i Ea
| |
~ Assessing Centrifugal
! Apump must suit the
ump VS ems system, not the other
way around
‘Scott Mansfield ee
Consultant Mavinur sachet =i ssoetiTe HL
: Operating level. = Building |
1 important job we are offen sum
! tasked with is assessing the op- oy Al
eration ofa centrifugal pump in eat
fa given situation. Oa some oo ay
t sions we will be asked to size a new [ie eet
pump for a given application. Other FIGURE 4. Ono of the
times we might be asked to find out ‘most common pumping
why a pump system is not behaving as
hhad been hoped. Finally, we might be
asked to change something in a pip-
ing system, perhaps a control valve or
a spray nozzle, and we want to make
isa) configurations Is forthe
(Wetl=20) transfer of liquid from a low
ration. The system
1 discussed in the text
CASE
sure the pump continues to perform | [Design flow rae | 280 gpm. contolled (average flowrate on flowshoets)
ae oe these tasks weal ree | [Mexia aw rate | 10-9pm, conionea
take a look at the basie How require. | [ Discharge tne 4904. dnchudes equlvalont lengths for fitings)
ments of the most. common pumping | [Suction ine ‘éhx in indhides equivalent length or fitings).
Frmggeration namely, the transier & | [Supply evel high | 16 #above suction
tlevation. Thon we ll have a cose | [SpYVIeveLiow <1 | SWaboveweton TT
look at this same system for a better | [Pump elevation | 1,000, above mean sea level
understanding of what is going on.
Next we will have a brief view of one | sider the proposed system shown in | exit are taken as one velocity head. Be-
other pumping system to appreciate | Figure 1. A manufacturing operation | cause length of pipe already includes
how a similar pump will respond to a | on a hill requires water from a source | allowances for fittings, fittings are not
different application. Finally, we will | down the hill. The supply water origi- | listed separately.
quickly cover some aspects of pump | natesin a treatment plant and is tobe | We are now ready for the first esti-
protection, accumulated in a storage tank at the | mate of control valve friction.
bottom of the hill. From there it will | Control valve: Because the head tank
Case 1: the basics bbe pumped up the slope (70 ft vertical | is operatod at a constant level a level
‘The starting point in pump analy- | rise) to the base of the structure, and | control valve is employed, as shown in
sis should be to establish the system | then vertically upward tothe discharge | Figure 1. We note in Table 1 that both
characteristic, or system curve. The | point in the side of a small constant- | the average and the maximum flow-
system curve is the curve that would | head tank. The design parameters are | rates must be lower than the flowrate
result from plotting pressure drop ver- | given in Table 1, Control-valve param- | corresponding to the wide-open con-
sus flow with the pump absent. The | eters are not included; they are listed | trol valvo. This is because a wide-open.
‘main reason for starting here is that | separately below. Note that the level | control valve is not controlling, yet the
‘the curve defines what is required of | of water in the supply tank varies. | specification in Table 1 is for the maxi-
‘the pump. The pump must suit the | Our calculations begin with an es- | mum flowrate to be controlled.
system; not the other way around. | timate of friction losses in the piping | To get started, we will assume a con-
Jt doesn’t make sense to examine a | for the maximum controlled flow (810 | trol-valve size and valve position (per-
pump catalog or to contact a pump | gpm). A breakdown of the losses is | contage open) for the 310 gpm case;
1 vendor until this easy task has been | given in Table 2. To save time, we use | then we will do some caleulations and
carried out. the tables in Cameron Hydraulic Data | review the results, Looking in a ran-
System curve: As an example, con- ' (I). Individual losses for entrance and | dom (or old) control-valve catalog, we
34 CHEMICAL ENGINEERING WHWWICHE.COM AUGUST 2005"aT FRITON LSSES aT sv GEM CWTTHOUT WANES
Description Loss toss, [vttls
(velocity heads)
Tnirance fo suction tne, 190 038 a7
2 Suction line 023 497
‘S.Dischargeling To ese gat
“4. Discharge fo head fonk 1.00 0.95 781
Z Sum: 2a Tae
2. opm Ht 117
Tonle, a7
50 fossa [o2sa [tier
joo farsa [air [1s
so [6792 [2650 | 126.4
200 fie far [1a
260 [va02 [7362 [1423
300 [2553 [ 1040 | Ysa
0 [ara farsa [155
‘3607 Saar [rau [1687
oo [4423 [veea | 1800
FIGURE 2. The = ‘curve for Case 1
fg shown here for a maximum controlled
flowrate of 310 gpm (control valve 90%
hhome in a 4-in, equal-percentage valve
‘that, at 310 gpm, would operate about,
909% open. This is acceptable for the
moment, so we set about calculating a
system curve with the valve 90% open,
‘To obtain the pressure drop across
the control valve we use the equation
for the valve coefficient,
ey = Q(ep.grlAPyu a
where Qs flowrate gpm), sp.gr-is spe-
cific gravity of the fluid (water sp.gr. =
1D, and AP is pressure drop across the
valve (psi).
‘The c, given in the vendor catalogue
for the 90% open, d-in. valve is 140. We
rearrange Equation 1 and (keeping in
mind that in this case, sp.gr. = 1) cal-
culate AP from,
AP = (Qle,)2= (810/140)? = 4.90 psi
which is 11.82 feet of water. To obtain
the total system friction loss, Hy, we
add the values for piping plus valve:
Hyj= 2114+ 11.82 = 88.46 HO
‘To build the rest of the curve, data
for pipe friction for other flows were
taken from Cameron [i] and compiled
‘open, maximum static head)
in the same manner as with Table 2.
‘Valve friction for these flows was esti-
mated by AP = 11,2(@/310)2. Several
flowrates were selected for flows from
0 t0 400 gpm. For each flow a value of
He was calculated. Since the system
cufve includes static head as well as
friction head, we add 117 ft to each of
the friction values. Table 3 lists the
results, Total head versus flowrate is,
plotted in Figure 2. This is the system
curve for the ease where the control
valve is 90% open, required static
head is highest (low level in the sup-
ply tank), and flow is at the maximum
controlled rate.
Before looking at how the pump
fits into this analysis; we will exam-
ine the system curve some more. To
start with, imagine Figure 1 without
the pump (and with the control valve
90% open). Without a motive force, the
flow would be zero, and the level in the
discharge leg would be at the same
level as the liquid in the storage tank.
Imagine applying air pressure slowly
in the supply tank, The level in the
discharge leg would slowly rise until
reaching the control level of the head
tank. During this process, the flowrate
would be quite-small, and the friction
Joss would be negligible. This part of
the system curve, coincides with the
vertical axis (Q = 0) up to 117 fof
water
If the (imaginary) air pressure on
the storage tank is fixed at 117 fe (60.7
psi), flow wil stop, ie. Q= 0. Ifthe air
pressure is increased to 117 + 38.46 ft
= 155 ft of water, low will oecur at the
‘maximum design rate: Q = 310 gpm.
Obyiously, if the pressure is increased
even more, the flow will increase cor-
respondingly. Our curve goes out to
nearly 350 gpm, well beyond the range
of interest. This is our first analysis of
the system curve.
Pump curve: Now we go through a
similar exorcise (using another random
‘or old catalog) for determining a plau-
sible pump curve. We select = pump
that at 1,750 rpm will deliver about
450 gpm. Our system conditions fit
into the center of the range of service