ENGINEERING PRACTICE
Optimal system
design requires
the right vapor
pressure. Here’s
how to calculate it
the pressure of the liquid at the inet
of centrifugel pump must be high
enough to prevent vaporization
within the pump, beeause this va-
porization (called cavitation) hinders the
‘pumping and can damage the impellers
‘This pressure requirement mustbe taken
{nto aocount when deciding how high to
place the pump feod vessel relative to the
Iheightf the pump itself
‘Basically, the pump suction pressure
rust be greater than the fluids vapor
pressure at the pumping temperature.
‘The difference between pump suetion
pressure and vapor pressure is the net
positive suction head (NSH). For eases in
which the liquid contains no dissolved
gases, the vapor-pressure determination
fs straightforward, With dissolved gases,
‘the situation is more complicated, because
‘vapor-pressure data for such systems are
usually notat hand,
‘Adding to the complication isthe fact
that centrifugal pumps generally can, ag
it happens, tolerate a small amount of
‘vapor about 2 to 8% by volume) atthe im-
pelle eve. Ifthe solubility ofthe dissolved
‘gas is low and the tomperature is far
below the boiling point of liquid, the
amount of vapor released in a pump is
{ikely not to exceed the tolerable value un-
Jess the pressure reduction is substantial
‘Accordingly ifthe engineer plays it
safe and assumes (for lack of data) that
the process pressure ofthe liquid-gas so-
Iution in the feed drum is its vapor pres-
sure, and if he or she consequently de-
mands enough feed-drum elevation to
fully compensate for friction losses be-
€.C.Chen
Litwin Engineers and Constructors, Inc.
twoen the drum and the pump, this ele-
vation may well be greater than neves-
sary. Conversely, ifthe engineer ignores
‘the presence ofthe gas and bases the el
vation caleulation onthe vapor pressure
ofthe pure liquid, the drum may not be
high enough.
For an economical pumping-system
design, an effective vapor pressure that
lies between the process pressure and
‘the liquid vapor pressure should be used
-ReBRCCAPERRY
in calculating the available NPSH, or
NPSHA. Presented here is a relatively
simple method to determine that effec-
tive vapor preseure.*
Setting the stage
‘The method builds upon two earlier ap-
proaches tothe dissolved gas problem pre-
sented in CB. First, Penney [1] developed
‘or infomstinn co dealing wth eis that carey
ssid aad pu CE bebe,
me
COPE WITH
DISSOLVED GASES IN
105 CHEMICAL ENGINEEAING/OCTOBER 1983basic equations to express the volume
fraction of flashed gas asa function of sol-
ubility, vapor and liquid densities liquid
vapor pressure and total pressure. He
used these equations to analyze some
pump-performance problems ereated by
dissolved gas, However, he did not provide
an explicit method to caleulate the effec-
tive NPSHA of a pump.
Four years later, Tsai (2] introduced
‘an “artificial,” or effective, vapor pres-
sure of a gas-saturated liquid for ealex-
lating the effective NPSHA. He defined
this as the pump-eye pressure at which
the volume of flashed gas is 2.5% of the
total volume. Then he determined this
effective vapor pressure by applying
‘trial and error to Penney's equations.
‘The present article offers an analytic
way to caleulate Tsai's effective vapor
pressure without trial and error, Fur-
thermore, it presents some eurves gener-
ated from the analytic equations, to fur-
ther simplify the determination when
the allowable gas volume at the pump
eve isin fact 2 to 3%,
Deriving the equations
When the pressure of a saturated solu
tion is reduced from the feed drum pres-
sure P, to the pump eye pressure P, the
volume fraction of flashed gas (including
liquid vapor) ean be written as:
Figures 1 (left) and 2 (right). These graphs can simply the calculation of effective va
Volume fraction of lashed gas 0,020
(Liquid Vey(Operating pressure) =
Set S205 — Set
S001 820.01 — $20.05
7 1
“PIR -RIRY O-B IR),
‘SPIRIG-PIP,)
where S = W,py/og, @
Equation (1), the basis for the analyti-
cal equations derived in the present arti-
cle, was developed by Penney. He as-
sumed that the gas obeys the ideal gas
Jaw, Dalton's law and Henry's law. W, is
the weight fraction of dissolved gas inthe
liquid in the pump feed drum. Ifsolubil-
ity data or a Henry's law constant is ap-
plied, partial pressure (not total pres-
sure) ofthe gas is employed to calculate
W,, They isthe liquid density, assumed
to'be constant (because only a small
amount of gasis flashing). The pq isthe
ans density atthe tomperatue ard pres-
‘sure (total pressure, not partial pressure)
in the pump feed drum,
For conservative calculation, the liq-
uid is normally assumed to be saturated
with the gas, as noted earlier. However,
in reali process operation the gas-liq.
uid contact time is frequently not long
‘enough to reach equilibrium. Ifthe per-
centage approach to equilibrium ean be
catimated by test data or operational ex
perience, the amount of dissolved gas
can be reduced by a saturation factor a
(Oto 1) and expressed as:
f a
W,=