain the minimum required flow through the pump. In the case of boiler-feed
pumps, the bypass flow usually is returned to one of the feed-water the water heaters.
Unless especially designed for cold starting, pumps handling hot liquids should be warmed
up gradually before being put into operation.
Radial Thrust Ideally, the circumferential pressure distribution at the impeller exit is
uniform at the design condition (as explained in Section 2.1); however, it becomes non-
¢h
dh
310h>0p2
T
dp 10h>0T2
p
dT4 a¢p C
p
J¢T
¢T
gH 11 h2
g
o
C
p
hJ
¢T
c