Page 231 - APPLIED PROCESS DESIGN FOR CHEMICAL AND PETROCHEMICAL PLANTS, Volume 1, 3rd Edition

Pumping of Liquids 203

Most standard pump curves illustrate the effect of When the performance of a pump handling water is
changing impeiler diameters on characteristic perfor- known, the following relations are used to determine the
mance (Figure 3-36A). Note change as reflected in the performance with viscous liquids [ 171 :
different impeller diameters. However, the slight change
in efficiency is not irecorded over the allowable range of
impeller change.
Recognizing the flexibility of the affinity laws, it is bet-
ter to select an original pump impeller diameter that is
somewhat larger than required for the range of anticipat- = CE(EW) (3-34)
ed performance, and then cut this diameter down after
in-service tests to a slightly smaller diameter. This new per-
formance can be predicted in advance. Once the impeller
diameter is too small, it cannot be enlarged. The only Determine the correction factors from Figure 3-56 and
solution is to order the required large impeller from the Figure 3-57, which are based on water performance
manufacturer. because this is the basis of most manufacturer’s perfor-
mance curves (except, note that the “standard” manufac-
Example 3-15: Reduicing Impeller Diameter at Fixed turer’s performance curves of head us GPM reflect the
M head of any fluid, water, or other non-viscous). Do not
extrapolate these curves!
If you have a non-cavitating (sufficient NPSH) operat- Referring to Figure 3-56 [ 171 :
ing %inch impeller producing 125 GPM at 85 feet total
head pumping kerosene of SpGr = 0.8 at 1750 rpm using 1. The values are averaged from tests of conventional
6.2 BHP (not motor nameplate), what diameter impeller single-stage pumps, 2-inch to %inch, with capacity at
should be used to make a permanent change to 85 GPM best efficiency point of less than 100 GPM on water
at 68 feet head, at the same speed? performance.
2. Tests use petroleum oils.
Qz = Qi (dz/di) (3-23) 3. The values are not exact for any specific pump.
85 125(dz/9)
dz = 6.1 in. diameter (new) Referring to Figure 3-57 [ 171 :

The expected he:ad would be 1. Tests were on smaller pumps, 1-inch and below.
2. The values are not exact for any specific pump.
H2 = HI (dz/di) (3-24)
Hz S5(6.1/9)n The charts are to be used on Newtonian liquids, but
=
= 39.0 ft (must check system new total head to deter- not for gels, slurries, paperstock, or any other non-uni-
mine if it will satisfy this condition.) form liquids [ 171.
Figure 3-56 and 3-57 are used to correct the perfor-
The expected brake horsepower would be mance to a basis consistent with thie conditions of the
usual pump curves. In order to use the curves, the fol-
BHPB = BHPl(d~/d1)~ (3-25) lowing conversions are handy:
BHPz = 6.2(6.1/9)3
= 1,93 BHP (use a 2- or 3-hp motor) Gentistokes = centipoise/SpGr

Eflects of Viscosity SSU = Saybolt Seconds Universal
= (Centistokes) (4.620) at 100°F
en viscous liquids are handled in centrifugal pumps, = (Centistokes) (4.629) at 130°F
the brake horsepower is increased, the head is reduced, = (Centistokes) (4.652) at 210°F
and the capacity is reduced as compared to the perfor-
mance with water. The corrections may be negligible for Example 3-16: Pump Performance Correction For
viscosities in the same order of magnitude as water, but Viscous Liquid
become significant above 10 centistokes (10 centipoise for
SpGi- = 1.0) for heavy materials. While the calculation When the required capacity and head are specified for
methods are acceptably good, for exact performance a ~~SCOUS liquid, the equivalent capacity when pumping
charts test must be run using the pump in the service. (text continued on page 206)