Document
EXPERI}ffiNTAL STUDY OF TURBINE FUEL THERMAL STABILITY IN AN AIRCRAFT FUEL SYSTEM SIMULATOR Alexander Vranos and Pierre J. Marteney United Technologies Research Center INTRODUCTION An experimental program has been conducted under contract NAS3-2l593 to investigate the thermal stability of aircraft gas turbine fuels. The objectives of this program were (1) to design and build an aircraft fuel system simulator, (2) to establish criteria for quantitative assessment- of fuel thermal degrada- tion, and (3) to measure the thermal degradation of Jet A and an alternative fuel. Accordingly, an aircraft fuel system simulator has been built and the coking tendencies of Jet A and a model alternative fuel (No. 2 heating oil) have been measured over a range of temperatures, pressures, flows, and fuel inlet conditions.
TEST APPARATUS A complete fuel coking assembly, consisting of fuel delivery system, sim- ulator unit, and controller was designed and constructed under this program.
The entire apparatus is shown schematically in Fig. 1. The fuel delivery system,shown schematically in Fig. 2, provided the option of preheating or de- oxygenating fuel with storage at ambient or elevated temperatures. A heat 6 6 exchanger charged with steam at 1.04 x 10 to 2.07 x 10 Pa (50 to 100 psi) was used to preheat the fuel which was stored in an insulated tank. Deoxygenation was conducted at room temperature in the holding tank by sparging for 12 hours or longer with nitrogen gas supplied by boil-off from liquid nitrogen. Run pressures of 6.9 x 105 to 2.07 x 10 Pa (100 to 300 psig) were set by Grove regulators; flows were measured with calibrated turbine meters.
The simulator test unit shown in Fig. 3 was designed to represent an air- craft nozzle/strut assembly, and incorporated provisions for (1) an isothermal wall, (2) variable test parameters, and (3) convenient change of specimens used for characterization of deposit rate by weight gain. Experience at United Technologies Research Center had shown that the weight gain of metal specimens immersed in thermally stressed fuel can be determined accurately, even for short test periods. Resolution of 1 ~g is easily attained with an electro- balance. In order to apply this method to the present problem, a means of placing preweighed specimens in the heated wall was devised which does not cause the flow 'to be disturbed significantly. This requirement was satisfied by using a rectangular channel and flat specimens. Various materials and surface characteristics can thus be. examined.
A schematic representation of the, fuel nozzle/strut simulator design is
shown in Fig. 3. Two closely fitted blocks of high conductivity Berylc~form
a channel of nearly constant wall temperature. Upper and lower block tempera- tures are regulated independently. The channel flow may be laminar or turbulent, and the fuel can be'heated to 533 K in either case. The heated fuel flows through a transition section and nozzle and subsequently dumps into a large cavity (not shown). The cavity walls are fitted with windows to allow photographing the spray. Nozzles are removable and can be inspected after each run. Deposit formation rate is determined quantitatively by measuring the weight gain of the implanted metal discs. The discs are clamped between two plates in a modular arrangement, and the entire assembly is bolted to the.
lower heating block. Four discs spaced uniformly.along the duct allo.w the axial dependence of coking rate to be determined. Mounting and dismountirig of the discs can be done away from the apparatus. Disc diameter is set by the minimum weight gain predicted for a four hour test at 422 K. The entire test assembly was nickel plated at a thickness of - .001 inch and overplated with gold to a thickness of - 50 microinch. Only working surfaces were subject to ' a thickness specification, but the entire assemblies were plated. Except for outer-surface wear. due to handling, no gold was lost from test areas.. Sample discs were 302 stainless steel.
The control system utilizes the following system blocks: (1) two flow- meter conditioners, (2) two pressure transducer conditioners, (3) a Fluke data logger with printer, (4) control and interlock panel, (5) three temperature controllers, and (6) three SCR heater power controllers with transformers.
The data loggerwas programmed to print run conditions every five minutes. The print included date, time, two metered flows and pressures and readings from 28 thermocouples which monitored temperatures of each of the two blocks (7 axially distributed thermocouples), the four modules, fuel at a number of loca- tions, and heater and storage tank and ambient air. Additional full prints or continuous monitoring of individual parameters were also optional. The eight heaters were 0.61 meters (24 inch) long x .01 meters (.375 inch) in diameter Chromalox units rated at 1500 watts each. Four were inserted into each half of the test assembly. Each bank of four was controlled, by one time-pro~ortion ing LFE controller in the console. A single thermocouple at the geometric center of the cover or base furnished a control signal. The rate of heating depended upon desired temperature and flow rate; heating rate was approximately 0.5 K/sec.
One of the aims of this program was to provide baseline tests with Jet A
and comparison tests with ERBS fuel. Properties of Jet A and ERBS are listed
in Table I. Principal differences are the higher aromatic and naphthalene con-
tent. However, ERBS (Experimental Reference Broad Specification Fuel) is not
readily available and a commercial substitute was sought. Number 2 Home Heat-
ing oil (No. 2 HH) was estimated to be an excellent representation of ERBS,
except for a slightly higher aromatic level. Analyses of Jet A and the No. 2
HH used are shown in Table I. The Jet A complied with all AS111 D-1655 require-
ments, except for the aromatic content of 21.7%. A temporary waiver to 22%
is in force, however, and the experimental fuel thus fulfills all requirements.
The No. 2 HH provided good representation of ERBS, with thermal stability
lying between Jet A and ERBS.
The range of variables employed in the tests is given in Table II. Fuel
3 3
flows were varied from 2.14 x 10- to 21.4 x 10- kg/sec (2.5 to 25 gal/hr),
temperatures from 422 to 672 K (300 to 750 F), and pressures from 6.9 x 105 to
2.07 x 10 Pa (100 to 300 psig) for as-received fuels', In addition, fuels
were preheated to 422 K and stored, or were deoxygenated before use. Samples
of fuel were collected after passage through the test section; little differ-
ence was found in composition compared to inlet fuels, Reynolds numbers
ranged from 800 at the inlet (for flow of 2.14 x 10- kg/hr and ambient temper-
ature fuel) to 34,290 at the outlet (for flow of 2.14 x 10- kg/sec with
preheated fuel).
RESULTS
Experiments with Jet A Fuel
Experiments were conducted to investigate the effects of fuel type, wall
temperature, flow rate, pressure, deoxygenation, and preheating on coking rate.
The coking rate for Jet A fuel is shown as a function of wall temperature,
flow rate, and axial position in Fig, 4; the results of deoxygenation are also
included for comparison. Inlet temperature for these tests is approximately
294 K. For the lowest temperature and flow rate, curve (a), maximum coking
rate is found at the location of the second sample disc. The maximum rate is
approximately twice the minimum. With increasing temperature and velocity,
curve (b), the maximum rate is found at the next position downstream, and the
rate is greater, Large increases in temperature at low flow produce even
higher rates, curve (c). The highest rate occurs before the first sample disc
because of the very rapid change in rate with temperature. Thus, no maximum
rate is indicated. Deoxygenation of fuels produces very low coking rates with
a relatively flat distribution.
The dependence of maximum coking rate on temperature is given in an
Arrhenius plot in Fig. 5, Curves are shown for flows of 2.14 x 10- and
6.85 x 10-3 kg/sec. The overall activation energy at 2.14 x 10- kg/sec is 9 kcal/mol,in reasonable agreement with previous work (Ref. 1). The point at 672 K has not been used in placing the curve since the maximum rate actually occurred prior to the first sampling location, as shown above. As expected, the indicated point lies below the curve. It is seen that at higher flow rate, the coking rate is lower at the same wall temperature, yet the dependence on temperature (activation energy) is approximately the same. The lower rate at increased flow points out the importance of liquid-phase heating and homoge- neous reaction effects on the formation of deposits. In both cases the magnitude of activation energy suggests a surface-catalyzed process. Taken together, the two curves of Fig. 5 suggest that an Arrhenius expression is reasonable for the description of coking rate over the range of temperature indicated. Therefore, it appears that the effect of preheating the fuel to 422 K (300 F) for I hour, as indicated by a single point, is to lower the coking rate. No conclusive explanation for this result is available, but it is possible that deposit precursors are removed in. the fuel system prior to exposure to the hotter reaction surface or that bulk phase reactions are not important at this temperature level. The results in Fig. 5 may also be com- pared with the results of previous "isothermal" tests conducted at UTRC, where the fuel was rapidly heated before reaching the working section, and coking was measured at conditions of equal wall and fluid temperatures. The UTRC results lie well below the present results, although the temperature dependence is nearly the same. Because of the bulk fluid heating effect which occurs in the present experiments, and the difference in surface to volume ratio, the two experiments are not directly comparable. For these reasons the magnitude of coking rate should not be comparable. However, the agreement in activation energies (slope of curves) is encouraging.
The effect of pressure on coking rate for both Jet A and No. 2 heating oil is shown in Fig. 6. It is seen that in both cases coking rate increases with pressure. For the given temperature, 533 K,the static pressure exceeds the vapor pressure; therefore, the results are not attributable to phase change.
A possible explanation is that dissolved oxygen comes out of solution at lower pressures, thereby lowering the rate of liquid-phase fuel-oxygen reactions.
No. 2 Oil Tests Coking rate distributions for No. 2 oil at low flow and 603 K are shown in Fig. 7, curve (a); a sharp maximum occurs at the second sampling location.
At the same temperature but higher flow rate, curve (b), the maximum value is considerably smaller and is shifted downstream. In this sense, the behavior parallels that observed with Jet A fuel. At lower temperature (533 K), curves (c,d,e) the effect of flow is reversed, and it is seen that increased flow leads to an increased coking rate. This may be attributed to more rapid heat- ing of the fuel as a result of the transition from laminar to turbulent flow
or this may signal a change in mechanism whereby coking occurs primarily at
the surface at lower surface temperature. In the latter case the coking rate
would be augmented by higher mass transport rates. The effect of flow rate is
summarized in Fig. 8.
The temperature dependence of coking rate for No. 2 oil is shown in Fig.
9. No simple Arrhenius function is observed. There appears to be no effect
of fuel preheating at 533 K (500 F). In view of the observed dependence of
coking rate on flow rate at higher temperature level, the single point at 602 K
(625 F) and 7.70 x 10- kg/sec might show an increased coking rate relative to
the unheated condition at the same temperature and lower flow. This result
would be in accord with the hypothesis that the bulk heating effect is more
important at higher temperature. On the basis of the results shown in Figs. 8
and 9 it is suggested that there may be two regimes of coke formation, a low
temperature, surface-dominated regime having low activation energy and a high
temperature, bulk-reaction dominated regime having higher activation energy.
In the low temperature regime, increased flow increases coking rate, while the
reverse is true at high temperature. Additional tests are required.
REFERENCE
1. Vranos, A.; Marteney, P. J.; and Knight, B. A.: Determination of Coking
Rate in Jet Fuel. International Conference on Fouling of Heat Transfer
Equipment, R.P.I~, Troy, N.Y., Aug. 1979.
TABLE I FUEL PROPERTIES Jet A ERBS Jet A No.2 HH (ASTM D-1655) (Specification) (Analysis) (Analysis) or nominal or nominal Specification Gravity 0.7753-0.8398 0.8438-0.8448 0.8128 0.8478 Viscosity, 80 F, CS 1. 79 3.36 Sulfur, % Wt.
0.3 0.05 0.06 0.23 Aromatics, % Vol. 20* 21. 7 30.4 Ole fins , % Vol.
0.3 0.3 0.3 0.3 Naptha1enes, % Vol. 3.0 7.5 2.0 12.3 Hydrogen, % wt.
13.7 13.0 13.75 12.9 HIe Ratio 1.91 1. 78 Breakpoint Temperature F 500 460 500 392 *Temporary waiver to 2.20 TABLE II TEST VARIABLES 2.5 to 25 gal/h: 3 Fuel Flow Rate (2.14 x 10 to 21.4 x 10- kg/sec) 300 to 750 F (422 to 672 K) Test Temperature 100 to 300 psip; 6 Test Pressure (6.9 x 105 to 2.07 x 10 Pa) inlet - 800 to 27,000 Reynolds Numbers outlet - 6,000 to 34,290 422 K, stored 1 hour before use Preheated Fuel sparged with nitrogen 16 hours Deoxygenated Fuel
FI
4,000 GALS STEAM DRAIN AT seR 100 PSIA POWER SUPPLY FIG. 1 4000 GAL.
STEAM HEATER TANK- FUEL NO.1 FUEL CONTROL AND DELIVERY SYSTEM 275 GAL.
TANK- FUEL NO.2 I CONTROLLER I ....
AND DATA LOGGER FIG. 2 CONVERGING NOZZLE IN MATING PART HAGO BURNER NOZZLE (VARIOUS SIZES) FIG. 3 FUEl: JET A PRESSURE: 2.07 x 10- Pa (300 psig) 1S0 • 672K (7S0F) 2.14 x 10- Kg/sec (2.S gal/hr.)
• S33K (SOOF) t:.. 533 K (SOO F) (OEOX.)
o 603K (625 F) (OEOX.)
V 589K (600F) 17.1 x 10- Kg/sec (8 gal/hr.)
7 2 0.28 x 10- Kg/m -sec l' N E u 0, ::to I w ~ 70 0: CJ z :;;:: SO 0 2 SAMPLE POSITION FIG. 4 COKING RATE VS. 1fT FUEL JET A 1000r---------------------------, o 2.14 x 10- Kg/sec V 6.85 x 10- Kg/sec .. 6.85 x 10- Kg/sec (HEATED)
J
E u 0, ::L UJ ~ a: Cl Z :£ 1.0 1fT 103/K I I I I 750 650 500 400 300 DEGREES F FIG. 5 o JET A • NO.2 H.H.
FLOW = 2.14 x 10- Kg/sec (2.5 gal/hr.)
T=533K
J
E .!2 0> ::L I UJ ~ a: Cl z :£ u a a PRESSURE - Pa x 10-6 FIG. 6 FUEL: NO.2 H.H.
PRESSURE: 2.07 x 10 Pa • 603K (625F) 2.14 X 10- Kg/sec (2.5 gal/hr.)
• 603K 17.1 x 10- Kg/sec (8 gal/hr.)
• 533K (500F) 2.t4 x 10- Kg/sec (2.5 gal/hr.)
o 533K 4.28 x 10- Kg/sec (5 gal/hr.)
A 533K 15.4 x 10- Kg/sec 500 r--------------------------------,
J
E ~100 I UJ ~ a: ~ 50 S2 106-~ ________ ~--------~--------~ SAMPLE POSITION FIG. 7 COKING RATE VS. lIT COKING RATE VS. FLOW RATE FUEL: NO.2 H.H. FUEL: NO.2 H.H.
SOOri------------------~~--, 400. A T= 603K o T=603K (DEOX.)
• T=S33K () Q)
J'
so
E 1: .!2 Ol N ::L E I ~ LU ::L ~ I cr LU (!)
I-' z ~ '-l cr \0 (!) 0 (.)
Z S
I
~
l' I , 1.S 2.0 2.S lIT 10 /K I I I 600 SOO 300
o 2
DEGREES. F FLOW - Kg/sec x 10 FIG. 9 FIG. 8