O

ne principal cause of damage to gas turbine hot-section

components is imbalanced fuel distribution due to dirty
or defective fuel nozzles. Having more fuel flowing to
one nozzle and less to another will also cause more NO
x
and CO
emissions because hot and cold spots are simultaneously existing
in the combustion section. Those same hot spots reduce creep
life of the combustion liners, transition pieces, and turbine nozzles
and blades. Luckily, these problems can be minimized by careful
refurbishment and calibration of the fuel nozzles.
Fuel nozzle malfunctions. Gas turbines typically have mul-
tiple fuel nozzles through which fuel is injected into the combus-
tion zone (Fig. 1). Ideally, fuel flowrate through each nozzle should
be uniform and mixing of fuel with air should be equally effective
for each nozzle. If liquid fuel is being injected, atomization of the
fuel droplets should also be identical for each nozzle.
However, flow maldistributions can often occur among the
different fuel nozzles in a turbine. Causes of nonuniform flows
can include:
Manufacturing defects (e.g., machining burrs or nicks)
Improper assembly (e.g., leaking gaskets, mismatched parts)
Changes in nozzle flow area due to erosion or coke and ash
deposits
Cracks in the nozzle due to fatigue
Faulty operation of check valves in fuel lines.
If maldistributions occur, the result will be uneven firing tem-
peratures around the combustion section circumference. Since most
gas turbines do not have thermocouples at the exit of the combus-
tors, these uneven firing temperatures are typically detected by ther-
mocouples mounted further downstream. Single-shaft gas turbines
usually have an array of thermocouples installed around the cir-
cumference of the turbine outlet, while multiple-shaft turbines typ-
ically have a thermocouple array around the power turbine inlet.
Almost all gas turbine control systems monitor exhaust tem-
perature spread (i.e., maximum minimum) and issue an alarm
when this reaches an OEM-specified value. The challenge for the
turbine operations and maintenance staff is then to figure out the
source of the high-temperature spread.
Some swirl is in the flow as it passes through the turbine. Con-
sequently, one cannot assume that an abnormally high or low
temperature at the 4 oclock position in the exhaust was caused
by the fuel nozzle located at the 4 oclock position in the com-
bustion section. It may come from the fuel nozzle at the 2 oclock
or 6 oclock position, or even further away.
Gas turbine performance monitoring software has now devel-
oped to the point that it can calculate the amount of swirl in the
flow and rotate the position of the exhaust thermocouples to indi-
cate the location of the fuel nozzle that produced the flow passing
by each thermocouple.
1
An example of the graphical output from
this type of analysis is shown in Fig. 2.
HEAT TRANSFER/ ROTATI NG EQUI PM ENT
Gas turbine fuel nozzle
refurbishment
Follow t hese guidelines t o improve reliabilit y and reduce emissions
J. N. PHILLIPS and P. SIMAS, Fern Engineering, Inc., Pocasset, Massachusetts
Wat er manif old
connect ions
Fuel nozzle body
Liquid f uel
connect ion
At omizing air
connect ion
Wat er inject ion
spray nozzle
At omizing
air cap
Fuel oil
t ransit ion
piece
assembly
Swirl t ip
A t ypical gas t urbine will have 6 t o 16 of t hese f uel
nozzles.
FIG. 1. Example of a combust ion monit or graphical out put . FIG. 2.
HYDROCARBON PROCESSING JANUARY 2004
I
1
Proof onlynot for publication.
Temperature spreads at the gas turbine exhaust would corre-
spond to even greater temperature spreads at the combustor sec-
tion outlet. If one models the turbine section as an isentropic
expansion of perfect gas, from basic thermodynamic principals it
can be shown that temperature at the turbine section inlet, T
t
, is
related to the exhaust temperature, T
x
, by the following formula:
where k is the ratio of specific heats, P
t
/P
x
is the pressure ratio
across the turbine, and the temperatures are taken as absolute val-
ues. Basic algebra then dictates that the difference between max-
imum and minimum temperatures at the turbine inlet, T
tmax
,
is related to the exhaust temperature spread, T
xmax
, by:
Using typical values of 12 for pressure ratio and 1.3 for k in Eq.
2 shows that, when the exhaust temperature spread is 50C, maxi-
mum and minimum temperatures at the turbine inlet will be 89C.
Impact of maldistributions. Nonuniform combustion will
cause higher emissions of either NO
x
(due to hot spots in the
combustion zone) or CO and unburned hydrocarbons (due to
cold spots and poor mixing or atomization).
If too much fuel is injected through one or more fuel nozzles,
the combustion gases exiting from that region will be hotter than
the average. This will reduce life of the hot section parts that are
exposed to the hotter flow.
Creep life of metal components in the hot section of a gas tur-
bine is extremely sensitive to metal temperature. Fig. 3 shows the
relationship between metal temperature and creep life of several
materials typically used in gas turbine hot sections. Note that a
temperature increase of only 50C will reduce the materials life by
an order of magnitude.
The consequences of hot-section component failures caused
by overheating can be quite costly. In extreme cases, combustion
liner failures can allow hot flames to impinge on the turbine pres-
sure casing, which can lead to catastrophic engine failure. Even
before casing failures occur, broken pieces of the liner can pass
into the expander section and cause extensive blade damage.
2
Fuel maldistribution correction. Due to rapid degradation
of hot-section parts life that can take place, corrective action should
always be taken whenever high exhaust temperature spreads are
encountered. While there can be other causes of the high spreads
(see below), the most logical place to start is the fuel nozzles.
When a set of fuel nozzles is removed for testing and refur-
bishment, it should be put through a multistep process called for
in OEM maintenance protocols. During the process, each nozzle
is disassembled, cleaned and inspected. Any worn-out parts are
repaired or replaced, and the nozzles are rebuilt, flow tested and cal-
ibrated to ensure uniform flow. Typical before and after pho-
tos of refurbished fuel nozzles are shown in Fig. 4. A key step in the
process is the inspection, which can help in the problem diagno-
sis. For example, many combustion problems can be diagnosed
by examining the wear patterns on the nozzles.
The flow tests are conducted with an apparatus that supplies the
liquid or gas fuel, and atomizing air and NO
x
suppression water
if applicable, to the nozzle at the OEM-specified conditions. Based
on the flow versus pressure results of each nozzle, adjustments are
made in the flow path geometry to ensure uniform flow at similar
conditions. Typically, a set of nozzles can be tuned to provide no
more than 3% deviation in flow among the nozzles.
Since temperature increase across a gas turbine combustor can
be on the order of 800C (1,440F), a 3% difference in fuel flow
should result in a firing temperature spread on the order of 24C
(43F). This is within typical OEM guidelines; however, based on the
earlier discussion of the impact of metal temperature on creep life, even
smaller spreads could yield a significant improvement in parts life.
In addition to the flow tests, the nozzle spray patterns should
be checked for irregularities, and the nozzle body should be pres-
sure tested to check for leaks and to verify proper gasket seating.
Also, the check valves in the fuel lines leading to the nozzles
should be cleaned, inspected and pressure tested to ensure leak-
free operation.
Case study. A West Coast refinery was experiencing excessive
temperature spreads in the exhaust of its gas turbines. It sent their
fuel nozzles to a lab for testing and refurbishment. Upon receipt,
each nozzle was flow tested to confirm whether the nozzles were
providing uniform flow. As shown in Fig. 5, the flow discharge
coefficients, Cd , of the nozzlesan indication of the flow deliv-
ered for a given supply pressuredeviated by almost 6%. After
cleaning and refurbishing the flow passages, the deviation in Cd
T T
P
P
t max t max
t
x
k k

1
2 ( )
T T
P
P
t x
t
x
k k
=

1
1 ( )
2
I
JANUARY 2004 HYDROCARBON PROCESSING
HEAT TRANSFER/ ROTATI NG EQUI PM ENT
Creep lif e of GT mat erials under 579 M Pa st ress
600
650
700
750
800
850
900
100 1,000 10,000 100,000
Time t o rupt ure, hr
M
e
t
a
l

t
e
m
p
e
r
a
t
u
r
e
,

C
Rene N5
Rene 125
Rene 80
Creep lif e of met al component s is ext remely sensit ive t o
met al t emperat ure.
FIG. 3.
A disassembled f uel nozzle bef ore (lef t ) and af t er (right )
ref urbishment .
FIG. 4.
was less than 2%. Once the reworked fuel nozzles were reinstalled
in the gas turbine, the temperature spread was reduced to 22C
(40F) versus an average of 72C (130F) before refurbishment.
The cause and effect were obviously identified.
Other causes of high exhaust temperature spreads.
Two other sources of high-temperature spreads can be the com-
bustion liner and the first-stage turbine nozzles. The metal liner of
the combustion zone is carefully designed to allow a specific
amount of air into the flame zone and then an additional amount
in the dilution zone. Air flow to each zone is determined by the size
and number of holes in the liner. As the liners wear, cracks can
occur and eventually produce new paths for the air flow, which will
distort the combustion process.
A third potential source of nonuniform fuel/air ratios is nonuni-
form flow areas downstream of the combustion zone. During nor-
mal operation, flow through the turbine section first-stage nozzles
is choked. To ensure uniform flow through each combustor upstream
of the first-stage nozzles, the cross-section flow area at the nozzle
throat should be identical. However, nonuniformities can occur
due to manufacturing defects, nozzle erosion or fouling nozzles. HP
LITERATURE CITED
1
Phillips, J. N., P. Levine and S. Tustain, Performance Monitoring of Gas
Turbines, Proceedings of COMADEN 2000 Conference, Houston,
December 2000.
2
Dundas, R. E., D. A. Sullivan and F. Abegg, Performance Monitoring of Gas
Turbines for Failure Prevention, ASME Technical Paper 92-GT-267, June
1992.
HYDROCARBON PROCESSING JANUARY 2004
I
3
HEAT TRANSFER/ ROTATI NG EQUI PM ENT
Jeff rey N. Phillips isvice president of Fern Engineering, Inc.
He holdsa BA in mathematicsfrom Austin College, a BS in mechan-
ical engineering from Washington University, and M S and PhD
degreesin mechanical engineering from Stanford University. Dr. Phillips industrial
experience hasfocused on analyzing gasturbine and combined cycle power plant per-
formance. He hasparticular interest in use of nonconventional fuelsin gasturbines.
Prior to joining Fern, he worked for the Royal Dutch/Shell Group on developing a
coal gasification process.
Paul Simas isthe manager of fuel nozzle refurbishment for
Fern Engineering, Inc. He hasover 20 yearsof experience in fuel noz-
zle testing, refurbishment and design. Recent projectshave included
the fuel nozzle design for landfill gasand development of a device that convertsa GE
dual-fuel nozzle to gas-onlyuse and eliminatesneed for purge air while also allow-
ing the nozzle to be easilyrestored to dual-fuel operation.
0.72
0.73
0.74
0.75
0.76
0.77
0.78
0.79
0.80
0.81
1 2 3 4 5 6
Fuel nozzle number
C
d

-

c
o
e
f
f
i
c
i
e
n
t

o
f

d
i
s
c
h
a
r
g
e
M ax. deviat ion
Bef ore: 5.8%
M ax. deviat ion
af t er: 1.9%
Bef ore rework
Af t er rework
Af t er cleaning and ref urbishment of t he f low passages,
t he deviat ion in Cd was less t han 2%.
FIG. 5.