Experimental investigation on heat transfer of rocket combustion chambers and cooling channels

Transcription

1 Sonderforschungsbereich/Transregio 40 Annual Report Experimental investigation on heat transfer of rocket combustion chambers and cooling channels By M. P. Celano, H. Rochlitz, D. S. C. Kowollik, O. J. Haidn, P. Scholz AND M. C. Haupt Lehrstuhl für Flugantriebe, TU München Boltzmannstraße 15, Garching Institut für Strömungsmechanik, TU Braunschweig Hermann-Blenk-Str. 37, Braunschweig Institut für Flugzeugbau und Leichtbau, TU Braunschweig Hermann-Blenk-Str. 35, Braunschweig For the validation of numerical tools a detailed understanding of the heat transfer processes taking place in cooled rocket engines is desired. A step by step strategy to investigate the heat transfer response of a sub-scale rectangular rocket chamber is presented. Each experimental configuration is planned to deliver a database to validate different aspects of numerical tools, such as the fluid structure interaction of a steady state or the material failure after multiple engine cycles. The injector characterization of a capacitive cooled combustor with a single-injector is presented and discussed. The design of planned experiments on a 5-injector combustor with identical characteristic geometry parameters compared to the single-injector design is presented. The multi-injector combustor is a modular concept, where multiple individually cooled short segments are used for calorimetric measurements. In a subsequent experiment a fatigue segment with a replaceable fatigue probe is mounted to the combustor in order to study the "dog-house" failure mode with multiple test specimens. Design aspects of such an experiment are presented. Furthermore, an over-scaled and electrically heated cooling channel experiment is presented. The experiment is designed to validate large eddy simulations. First results on Particle Image Velocimetry and on Laser Induced Fluorescence measurements are discussed. 1. Introduction An experimental database of the complex heat transfer processes and the structural response of cooled rocket engines is mandatory for the validation of numerical tools. Therefore, different experiments are designed to analyse different physical phenomena step by step such as the injector characteristics on heat transfer, the so-called "doghouse" effect as structural failure mode and the cooling channel flow characteristics. Injector characterization is a key issue for heat transfer in rocket engines since injectors are key components for propellant mixing and combustion thus for axial and radial

2 330 Celano, Rochlitz, Kowollik, Haidn, Scholz & Haupt distribution of heat release and heat transfer to the combustion chamber walls. The stepby-step development logic of the injector and combustion chamber foresaw development and testing of a single-injector, capacitive cooled combustor with square cross section having identical injector dimensions and contraction ratio and thus identical chamber Mach number as the 2D 5-injector combustor. Any proper injector characterization requires aside its combustion performance, the determination of its resulting axial heat release and wall heat load distribution for a sufficient broad range of operating conditions. Once such data sets are available, they are suited for the validation purposes of numerical tools and a first effort has already been reported in [1]. In order to provide a well-proven data set as a test case for validation of models and numerical methods, great effort has been put into cross-checking all individual measurements and predicted data. In addition to the detailed characterization of the 2D 5-injector combustor an extra fatigue segment with a replaceable fatigue probe is planned for the modular experiments of the cooled rectangular rocket chamber. A lot of life time experiments on combustion chamber structures have been investigated in the past, for example see Quentmeyer [2] and Anderson et al. [3]. A rectangular combustor with replaceable fatigue probe has not been investigated so far. This novel approach allows for the investigation of several fatigue probes under equal boundary conditions. With this approach the structural damage can be inspected and measured in between test cycles. The fatigue probe cooling channel geometry is designed in collaboration with our project partners in order to generate a controlled early failure of the cooling channel wall after a few engine test cycles [4]. A two step approach is planned for the complex fatigue experiment. At first, the experiment is planned with a pure actively cooled segment. If the fatigue probe design proves to be suitable to generate the dog-house failure mode, a second fatigue segment with two windows and film cooling will be built in order to investigate the flow field and the deformation of the cooling channel structure. A detailed understanding of the cooling channel flow characteristics and of the coolantside heat transfer are necessary for designing an optimized cooling circuit of a rocket engine. The design process is usually based on stationary RANS computations, compare [5] for example. The computations are validated and compared with experimental results. In most of the experiments, only integral values of the cooling channel are taken, for instance [6]. Rarely, measurements of time averaged flow fields in cooling channels using gaseous fluids are recorded using large scale channels, for example [7]. Quantitative measurements of the flow characteristics including local fluctuations of the velocity and temperature field under well-defined boundary conditions are not available. Experimental investigations using optical measurement methods can help to understand the cooling channel flow and to validate the numerical models. Therefore, a generic experiment using the high-quality measurement techniques Particle Image Velocimetry (PIV) and Laser Induced Fluorescence (LIF) is designed to investigate the cooling channel flow with a spatial resolution. Applying PIV and LIF simultaneously provides information about the velocity and temperature field at a certain time. 2. The injector heat transfer characterization In order to gain a deeper knowledge on the boundary conditions for the future life duration experiments, efforts are initiated to fully characterize the injector behaviour. For this reason a heavily instrumented heat sink single element combustion chamber is forseen

3 Heat transfer studies on combustion chamber structures 331 Chamber length [mm] 290 Chamber width [mm] 12 Chamber height [mm] 12 Throat height [mm] 4.8 Contraction ratio A cc/a th [-] 2.5 TABLE 1. Combustion chamber dimensions GOX diameter [mm] 4.0 GOX post wall thickness [mm] 0.5 GCH4 diameter [mm] 6.0 Injector area ratio A GCH4 /A GOX [-] 0.7 TABLE 2. Injector dimensions as intermediade step to the final hardware described in [8]. Extensive test campaigns are conducted on a gas-gas shear coaxial injector element over wide range of pressure and mixture ratios. The goal is to fully characterize the chamber wall heat transfer, due to its great impact on the strength, life cycle and effectiveness of the cooling system in a LPRE (Liquid Propellant Rocket Engine), for methane and oxygen as propellants. This set of data is then a benchmark suited for the validtion of models and tools Hardware description In this section a description of the instrumented single element rocket chamber, injector geometry, flow conditions and data analysis procedures, used for the wall heat flux characterization, is presented. The presented test campaign is performed using a modular heat-sink combustion chamber with a inner square cross section, designed for a testing time of up to 4s at a chamber pressure of P c = 20bar and mixture ratio of 3.4. The single-element rocket combustion chamber is depicted in Fig. 1. The inner chamber dimensions are shown in Table 1. In recent years Pennsylvania State University, University of Florida, NASA MSFGC and Beijing University of Aeronautic and Astronautic made many attempts [9] on studying heat transfer in heat sink chambers. A heat sink chamber design is in fact preferable for research purposes because of its simplicity, low structural costs, ease to manufacture and high accessibility for thermocouple installation. The material used for the chamber segments and the nozzle segment is oxygen-free copper (Cu-HCP). For the current study, a single shear coaxial injector element is integrated as shown in Fig. 2. For simplicity, the gaseous oxygen (GOX) post is configured flush with respect to the injection face (no recess). Table 2 shows the main injector characteristic dimensions. To ensure homogeneous injection conditions, in terms of temperature, pressure and velocity profile and to reduce the influence of the upstream feed lines, two porous plates are placed in the oxidizer and fuel manifolds respectively. More details about the experimental set-up can be found in [10]. In a coaxial injector the shear forces between the propellants establish the mixing efficiency. Non-dimensional numbers such as the velocity ratio VR (Eq. 2.1) and the mo-

4 332 Celano, Rochlitz, Kowollik, Haidn, Scholz & Haupt d 25 thermocouples second segment gment 9 pressure transducers 4 thermocouples porous GCH4 porous plate FIGURE 1. Heat-sink combustion chamber fins A GOX GCH4 A GCH4 GOX inlet GCH4 inlet FIGURE 2. Single shear coaxial injector mentum flux ratio J (Eq. 2.2) are employed to characterize injection conditions. VR= v GCH4 v GOX (2.1) J= (ρv2 ) GCH4 (ρv 2 ) GOX (2.2) From the detailed analysis of the test data it is possible to highlight the influence of the variation of these parameters on physical phenomena controlling the injection and the combustion phenomena. In this direction as an example of the test results obtained in the current test campaign, in the following section test case A (nominal P c = 20bar, O/F = 2.66) and test case B (nominal P c = 20bar, O/F = 3.48) are shown in more detail, see Table 3. The values reported are obtained from experimental measurements. For the design of the operating points, the characteristic velocity c is calculated with

5 Heat transfer studies on combustion chamber structures 333 O/F ṁ T J VR Ratio GOX GCH4 GOX GCH4 [ ] [g/s] [g/s] [K] [K] [ ] [ ] Case A Case B TABLE 3. Operating conditions t 0 t 1 t 2 FIGURE 3. Temperature and pressure P c build-up CEA2 [16] and a combustion efficiency ofη C =1is assumed. The mass flow rates in the combustion chamber (GCH4, GOX, GN2 for the purge) are set by sonic orifices in the feed lines and the upstream pressure. Since the pressure values upstream the orifice are influenced by the mass flow rates, flow checks are required in order to accurately set the pressure regulators. The actual mass flows are calculated from the recorded pressure, temperature signals and the orifice calibration data after the test. The use of a heat sink hardware limits the duration of every firing test, so the burn times are chosen to reach stable operation, required for the thermal load measurements. Each of the operating points is run at least two times to ensure the repeatability of the recorded test data. Good agreement is obtained for all load points. To minimize the influence of the igniter on the temperature measurements, the igniter runs at minimum power for only600ms (200ms prior to the opening of the main valves).a typical chamber pressure and temperature output is shown in Fig. 3, where it is possible to recognize a lower value of the combustion pressure associated to the not ideal combustion efficiency. Due to the transient nature of the problem, three time intervals are chosen for the evaluation of the test data. A representative time interval for the starting conditions t 0, a characteristic hot run time stept 1 and a shutdown condition timet 2. The thermocouples, calibrated before the test campaign, ensure a response time of 100 ms, sufficiently small to resolve the transient behaviour of the hardware.

6 334 Celano, Rochlitz, Kowollik, Haidn, Scholz & Haupt Temperature (K) Time (s) FIGURE 4. Temperature signal at 1mm distance from the hot wall, P c=20bar and O/F= Heat Flux Calculations The characteristic of an injector element is mainly defined by the heat flux distribution on the hot wall along the chamber axis. The mixing mechanisms in the near injector field determine the flow conditions and influence flame and flow dynamics. In the present paragraph the distributions of the temperature and heat flux on the inner chamber wall and their transient behaviour during the hot run are shown. The axial and perpendicular distribution of the thermocouples allows the determination of the heat flux evolution along the axis and the reconstruction of the thermal field in the chamber wall. The temperature variations at the measurement locations are direct results of the unsteady inner heat flux, which depends on the local flame and flow structure. Thus theoretically, the temperature measurements at these points can be used to calculate the temperature and the heat flux at the inner walls. Figure 4 shows the thermocouple time traces at 1 mm distance from the inner wall. Measuring transient temperatures with sufficient accuracy for heat flux determination has been found to be challenging due to a significant influence of the response time and sensitivity to positioning of the thermocouples. Previous studies [9] have already made attempts in this direction based on a 1D approximation of the heat transfer in the copper material and applying a correction for the unsteady term. The heat flux calculations based on a 1D approximation are however subject of error, due to the multidimensional nature of the problem. Due to the rectangular cross section of the combustion chamber, the temperature gradient is not significant for only one coordinate direction and it is necessary to account for multidimensional effects. In the context of the present study the heat flux is calculated using two different approaches: a detailed numerical scheme and an intuitive scheme based on experimental data observation in each location where thermocouples are present. In the numerical method, the heat fluxes are obtained by solving the 2D unsteady heat conduction equation, while the experimental approach is based on the concept of accumulation of heat. The methods are explained in detail in [1]. A detailed analysis of the test data is performed using the described procedures. In all of the following figures the heat fluxes are calculated towards the end of the combustion time, when the well defined conditions

7 Heat transfer studies on combustion chamber structures 335 Heat Flux (W/m 2 ) 10 x O/F = 2.6 O/F = 2.6 O/F = O/F = 3.0 O/F = 3.4 O/F = Length (mm) Heat Flux (W/m 2 ) 10 x O/F = O/F = 2.6 O/F = O/F = 3.0 O/F = 3.4 O/F = Length (mm) FIGURE 5. Heat flux distribution along the chamber axis (z axis), P c= 20bar at t 1 FIGURE 6. Heat flux distribution along the chamber axis (z axis), O/F= 3.4 at t 1 are reached at all axial positions and no influence can be assumed to be seen from the start-up transient. Fig. 5 and Fig. 6 show a comparison of the heat flux along the chamber length (z axis, where z=0is the faceplate) at time interval t 1 for different chamber pressures and for different mixture ratios. In both cases the characteristic of the heat release is not significantly shifted along the chamber axis. Due to the small changes in velocity ratio and impulse ratio at the injection, no significant impact can be recognized between the different operating conditions in terms of mixing processes. The dominating phenomenon for the heat flux release is the combustion itself, as can be seen from the increase in steepness and absolute value of the heat flux profile with increasing pressure in the combustion chamber when the velocity ratios are varied. Test repetitions for the different load points are included in the figures and present good agreement. A difference is noticed only with the variation in mixture ratios, between O/F = 2.6 and O/F = 3.4. For the same combustion pressure, in a gas-gas injector, by changing the O/F the total mass flow rate stays almost constant but the injection velocities are varied. In the specific the velocity ratio changes from 0.83 to 1.07, meaning that the annular flow of methane becomes faster than the one of the central jet of oxygen. The direction of rotation of the eddies in the inner mixing layer depends on whether the velocity ratio is greater than or less than unity [11]. Furthermore, the scalar properties in the near field are defined by the effective Reynolds number, the Schmidt number (for gas-gas=1) and the stoichiometric mixing length L s [12]. Smaller values of L s are an indicator of faster mixing. For methane according to [13] this value increases linearly with the velocity ratio. This and the change in vortex direction can explain the difference in heat flux profile shown in Fig. 5 as a consequence of the change in the local mixing processes in the near field region of the coaxial jet. The total length required for the completment of the chemical reaction appears to be constant. The temperature profiles show a decrease in the gradient along the chamber axis and they show a smaller plateau in the last millimiters. This change in shape can be associated to full combustion that seems to be achieved in both cases only towards the end of the chamber. Effects derived from the change in velocity of the coaxial jets affect only the primary aerodynamic mixing in the near injector field, while the chemical process is the dominant mechanism for the completeness of the reaction and the flame length. Exausting explanations still have to be found in order to explain the change of the temperaure profile between the two cases analysed only starting from a position 120 mm downstream the faceplate. Inspection of empirical heat transfer correlation available in literature such as Bartz or Dittus-Bölter indicates that the heat transfer coefficient scales by the pressure to the power of 0.8. For

8 336 Celano, Rochlitz, Kowollik, Haidn, Scholz & Haupt 10 x 106 Heat Flux (W/m 2 *(18.79 bar/pc) 0.8 ) PC =10bar PC =10bar 2 PC=15bar PC=15bar PC=20bar PC=20bar Length (mm) FIGURE 7. Normalization of the heat flux profile, O/F= 3.4 at t 1 the results discussed here, the temperature of the combustion products (3000K) is significantly larger than the measured wall temperature variation in axial direction (ca. 5K, for the p c = 20bar case). Therefore, for the first approximation, the wall heat flux should also scale to thepc 0.8, since the wall heat flux is proportional to the product of heat transfer coefficient and the temperature difference between the fluid and the wall. The results of this non-dimensional analysis are shown in Fig. 7. It can be seen that all heat flux profiles collapse to common trend and all show the same quantitative trends, which means that the heat flux in a gas-gas injector combustor correlates well with the pressure as predicted. The fact that the heat flux profile remains similar across pressures, suggests that the dynamic structures with the combustion flows are pressure independent if all factors remain constant. Similar conclusions are also obtained for GOX/GH 2 in [9,14] Thermtest Simulations The need for a reliable prediction of the thermal behaviour of the institute s rocket combustion chambers has led to the development of the engineering tool Thermtest at TUM [15]. Thermtest allows the simulation of steady as well as transient thermal behaviour of cooled or uncooled structures over a wide scope of chamber materials and cooling fluids. While the heat conduction inside the chamber material is solved by a 3D finite difference method, the convective heat transfer is implemented by empirical Nusselt correlations. The advantage of this approach is a satisfying accuracy maintaining a reasonably fast simulation of the conjugate heat transfer from the hot gas into the cooling fluid. Thermtest utilizes 1D hot gas properties acquired from the NASA computer program CEA2 [16].The evolution of temperature caused by reaction kinetics and atomization processes is generally neglected as it is not taken into account in CEA2. The fluid properties needed for heat transfer calculations near the hot chamber wall are obtained assuming an equilibrium composition frozen reactions Temperature-Pressureproblem. The hot wall heat transfer coefficient is usually calculated from a modified formulation proposed by Sinyarev. Information on Thermtest as well as a comparison with experimental data and calculations from commonly available CFD code has been published [17]. The implementation of the propellant combination methane/oxygen and

9 Heat transfer studies on combustion chamber structures 337 Temperature (K) mm 1mm 2mm 3mm Heat Flux (MW/m 2 ) Length (mm) FIGURE 8. Temperature vs axis at z = 273 mm, P c= 20bar, O/F= 3.4 at t Length (mm) FIGURE 9. Heat flux along the chamber axis, P c= 20bar O/F= 3.4 at t 1. the adaption to the new injector characteristic and chamber design has required code adaption and validation. Inspired by work of [18], a correction function for the heat load characteristic of generic coaxial injector elements has been applied, in order to take into account the injector mixing behaviour. The complete correlation is presented in Eq The hot gas film coefficient is described with three parameters: c 1, c 2 and l max. In order to account for the reduced heat transfer at the beginning of the combustion chamber, where the combustion evolves, the heat flux is shaped as a function of the chamber axis z. The length l max is the axial position where the heat transfer is at its maximum value and the combustion can be considered to be finished. α w,corr =c 1 α w,0 ((1 c 2 )+c 2 tanh( z l max π)) (2.3) The values of the parameters are established via an optimization problem based on the experimental data. Calculated and measured temperatures are compared and the error between these two values is minimized. The error is normalized over the number of experimental tests and sensors taken into account. Simulation results obtained with the optimization procedure are shown in Fig. 8 and in Fig. 9. Interesting is to observe the values for the temperature also in the nozzle region, values that can be not experimentally determined with the current experimental set-up. The values obtained are then extrapolated by means of the use of the empirical correlation. For the temperature profile along the chamber axis showed in Fig. 8 at time t 1,the results obtained from Thermtest simulation are compared with experimental ones for thermocouples at 3 different depths in the chamber wall. They show a good agreement both in terms of absolute values and trend. 3. Final combustion chamber configuration The preliminary studies presented in the previous paragraph are foreseen as a preliminary step for the final definition of the design of the five element combustion chamber reported in Fig. 10. The chamber will be water cooled and features a large number of short segments in order to allow the definition of the heat flux profile from the variation of the cooling water enthalpy. The modular configuration of the chamber will also allow the variation of the chamber length, in order to evaluate the influence of the chamber characteristic length on the chamber performance and injection processes. Furthermore,

10 338 Celano, Rochlitz, Kowollik, Haidn, Scholz & Haupt FIGURE 10. Combustion chamber final design this design facilitate the installation of the segment for the fatigue studies of section 4 in the most opportune position in order to achieve complete-combustion and prescribe well established boundary conditions. 4. Fatigue segment for cooling channel damage characterization The planned 2D 5-injector rectangular combustor presented in section 3 is a modular concept. In order to analyse the life time of cooling channel structures separate test campaigns with an additional fatigue segment mounted to the combustor are planned. The fatigue segment is placed behind the axial position of finalized combustion. In the previous study of section 3 the heat flux distribution is characterized with the calorimetric short segment measurements Fatigue segment design A cut view of the fatigue segment assembly with an actively cooled inspection plate opposite of the fatigue probe is shown in figure 11. The entire segment is 160 mm long. The fatigue segment itself and the inspection plate are overcooled with water in order to guarantee material integrity for the experimental investigation of several fatigue probes. Cross cooling channels are placed at the beginning and at the end of the fatigue segment. A separate faceplate is used to simplify the fatigue probe geometry and to provide a uniform clamping surface between the fatigue probe and fatigue segment. The evolution of cooling channel damage shall not be influenced by side effects. An inspection plate for deformation measurements reduces side effects to a minimum because the fatigue probe does not need to be disjoined between test cycles. Therefore, the sealing and tensioning of each fatigue probe is not altered through the life time of the cooling channel structure Fatigue probe design The fatigue probe cooling channel geometry is defined through an extensive parameter study presented in [4,8]. The cooling channel length is 100 mm. The hot gas wall thickness is defined with 1 mm. The 15 cooling channels are 8 mm high and 2.5 mm wide with a fin thickness of 2 mm. The rear wall thickness is 10 mm. The design temperature of the hot gas wall for the center cooling channels is 950 K. At these high wall temperatures the desired weakening of the material properties of the copper alloy occurs.

11 Heat transfer studies on combustion chamber structures 339 FIGURE 11. Cut view of the fatigue segment assembly with an actively cooled inspection plate opposite of the fatigue probe (Arrows indicate hot gas and cold fluid flow directions). The heat flux is expected to reach up to 7 MW/m 2 at P c = 20 bar combustion pressure and around 14 MW/m 2 at P c = 40 bar combustion pressure. A small thermal gradient generates an elevated cooling channel wall temperature above 900 K. At these high temperatures water boiling in the cooling channel boundary layer cannot be avoided at cooling pressures up to 100 bar. Therefore, a water cooled fatigue probe is not feasible. Supercritical nitrogen gas is used instead. Nitrogen gas operated at 70 bar guarantees no phase shift during the experiment. Special care is taken for the fatigue probe nitrogen cooling system. The area of interest for validation is focused on the three center channels of the fatigue probe. A symmetric cooling around the center channel is planned in order to be able to apply symmetry conditions for the numerical models. The symmetric properties are verified by temperature measurements using thermocouples in the cooling channel fin of the fatigue probe at several positions of the cross section. In addition to this, the heat up of the fatigue probe is measured in chamber axis for one cooling channel fin. The cooling fluid flow is shown through arrows in figure 12. In order to provide appropriate boundary conditions for validation a closed loop pressure and mass flux control is chosen. In addition to this, pressure and temperature are measured in the vicinity of the inlet and outlet of the fatigue probe cooling channels. The inlet flow is pressure controlled in a closed loop for the entire number of cooling channels. A perpendicular inlet with a comparatively large chamber is used to provide an evenly distributed inlet condition for the cooling channels. The closed loop mass flux control is realised individually for the three center cooling channels. The outside cooling channels are combined into two outlets whereat each mass flux is separately controlled in a closed loop, see the outlet manifold in figure Measurements The fatigue measurements are planned in two steps. The first step consists of the previously described fatigue segment without optical access but with an actively cooled

12 340 Celano, Rochlitz, Kowollik, Haidn, Scholz & Haupt FIGURE 12. Fatigue probe and manifold design inspection plate, see figure 13(a). The overcooled segment allows for long time test campaigns, where multiple fatigue segments can be tested in order to generate a database for material failure and validation. A typical test cycle consists of 2 seconds pre cooling, 20 seconds hot run and 18 seconds post cooling. In between the test cycles the inspection plate is dismounted and stereoscopic measurements of the fatigue probe surface are planned in order to measure the growing cooling channel deformation response. In a second step a fatigue segment with more measuring equipment is planned. The idea is to measure the flow field in the vicinity of the center channels of the fatigue probe through PIV and to measure the deformation of these cooling channels throughout a test cycle. This is a novel and challenging approach for combustion chamber measurements. Two windows are necessary for the PIV measurements. A cut view of the second fatigue segment is shown in figure 13(b). The window perpendicular to the fatigue probe is needed for the PIV camera. The opposite window is big enough to place the necessary laser sheet at different locations of the probe. In addition to this, a stereoscopic measurement system is calibrated for this optical access. A combination of regenerative cooling and film cooling is necessary to guarantee enough cooling for the frame, the sealing and the two windows. 5. Cooling channel flow characterization 5.1. Hardware setup and qualification The generic cooling channel consists of a straight and a curved part. Thereby the influence of the curvature on the flow and on the heat transfer can be measured. The cooling channels design is restricted by various requirements. The best suitable cooling channel geometry and proper boundary conditions is achieved considering a detailed parameter study and are given in [19] and [20]. It is found that the best suitable cooling channel geometry has a width of 6 mm, an aspect ratio of 4.3 (the height is 25.8 mm) and a length of 600 mm. The radius of curvature is set to 60 mm. The bulk temperature is set to T b = 333 K and the wall temperature to T w = 373 K. Two cases with different

13 Heat transfer studies on combustion chamber structures 341 (a) With actively cooled inspection plate. (b) With two optical accesses. FIGURE 13. Cut view of the two fatigue segments. feed line secondary structure PMMA-cooling channel pressure transducer PT1000-RTD for measuring and controlling the wall temperature insulation heat nozzle PT100-RTD for measuring the bulk temperature FIGURE 14. Generic cooling channel (straight part). flow rates (50 l/min and 200 l/min) are defined. The first case has a feasible number of cells when doing LES computations and hence, can be used as a validation for LES simulations. The second case has a large heat flux density, Prandtl, Reynolds, Nusselt and Dean number and a large heat-transfer coefficient which are not achieved in the first case. The generic cooling channel consists of a heated bottom area made from copper Cu-HCP. The bottom area is the tip of a heat nozzle distributing heat added by cartridge heaters uniformly in time and space. The wall temperature is controlled by PT1000-RTDs which are placed next to heat nozzle s tip. The generic cooling channel s upper side and the side walls are made from polymethyl methacrylate (PMMA) because of its high transmittance. The high transmittance allows optical access to the flow which is necessary for applying optical measurement methods (PIV and LIF). The setup of the generic cooling channel is shown in figure 14.

14 342 Celano, Rochlitz, Kowollik, Haidn, Scholz & Haupt l Tempe Temperature (a) Infrared thermography. (b) Extracted temperature distribution. FIGURE 15. Qualification of the experimental setup. A qualification of the setup is conducted to ensure that the boundary conditions are defined as expected. The heat nozzle s wall temperature distribution in time and space is measured using infrared thermography. The core temperature in the heat nozzle can be assumed to be constant if the surface temperature is constant. The heat nozzle is coated with Nextel Velvet Coating (RAL tiefschwarz matt). Nextel Velvet Coating is chosen because of its very well known properties and its high emissivity of The infrared camera is of the type Infratec ImageIR 8300 and a 25 mm objective is used. The exposure time is set to 600µs allowing to measure a maximal temperature of approximately 150 C. A desired wall temperature of 100 C is set. An infrared image of the temperature distribution is given in figure 15(a). The temperature along the plotted line which is almost the tip of the nozzle temperature parallel to the cooling (spatial distortions are existent due to lens distortions) channel is extracted and shown in figure 15(b). The temperature is almost constant except a global minimum at 53 % of the relative position. The minimum is a result of a water droplet which covered the temperature nozzle due to condensation. The mean deviation is 0.27 K despite the described minimum. This temperature distribution is almost equal over space and is accepted. The temperature mean deviation over time is small compared to the mean deviation over space. Hence, the test results yield that the boundary conditions are in a very constant and well known state and the experiment can be used for detailed measurements Measurement technique and experimental setup Particle Image Velocimetry (PIV) and planar Laser Induced Fluorescence (LIF) are used as the main measurement techniques to qualify the flow Measurement technique PIV is an optical, non-invasive measurement technique which is based on the shift of seeding particle in the flow. The particles are illuminated twice by a laser pulse. A camera takes one frame at each laser pulse. Hence it is possible to reconstruct the vector

15 Heat transfer studies on combustion chamber structures 343 field using the information about the particle shift and the time between the pulses. The seeding particles used are silver-coated hollow glass spheres with a diameter of 10 µm. LIF is also an optical, non-invasive measurement technique that uses temperature sensitive fluorescent dyes excited by laser light to determine the temperature. The dye is activated by the photons of a laser light with an appropriate wavelength to an excited state. Fluorescence is the spontaneous transition from the excited state to the ground state. Part of the absorbed energy is emitted during the transition. Hence, fluorescence occurs at a longer wavelength than absorption. Rhodamine B is used as the fluorescence dye. The fluorescence intensity depends upon temperature allow gathering information about the temperature field. Rhodamine 110 is used as an additional fluorescence dye. The fluorescence intensity of Rhodamine 110 is nearly temperature independent providing information about the incident laser light sheet. The emission wavelength of Rhodamine 110 is different to that of Rhodamine B allowing their combined use. This combination of two different dyes is called two-color / two-dye technique. The ratio between the two fluorescence intensities allows an accurate temperature determination because the ratio is almost independent of the incident laser intensity Experimental setup The PIV and LIF experimental setup is presented in figure 16. The laser light source is a Nd:YAG double pulse laser with a wavelength of 532 nm and an energy of 60 mj per pulse. The laser light sheet with a thickness of approximately 1 mm is focused by a plano-concave lens with a focal length of -50 mm and a plano-convex lens with a focal length of 100 mm. The light sheet is formed by a cylindrical lens with a focal length of -25 mm. Three Imager pro / prox 11M cameras with Tamron SP AF 180 mm objectives are used. One camera records the PIV Images, one camera (LIF-T) records the fluorescence of the temperature sensitive dye Rhodamine B and the third camera (LIF-C) records the fluorescence of the temperature insensitive dye Rhodamine 110. Suitable bandpass filters transmit the wavelength of interest to each camera. The bandpass filter for the PIV camera transmits the green laser light and the filters for the LIF cameras transmit the wavelengths in the regions of maximum emission intensity of the fluorescent dyes. The camera apertures are set to 11 for the PIV camera to increase the depth of focus and to 3.5 for the LIF cameras to maximize the brightness of the fluorescence images. The LIF cameras are operated with 2x2 hardware binning to increase the signal to noise ratio. All three cameras are focused on the area of interest (AOI) as shown in figure 17. An image correction is applied during post processing to consider the different viewing angles. Thus, a beam splitter is not used resulting in an intensity gain. The synchronization between the laser and the three cameras is timed by a programmable timing unit. The maximum recording frequency is 1.5 frames per second limited by the data transfer rate between the cameras and the data acquisition device Preliminary results An example for a mean velocity vector field is given in figure 18. The field of view has a length of 50 mm and a height of 14 mm. The raw images are corrected in space and the average of all images is subtracted to increase the signal to noise ratio of the intensity peaks caused by the seeding partikels. Then, the vectors are calculated for every double frame image using the multipass technique with decreasing window size from 64x64 px down to 16x16 px and 50 % overlap. The mean velocity vector field is based on 1500 double images. Only every third vector is shown (in x and y direction) for a better visual clarity. The flow is mostly uniform from left to right and the velocity boundary layer on

16 344 Celano, Rochlitz, Kowollik, Haidn, Scholz & Haupt Light sheet optics Test section Camera LIF-C Camera PIV Camera LIF-T FIGURE 16. Experimental setup. the bottom is clearly visible. Further data analysis considering for example RMS values, wall and bulk temperature variations etc. need to be performed. A first fluorescence intensity decay study of Rhodamin B illustrates figure 19 for a concentration of 54 µg/l with changing temperature. The fluorescence intensity drops approximately 2-4 % / K which is in accordance to the literature (for example [21]). LIV data will be evaluated and described in more detail in prospective work including temperature distribution in the cooling channel. 6. Summary Experiments are conducted over a large range of pressures (5 to 20bar) and mixture ratios (2.6 to 4.0) in order to satisfy the need for a major understanding of the injection and combustion processes and of more reliable prediction of the thermal behavior for the propellant combination methane-oxygen. A single-element combustion chamber is

17 Heat transfer studies on combustion chamber structures 345 LIF-T PIV LIF-C a FIGURE 17. Schematic illustration of the camera setup. [m ] y [mm] [mm] FIGURE 18. Mean velocity vector field te it [ u t ample temperature [ C FIGURE 19. Fluorescence intensity decay study of Rhodamin B.

18 346 Celano, Rochlitz, Kowollik, Haidn, Scholz & Haupt designed and tested. Detailed wall temperature measurements and derived heat flux data sets are obtained for GOX/GCH 4. These data sets are valuable for both injector design and code validation. The temperature and the chamber wall temperature distributions seem to have no direct pressure dependence. The lack of pressure dependence and only slight dependence on the propellant injection velocities suggests that the basic dynamic structures of the combustion process are mainly dominated by the injector and chamber geometry. The same trend was observed for all investigated load points. Difference in the heat flux profile can be noticed if the O/F is changed between 2.6 and 3.4, while keeping the combustion pressure constant. The heat flux peak is shifted along the chamber axis but the total length required for the completment of the chemical reaction appears to be constant. Effect derived from the change in velocity of the coaxial jets affect only the aereodynamic mixing in the near injection field, while the chemical process seems to be the dominant mechanism for the completeness of the reaction and flame length. Full combustion seems to be achieved only towards the end of the chamber. Simulations, conducted with Thermtest, shows good agreement with the experimental results. An optimization, according to the test data, of the heat transfer model in order to match the injector characteristics is performed and it shows to approximate the injector thermal behaviour. Futhermore, the two step design concept of a fatigue segment with replaceable fatigue probe to analyse material failure of cooling channel structures is presented. An actively cooled inspection plate on the opposite side of the fatigue probe allows the measurement of structural response in between test cycles. The challenging aspects of a design with individually fed cooling channels and optical accesses are presented and discussed. In addition to the combustion chamber experiments, an experiment of a generic cooling channel flow is designed. The motivation for the generic cooling channel experiment is given and the hardware setup and qualification are described and evaluated. The measurement technique (PIV and LIF) and the experimental setup are depicted. Preliminary results are described and discussed and an outlook for future work is given. Acknowledgments Financial support has been provided by the German Research Foundation (Deutsche Forschungsgemeinschaft DFG) in the framework of the Sonderforschungsbereich Transregio 40. References [1] CELANO, M. P., SILVESTRI, S., SCHLIEBEN, G., KIRCHBERGER, C., HAIDN, O. J., DAWSON, T., RANJAN, R. AND MENON, S. (2014). Experimental and Numerical Investigation for a GOX-GCH 4 Shear-Coaxial Injector Element. Space Propulsion Conference 2014, Köln. [2] QUENTMEYER, R. J. (1977). Experimental fatigue life investigation of cylindrical thrust chambers. AIAA , AIAA/SAE 13th Propulsion Conference, Orlando, Fla., July [3] ANDERSON, W. E., SISCO, J. C. AND SUNG, I. K. (2003). Rocket Combustor Experiments and Analyses 14th Annual Thermal & Fluids Analysis Workshop (TFAWS), August

19 Heat transfer studies on combustion chamber structures 347 [4] FASSIN, M., KOWOLLIK, D. S. C., REESE, S. AND HAUPT, M. C. (2014). Design studies on the structural failure of cooling structures by a viscoplastic damage model. SFB/TRR 40 Annual Report [5] TORRES, Y., STEFANINI, L. AND SUSLOV, D. (2009). Influence of Curvature in Regenerative Cooling System of Rocket Engine. Advances in Aerospace Sciences Vol. 1 - Propulsion Physics, [6] MEYER, M. L. (1997). The Effect of Cooling passage Aspect Ratio on Curvature Heat Transfer Enhancement. NASA TM [7] NEUNER, F., PRECLIK, D., POPP, M., FUNKE, M. AND KLUTTIG, H. (1998). Experimental and Analytical Investigation of Local Heat Transfer in High Aspect Ratio Cooling Channels. AIAA-Paper [8] CELANO, M. P., ROCHLITZ, H., KOWOLLIK, D. S. C., HAIDN, O. J., SCHOLZ, P. AND HAUPT, M. C. (2013). Subscale GOX - GCH4 thrust chamber design for life time prediction. SFB/TRR 40 Annual Report [9] CONLEY, A., VAIDYANATHAN, A. AND SEGAL, C. (2007). Heat Flux Measurements for a GO2/GH2 Single-Element, Shear Injector. Journal of Spacecraft and Rockets, 3-44, [10] CELANO, M. P., SILVESTRI, S., SCHLIEBEN, G. AND KIRCHBERGER, C. (2013). Injector Characterization for a GOX-GCH 4 Single Element Combustion Chamber. 5TH European Conference For Aeronautics and Space Sciences (EUCASS), München. [11] WERNER, J. A., CLIFFORD, D., FRIELER, E., GR&EACUTE AND TAR TRYGGVA- SON (1992). Vortex structure and dynamics in the near field of a coaxial jet. J. Fluid Mech., 241, [12] SCHUMAKER, S. A. AND DRISCOLL, J. F. (2009). Coaxial turbulent jet flames: Scaling relations for measured stoichiometric mixing lengths. Proceedings of the Combustion Institute, 32-2, [13] SCHUMAKER, S. A. AND DRISCOLL, J. F. (2012). Mixing properties of coaxial jets with large velocity ratios and large inverse density ratios. Physics of Fluids, [14] MARSHALL, W., PAL, S., WOODWARD, R. AND SANTORO, R. (2005). Benchmark Wall Heat Flux Data for a GO2/GH2 Single Element Combustor. AIAA [15] KIRCHBERGER, C., SCHLIEBEN, G. AND HAIDN, O. J. (2012). Assessment of Analytical Models for Film Cooling in a Hydrocarbon/ GOX Rocket Combustion Chamber. 48th AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, JPC. [16] MCBRIDE, B. J. AND GORDON, S. (1996). Computer Program for Calcualtion of Complex Chemical Equilibrium Compositions and Applications. NASA Reference Publication [17] KIRCHBERGER C., WAGNER R., KAU, H. P., SOLLER, S., MARTIN, P., BOUCHEZ M. AND BONZOM C. (2008). Prediction and Analysis of Heat Transfer in Small Rocket Chambers 46th AIAA Aerospace Sciences Meeting and Exhibit, 7-10 January 2008, Reno, Nevada. [18] VASILIEV, A. P., KUDRYAVTSEV, V. M., KUZNETSOV, V. A. ET AL. (1993). Fundamentals of rocket engines theory and calculations. Book 1 and 2, Moscow Russia. [19] CELANO, M. P., ROCHLITZ, H., KOWOLLIK, D. S. C., HAIDN, O. J., SCHOLZ, P. AND HAUPT, M. C. (2013). Subscale GOX - GCH4 thrust chamber design for life time prediction. SFB/TRR 40 Annual Report [20] ROCHLITZ, H. AND SCHOLZ, P. (2014). Wärmeübergangs-Versuche an einer generischen Kühlkanal-Geometrie. Deutscher Luft- und Raumfahrtkongress 2014.

20 348 Celano, Rochlitz, Kowollik, Haidn, Scholz & Haupt [21] COOLEN, M. C. J., KIEFT, R. N., RINDT, C. C. M. AND VAN STEENHOVEN, A. A. (1999). Application of 2-D LIF temperature measurements in water using a Nd : YAG laser. Experiments in Fluids, 27,