Simulations of Bleed-Air Duct Rupture

Ramon Papa Empresa Brasileira de Aeronáutica SA – Embraer

Environmental and Control System Division Av. Brigadeiro Faria Lima, 2170

12227-901 São José dos Campos – SP – Brazil ramon.papa@embraer.com.br

Bento Silva de Mattos

Empresa Brasileira de Aeronáutica SA – Embraer Technological Development Division

Av. Brigadeiro Faria Lima, 2170 12227-901 São José dos Campos – SP – Brazil

bmattos@embraer.com.br

Luis Carlos de Castro Santos Empresa Brasileira de Aeronáutica SA – Embraer

Environmental and Control System Division Av. Brigadeiro Faria Lima, 2170

12227-901 São José dos Campos – SP – Brazil luis.castro@embraer.com.br

Abstract Aircraft engine bleed-air system is of vital importance concerning flight safety and ground operations. The architecture for the duct placement in the aircraft has to take into account the proximity of fuel lines, fuel tanks, avionics, and cargo in order to assure a safe design. To accomplish this, extensive testing and numerical simulations are required. By using computational fluid dynamics analysis hazardous unforeseen situations as well as the timing that they occur can be accurately identified providing the designer with a useful tool to take corrective and preventive actions. The present work reports some two- and three-dimensions simulations of engine bleed-air rupture performed with the fully unstructured CFD code FLUENT1. Some validation cases using the FLUENT code are also provided. The careful choice of flow models, especially turbulence models, and several geometric arrangements allow the evaluation of important design features and characteristics relevant for the safe design of this system.

Introduction Bleed-air system The air-bleed system is the primary air source for the environmental control system (ECS) including anti- and de-icing functions. The primary sources of air supply are the compressor stages of the engines. The auxiliary power unit (APU) acts as a secondary source, while sometimes, during ground operation, an external air supply is also used. The external air enters the engine compressor at atmospheric conditions and is compressed by the several stages raising its pressure to up to 50 psi with temperature usually toping 450 K. A small portion this air is then extracted from the engine through one or more bleed port openings on the side of the engine. The choice of the number and the conditions from which bleed port extracts the air depends on the design and requirements of the system. Commonly one port bleeds the compressor at higher pressure and temperature, being called “high stage”, and is primarily used at extreme energy output requirements, such as anti-ice operation. For normal operation lower settings are required, commonly called “low” or “intermediate stages.” The exact stage can vary depending on engine type, and energy requirements. Copyright 2003 by the American Institute of

Aeronautics and Astronautics, Inc. All rights reserved.

21st Applied Aerodynamics Conference23-26 June 2003, Orlando, Florida

AIAA 2003-3678

Copyright © 2003 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.

Although the choice of bleed ports attempts to minimize the waste of energy, depending of the flight altitude and engine regime in order to guarantee that bleed temperatures remain on safety limits excess energy must be discarded as waste heat. The function of the pre-cooler is to automatically discharge excess energy back into the atmosphere as waste heat. Ensuring that the temperature of the pneumatic duct line is always well below the ignition temperature of fuel, and/or below the critical level for surrounding equipment. For turbine engine bleed air systems, the Federal Aviation Administration elaborated the following rules. (a) No hazard may result if duct rupture or failure occurs anywhere between the engine port and the airplane unit served by the bleed air. (b) The effect on airplane and engine performance of using maximum bleed air must be established. (c) Hazardous contamination of cabin air systems may not result from failures of the engine lubricating system. Usually the hot high-pressure air coming from the engine compressor is cooled to a safer temperature level before it is distributed throughout ducts to pneumatic aircraft systems. The ducts can cross or pass close to many high-sensitive areas like baggage and fuel storage compartments or be located beneath the passenger cabin floor.

The bleed-air rupture computational simulation objective is to assess the local temperature level of neighboring compartments that surround the bleed line, suggest the adequate thermal protection and provide the necessary information for the positioning of sensors in order to detect a leak in minimal time. This is a formidable problem to the flow solvers due to its high complex nature and transient behavior. Turbulence Modeling Direct simulation of the Navier-Stokes equations is only possible for simple low-Reynolds flow. Instead, the Reynolds Averaged Navier-Stokes equations (RANS) are solved (Table. I). RANS equations require closure for Reynods stresses in order to develop turbulence models. For two-equation models, turbulence viscosity is correlated with turbulent kinetic energy (TKE) and the dissipation rate of TKE.

Turbulent viscosity:

Transport equations for TKE and dissipation rate are then solved so that turbulent viscosity can be computed for RANS equations. The k-ε turbulence model is the most widely used in the industry. Therefore, its strength and weaknesses are well documented. The k-equation is derived by subtracting the instantaneous mechanical energy from its time-averaged value, and the ε-equation is formed from physical reasoning. This model is only valid in fully turbulent flows. In order to overcome the intrinsic drawbacks of the Standard k-ε model, some derivatives were developed.

Table I – Solving the Navier-Stokes equations.

One of them is the Realizable k-ε. The basic distinctions of the Realizable k-ε are given as follow

• Alternative formulation for turbulent viscosity

where

is now variable.

• A0, As and U* are functions of velocity gradients.

• Ensures positivity of normal stresses; 02 ≥iu

• Ensures Schwarz inequality ;

( ) 222

jiji uuuu ≤

• New transport equation for dissipation rate, ε;

Concerning the near-wall treatment, Table II provides a good comparison between the existing usual approaches1.

Table II – Comparison of near wall treatments.

Aerothermal Modeling Validation

In terms of fluid dynamics the accurate modeling the bleed rupture phenomena requires the adequate computation of both flow field and heat transfer characteristics of both free and impinging jets. This section presents some validation results in order to raise important issues concerning the application. The choice of FLUENT was the CFD tools is based on previous successful experience by de Mattos et all [2– 5] in the application of FLUENT to usual practical problems, typical of the aerospace industry, in both aerodynamics and thermal analysis. Jet Impingement As a validation case for jet impingement the experiment described in [6 - 11] provides a basis for evaluation of the computational tool used. An axi-symmetrical numerical grid of 26 by 26 mesh points is used to compute the heat transfer Nusselt coefficient on the wall (Figs. 2-3). Although this mesh is coarse for academic purposes it is representative of the limitations of the application CFD for practical problems where the excessive use of grid points in complex configurations may lead to unfeasible computational costs. Another issue is the choice of turbulence models. Although the references indicate several specific/customized turbulence models for jet impingement, the practical approach was to use a general model, available commercially a variety of software tools. In this study FLUENT was used and the realizable k-ε model selected. It predicts more accurately (than the Standard k-ε) the spreading rate of both planar and round jets. It is also likely to provide superior performance for flows involving rotation, boundary layers under strong adverse pressure gradients, separation, and recirculation. For flows with high swirl numbers, high-Rayleigh-number natural convection, high-speed rotating flows, flows involving porous media, and flows in strongly curved domains, the combination of the Realizable k-ε (RKE) with the PRESTO pressure integration scheme provides a reasonable computing time with more than satisfactory accuracy. Since the grid is coarse Enhanced Wall Functions are used. Fig. 4 displays the comparison between the experimental curve and the

results obtained using FLUENT. The advantages of the combination suggested above are evident. Free Jet Expansion

Physical observation indicates that in nature the free jet displays four distinct regions as shows Fig. 5. An important feature of the numerical simulation is the accurate prediction of the rate of expansion of the jet. Experiments are used to propose relations shown Fig. 6, see ref. [12], such that they can be used as a reference for computation. Another axi-symmetrical model is used (2500 cells) in order to evaluate the accuracy of the simulation of the jet velocity decay. Applying FLUENT with both the realizable k-epsilon model and PRESTO pressure integration scheme provides accurate results in terms of jet velocity decay for a wide range of flow velocities (Fig. 7). The results of both validation cases qualify the computational model built on the FLUENT code as a reliable tool to analyze the phenomena arising from a bleed duct rupture. Now, a more specific test case is proposed, simulating the typical geometrical arrangement present in several aircraft designs.

Bleed Air - Duct Rupture Simulations Three-dimensional case To analyze the phenomena of interest a geometrically simplified model was constructed as shows Fig. 8. Consider a prismatic enclosure, which is traversed by a bleed air duct. The practical problem consists on evaluating what would be the temperature of the surrounding compartment walls in case of bleed rupture. In order to evaluate the effect of modeling four cases are proposed: case a) free jet, b) partial duct obstruction c) full obstruction without duct d) misaligned free jet. Case a) is the easier to model, specially considering a more complicated special arrangement. Case b) represents the more typical behavior of a joint failure still leaving the ducts aligned enough to direct the flow downstream. Case c) represents an extreme case where jet impinges directly on an obstruction; case d) can represent the event of the random displacement of the bleed line favoring impingement on the surrounding walls. All the models are in the range of 140,000 hexahedral cells resulting in maximum y+ at the wall in the range of 300 in the worst case, well within suggested range for the use of enhanced wall functions. Based on the previous experiments the Realizable k-ε model with enhanced wall functions is the natural choice. A mass flow of 0.3 kg/s, at a total temperature of 500 K is imposed. The inlet turbulence intensity is 10%. The temperature distribution along a mid-plane crossing the domain is displayed on Fig. 9. The transient behavior of the maximum wall temperature, regardless where it happens, is presented on Fig. 10.

In case d), the direct impingement of the jet core leads the peak temperature to 440 K. This case illustrates both the need of thermal protection on the surrounding walls and also some mechanical restriction of the duct movement. The temperature rises sharply minimizing the possibility of early detection, this clearly represents the most critical case. Case b), on the other hand, represents the least hazardous possibility, when the duct ruptures, but remains mostly aligned conducting the flow downstream. The temperature rise is mild, which would indicate a light thermal protection, but again the need to restrain the duct movement is greatly emphasized. Although case a) might represent a severe geometrical simplification its solution behavior leads to a balance between a more severe restriction in transient temperature analysis, since its peak is higher than the obstruction cases b) and c), although in steady state the final temperature drops to a lower value than the other cases. Case a) stresses the importance of the transient analysis: The temperature peak is significantly higher than the steady state values; therefore the thermal protection has to be designed for the worst case. This arrangement also clearly allows the analysis of the minimum response time required by sensors to detect the leak. The hypothesis of a full blockage of the duct after the rupture, case c), displays a similar transient to case a) leading to a peak of the same level at the same time, but a more pessimistic steady state temperature, which, in terms of design, calls for more thermal insulation. These are some possibilities of massive rupture of the bleed duct. Depending on the size of the enclosure and freedom of movement of the duct several other possibilities exist. The important issue is that CFD can be used as a tool to take the relevant issues into account supporting design decisions and risk analysis. Two-dimensional case A section of an aircraft aft fuselage that contains a bleed duct and a protective shroud was selected to perform a rupture simulation (Fig. 11). A kawool mantle provides thermal insulation for the duct. The bleed air inside of the duct is maintained at 577 K and 200 psi. The numerical calculation for this case was performed employing the Standard k-ε turbulence model instead of the Realizable one as for the previous cases. The thin-wall approach was employed in order to model the shroud. Fig. 12 shows as the temperature contour develops before the air hits the shroud. Just after the hot air leaves the duct through the breach the wave front reaches Mach number 3 driving a significant temperature drop in the region behind the front. A close-up of the flow pattern of the hot-air jet impacting of the shroud reveals that the shroud was

not able to properly avoid the temperature reaching undesirable levels downstream (Fig. 13). Covering the shroud with an adequate insulation provides a typical solution for this problem. Finally, Fig. 14 displays as the flow evolves after the hot air impacted on the shroud. Two important aspects of the flow pattern at this stage can be observed: a ring structure in the region close to the shroud; and a heating of the outer parts of the kawool insulation by the backflow of the hot air. The temperature reached critical levels 0.001 s after the rupture.

Concluding remarks The careful use of a CFD tool (such as FLUENT) can be useful for the design an evaluation of bleed duct ruptures protection systems. The careful judicious of turbulence models and integration schemes added to the use of enhanced wall function can be combined to produce a reliable and consistent analysis procedure. The flow pattern can eventually evolve in a completely different way in a three-dimensional simulation when compared to a two-dimensional one depending on the configuration taken into account. However, the two-dimensional analysis presented here revealed that the temperature reached critical values in sensitive areas in a very short period of time. Even if one considers that a three-dimensional simulation may provide a time scale ten times longer, the period of time is short enough in order to prevent certain probes of properly shutting down the bleed-air flow in the duct. In addition, the two-dimensional analysis also showed that the backflow of the hot air led to a heating of the outer parts of the insulation. A more sophisticated three-dimensional simulation would certainly capture this kind of situation, too. Considering the amount of computational power required for transient simulations as shown in the present paper, a two-dimensional bleed rupture simulation can be attractive by providing some insights of the flow pattern involved. On the other hand, due to the inherent perils of a bleed air-duct rupture, a three-dimensional simulation of the configuration under study should be always performed in order to correctly capture and predict all flow phenomena and situations related to.

References 1. FLUENT 5.3 User’s Guide, Vol. 1-4, FLUENT

Inc., Lebanon, NH, USA, 1996.

2. de Mattos, B. S., Ferarri, M. A. S. and Leahy-Dios, F, “Transonic Euler Flow Calculation around a Transport Configuration with Powered-Engine Effects,” 37th AIAA Aerospace Sciences and Exhibit, AIAA Paper # 99-0529, January 1999, Reno, NV, USA.

3. de Mattos, B. S. and Oliveira, G. L., “ Three-

dimensional Thermal Coupled Analysis of a Wing Slice Slat with a Piccolo Tube,” 18th AIAA Applied Aerodynamics Conference, AIAA Paper # 2000-3921, August 2000, Denver, CO, USA.

4. de Mattos, B. S. and Oliveira, G. L.,

“Aerothermodynamic Flow Simulation Inside an Aircraft Cabin Section with Consoles and Screens,” European Congress on Computational in Applied Sciences and Engineering , ECCOMAS 2001 Conference, September 2001, Swansea, UK.

5. Fernandes, F. C., and de Mattos, B. S.,

“Aerothermodynamic Flow Simulation Inside an Aircraft Cabin with Consoles and Screens,” European Congress on Computational in Applied Sciences and Engineering, ECCOMAS 2001 Conference, September 2001, Swansea, UK.

6. Baughn, J., Hechanova, A., Yan,X., “An

Experimental Study of Entrainment Effects on

the Heat Transfer for a Flat Surface to a Heated Circular Impinging Jet”, Journal of Heat Transfer, 113,pp.1023-1025, 1991.

7. Baughn, J., Shimizu,S., “Heat Transfer Measurements form a Surface with Uniform Heat Flux and an Impinging Jet.”, Journal of Heat Transfer, 111, pp.1096-1098, 1989.

8. Behnia, M., Parneix, S., Durbin, P., “Simulations

of the Jet Impingement Heat Transfer with the k-e-v2 Model”, Annual Research Briefs, Center for Turbulent Research , NASA Ames/Stanford University, pp. 3-16, 1996.

9. Behnia, M., Parneix, S., Durbin, P., “Accurate

Modeling Impinging Jet Heat Transfer”, Annual Research Briefs, Center for Turbulent Research, NASA Ames/Stanford University, pp. 149-164, 1997.

10. Cooper, D., Jackson, D.C., Launder, B. E., Liao,

G.X. “Impinging Jet Studies for Turbulence Model Assessment- I. Flow-field Experiments”, Int. J. Heat and Mass Transfer, 36(10), pp. 2675-2684, 1993.

11. Craft, T. J., Graham, L.J.W. , Launder, B. E.,

“Impinging Jet Studies for Turbulence Model Assessment- II. Examination of the Performance of Four Turbulence Models”, Int. J. Heat and Mass Transfer, 36(10), pp. 2685-2697, 1993.

12. ASHRAE Handbook 1997.

Figures

Fig. 1 – Schematic of a typical bleed-air system of aircraft with rear-mounted engines.

Fig. 2 – Geometry for the impinging jet validation test case.

Fig. 3 – Computational Mesh and Contours of Velocity Magnitude.

D = 0.0403m

13 x 13 D Impingement Surface

Symmetry Axis

Flow Direction

20

40

60

80

100

120

140

160

180

200

220

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2

Distance

Nuss

elt

Rke_Enh_PrestoSke_Enh_PrestoExperimentalRke_Enh_2orderSke_Enh_2order

Fig. 4 – Nusselt number along the wall.

Fig. 5 – Schematic description of the free jet expansion phenomena.

Fig. 6 – Excerpt from ASHRAE Handbook, presenting the relations for free jet expansion computation.

Flow direction

Contour of velocity

Nozzle

Center Line Velocity Decay (V=10m/s)

y = 3,0x-1

0

2

4

6

8

10

12

0 0.5 1 1.5 2 2.5 3distance

velo

city

ASHRAE Handbook 1997

Fluent_Rk-e

Center Line Velocity Decay (V=200m/s)

y = 45.9x-1

0

50

100

150

200

250

0 0.5 1 1.5 2 2.5 3distance

velo

city

ASHRAE Handbook 1997

Fluent_Rk-e

Fig. 7 – Comparison between numerical simulation (FLUENT) and experimental data – Free jet.

Case A) Free Jet

Case B) Partial Duct Obstruction

Case C) Full Obstruction

Case D) Free Jet with 30 deg incidence

Fig. 8 – Test Cases for the Bleed-Rupture Simulation Model.

Case A

Case B

Case C

Case D

300

320

340

360

380

400

420

440

0.00 0.05 0.10 0.15 0.20 0.25 0.30

Fig. 9 –Flow solution of the bleed-rupture simulation model.

Case B

Case A

Case C

Case D

Fig. 10 – Time history of temperature on the top wall.

Fig. 11 – Geometry of the two-dimensional application case.

Fig. 12 – The sequence above shows the bleed air leaving the duct before it impacts on the

shroud (Temperature contours).

Fig. 13 – Details of the jet impacting on the shroud (Temperature contours).

Fig. 14 – Sequence showing the evolution of the temperature contours after the hot air

encountered the shroud.