Académique Documents
Professionnel Documents
Culture Documents
t) F Ltfu A ru
"I'A-f
l ru
{i
F ['i E i\i
ffi";,'.
'g'g5qli;Fr
,\;q
'r"i
,,,..,4,
ffifu
Elj
from its original design. However, obtaining accurare per[ormance data in the field can be very challenging.
This aticle wiil explain the relarive importance that different process variables have in performance calculations, as ,.yeil
as specify the necessary instrumentation to obtain process data
with an acceptable uncertainty. Since original design conditions almost neve match actual operaring conditions, how to
. , compare actual field data to design data usingn-ondimeqpional
head and efficiency will be demonstated. Likewise','the limits
on these comparisons wiil be tiilined fo users. Several example
fiel{ performance evaluations are discussed with some common
' ' pitfalls that can be avoided. Examples of the effects of inaccurate
process data are also included
in the
discussion.
Q=
e\
o
ation are:
. Gas composition
. Rotational speed
. Dliver load.
.'
are:
within
where
=e.tr"lr-w.
-\ P, T^ Z^ Mw^
: f (p,T,Mw)
forstd. vorume
(l)
flow
h, = ,t.trW" Y,
^ ! Po Tn Z^ Mut,
f,or nrass
o.=o[ut"z"**.
\ -\ P^To Zo Mtan
flow
12)
(3)
0ow
(4)
: ,
KJtr
. il,,i.,
tlq
| :s
i:i"{"I{il
Fii
lJr/ i{il'jr,'
Ftowtoohigh
tru
fi
"4
[J
ii F,]
i1,4
fui'l-
f,iTA-T
Liken,ise, in a wet ga.s application, an accurate gas compusition ir; also the most difficult piece of tesr data to obcain. r\t
least two gas samples should be taken during the performance
test. lfwo samples are required to vaiidate the gas analysis
(for comparison) as well as when one of the samples is iosr or
invalid (i,e.,. qirhout the gas composition, the other process
dara are wUrthlei$, fhe sample should be obtained using a
Actual curve
\'**o.y'
ffi
hydro1en.
As an example, a sample taken from the discharge of a cokcr
wet gas compressor was analyzed at the lab ambient ternperature
(approximately 7 5"F) and at 27 5"F (sample temperature, Table 1).
275'F
Molecular weight
Flow MMscfd
27.4
32, s
Shaft horsepower, hp
5,77s
7,036
oo
] nucusr
2007 HyDRocARBon
pRocESsrNG
"***;ffi
e/l
*f:ffi
sffi
Effy.
@a*?**s!
mf
1.
l-Lffii!:) Ffl"t'1ji/ ll
ftr3U",{,f II\f
Example-inclrrect gas cornposition. ]f the measured gas composition is lower than the actual gas composition in the compressor, it will have the most pronounced effect on borh the correcred
flowrate and calculated polytropic head. An incorrect low molecular weight will cause the corrected flowrate (if it is measured
in standard cubic feet) to be higher (Eq. 2). Likewise, the low
molecular weight will cause the calculared polytropic head to
be higher than it actually is for the given compression ratio. A
molecula weight that is coo low wll al^so.,pause the calculated
tl(4f
polytropic .ffi..,.y to be higher,
as pronounced.
"o, lower
Additionall the lower molecular weight will
rhe calculated mass flow, which will in tun lower
the calculated horsepower. Note, these are
the effects produced by an incorrect gas
composition being used as the input for
a field test. These are not the results of a
compressor that is operaring in a gas that
is actually lower in molecula weight than
its design. If the measured gas composition
is too high, the results are just the inverse
of the above (i.e., the flow, head and efficiency are all lower). A summary is shown
in Table 2.
{r
Hq}{.JIF}lVlE[ri"I
Applying the first law of thermodynamics to the control volume around the compressor is shown in Fig. 3.
rtiP
Shp
hp*,
2.545
(6)
uecu
(7)
bpuz
'.
Once
"
these parameters).
t/E
61
;SPECIALRE
*:
I_ffi
[}
[.] : LW /\N
iq "8 "\r
r,d
ffi
{- i
p \{
H [\{ T"
Wo seal losses
Wseal losses
----7
P2
Loss in
pressure due
to higher f7
lncrease in
volume flow
due to higher
11 and leakage
from:
the reservoir.
.
en lrn atri-f',rrc'
''1r
il:':
o':ilT
5 i
:'-!i''
the seai leakage on the inlet is normally less than 170, but it can
be easily calculated at the orifice in the ventrofF'the sealipoti.
Note, the internal seal losses (i.e., balance piston and impller
thc^LT.,
t. _o,t
labyrinth
_
L^
,'f
M:CH
p.rfor-".r..
data'lirmr-,J"t'Jst'ii'
,t.
(g)
33,000
seals.
"
YABL 4"
-3,000
differential.
Gearbox efficiencies
Mechanical losses,%
Gear type
Etficiency, %
3.0
Helical
97-99
- 6,000
2.5
Herringbone
6-99
6,000- 1 0,000
2.0
Straight bevel
95-98
I0,000 +
1.5
Spiral bevel
96-98
3,000
,r
lq
<-
'.
m P
T1,
hi+
= nl1 *
ric
= otH
lmpeller flow,
lfl6p Ptu Tt',
IIll 't
h1
never exactly match the original design, certain nondimensional parametes must be calculated so that the field performance can be compared to the OEM shop test or predicted
performance data.''7hile these nondimensional parameters
will enable "apple-to-apple" comparisons for different conditions, they have very real limitations based on the aerodynamic characteristics of the impellers. These nondimensional
parameters includei
Polyrropic head coefficient, p7
ht'
hoo
fltp
P.2.
,-l
P.2
,
(e)
NT
0.51trP*rH
,lznr
llp =
/t"\
(12)
lnDN)'
Poiytropic efficiepc rp
(10)
.--L -
H
\t, =;J;
frt-bl
,.LL.t
(1
3)
Flow coefficient, @
O: Q. or g
/
ND'
(14)
Continuecl
Ma
?'/tl
] o:
FLI,'ID
::t:i::.
QL'NF'MEzu'[',
.'
=a 0.1
o.o
.0,1
-0.2
,41
(1
5)
The first step in any comparison is to obtain a set of nondimensional curves for rhe shop test or predicted data. If
they were not provided by the OEM, they can be obtained
by iteratively calcuiating the p and r2values from the given
values of discharge pressure and shaft horsepower. This can
be accomplished by guessing a discharge temperarure for the
given discharge pressure until the correct shaft horsepower
has been reached. Once the correcr discharge remperarue
is known, the polytropic head coefficient and efficiency can
be calculated to give a set of nondimensional curves (Fig.
6). These curves will predict the compressor performance
for the given mach numbe. They can be used ro compare
against the existing operating condirions if rhe field mach
numbe is close enough to the mach number for the curves.
The ASME PTC-10 test code defines the maximum shift in
mach number ratio for a certifiecl shop performance resr (Fig,
7). These limitations are good to apply in the field as well. If
pRocESSTNG
blq
F;g"rrr.
1,0
Fn'u l_i
L\flf
tr\
RTAT
ffi
[I{J
iP[Vd
iT
l7\*
l:iff:-*
2.A
'Tr
calculate the field discharge conditions (pressure, temperature, horsepower, etc.) based or the field inlet conditions and
the performance curves (Fig. 8). Even though seal losses do
not affect the calculated horsepower (i.e., they are included
in the lower measured efficiency), they do greatly affect the
predicted horsepower. The seal losses increase the calculated
value of , which moves the predicced operating point farther
ro the right on the performance curves, which always increases
the horsepower.
Exa m pl e perf
Reformer hydrogen
Pz = 275 Psia.
Balance piston labyrinth dimensional data:
data:
Pr = 150 Psia
Tr = 80'F
Pz = 248 Psia
T2 = 191'F
P= o'gz
E = 0.957
1) Calculate the horsepower supplied to the
'
Amperage = 978
Voitage = 4,0A0
Speed = 1,784
Original design:
'7,940 rpm
66
= 100'F
7h
compretii))r,rr,
)F
F t" tJ it l1-) F
1l
'"Affitl",l$i
i \i (;; H i.?
tli
il
Paranreter Measuredfield
pe
248
P2
lbf-ft/lbm
Efficiency, %
Shp
fn
Vf
flD
J tkator
If the
+7 I
= 7,988 hP
Predicted
258
l/tr
191
Head,
Predicted
performance performance
(w/o seal losses) (w/seal losses)
68,898
79
6,129
64,504
65
7,641
182
66,411
76
7,030
(Fig. 10).
The discharge condirions are repredicted with and withouu
seal losses to compare against the measured field results (Tables
5 and 6). The horsepower fo the existing field performance data
must be close to the horsepower supplied by the driver for the
results to be considered valid. As you can see, the con-rpressor efficiency is considerably iowe than what is predicted by the OEM
perfornrance curve. Adding the seal losses (approximately 4-5o/o
of the r:otal flow) to the predicted curves brings the predicted
and actual conditions closer together. However, the rneasuecl
efficiency is just too low to be a balance piston problem alone.
Likewise, the thrust bearing temperarure and axial position were
relative:ly low. Based on the past fouling history in this compressor as well as the high balance piston differentiaI pressure,
the loss in efficiencywas rhought to be a result of fouling. The
comprc:ssor was still moeting the desired discharge pre.ssure, bur
the lou, efficiency was causing excessive horsepower consumption, which was limiting unit charge rate. Because the motor was
oversized and the loss irr efficiency had been gradual, operations
was unaware that the pr:oblem was in the comPressor and not the
motor (i.e., they thought the motor was dirty).
The compressor wa$ pulled because it couldn't be washed in
place due to a lack of adequate case drains. Alarge amount of
ammonia chloride buildup was found in the compressor station.\qry.cpryp.p,rlfll (the diffuser channels had approximately 40o/o
amplitudes were relatively low (< 1 mil) because the fouling was
3o/o
of the predicted.
NOMENCLATURE
Q = sPecific heat, constant Pressure, Btuilbm, 'F
C = specific hear, consrant volume, Btu/lbm, "F
D = imPeller diamerer, in.
= enthalpy, [iu/lbm
,? = head, lbf:ftllbm
hP =
hotstPo*e'
l/=
l/S
speed, rpm
= specific spt:ed,
P - pressure, psia
non-dim.
Select 1 60 at www.HydrocarbonProcessing.com/RS
blq
i"il
ffi
SrT t f\i
G ffi f; {"
f,
P $\fi E hT
Test data
Suction
Discharge
conditions:
_/
'
Predicted data
lnlet
0utlet
lnlet
Discharge
conditions:
Press, psia
150
248.0
150.0
248.0
150.0
252.1
Temp, "F
80
191 ,0
84.9
191 ,0
84.6
182.1
Flow
lcfm
lbmimin
Flowmeter
0 Mole weight
0 Temp., "F
1,940
7,940 rpm
Number
Diametec in.
23
Calculated data
a1
Mole weight
Properties
Compressibility
Enthalpy, Btu/lbm
Specif ic volume, ft3/lbrn
1.975.
outlet
lnlet
0utlet
lnlet
.0026
I.0059
1,0026
1.0059
56.1
561 .6
683.4
5.39
683.4
2 0/
1.129
1.115
5.44
3.94
.131
1.l75
cplcv
,33
1,32
1.33
1.32
3.76
3.84
3,77
3.84
Flow (corrected)
MMscfd
Predicted Predicted
inlet
1.0026
561 .3
5.44
.130
1.33
3.77
1
Flow (corrected)
outlet
1.0059
673,0
3.83
1
.112
1.32
190.0
lbm/min
2,496.0
2,6'10.0
2,616.1
1.695
lmpeller q/n
1.789
1.792
14.0
Performance
120.1
Predicted
Performance
(corrected)
pefformance
1.61
1.56
1,48
Polytropic head
64,509
64,195
66,417
0.545
0.541
0.561
1.000
1.000
1.000
Polytropic effy.
0.65
0.68
0.76
Gas horseoower
7,497
7,497
6,892
Shaft horsepower
7,647
7,647
7,030
Mach no.
0.357
0,35 9
0.35 5
Volume ratio
1,361
1.379
1,421
Polytropic exponent
wl
'F
= compressibiliry
Subscripts:
I = inlet condirions
2 = ourlet condirions
P = balance pisron
D = design condirions
= mechanical
M= mechanical
P = polytropic state
s = isentropic state
.t = corresponds to seal conditions
- 43.
to Improve
,...
".
Symbols
MECH
,,
Seal losses, lbm/min
Flow (predicted)
13,456.9
3.82
lcfm
Q/N
0 Press., psia
lmpeller
161_175.
'
design data
Speed
190
Polytropic
'
q/q
HYDROCARBOI
PROCE5STNG AUGUST
2007
69