30-06-2012, 03:52 PM
DESIGNING COMPACT AND ROBUST ROCKET ENGINE COMPONENTS FOR SUSTAINABLE SPACE EXPLORATION
ABSTRACT
Propulsion systems that are light and compact, have the necessary strength and possess robust operating characteristics are essential for space exploration. Here we explore the use of advanced design optimization methods in designing propulsion components such as turbine airfoils with these characteristics. These design methods are applied to a real world design optimization problem derived from the Space Shuttle Main Engine. The system under consideration is an axial turbine with liquid oxygen as the working fluid, the Low Pressure Oxidizer Turbo-Pump. Inspection of the first row of vanes in this turbine showed evidence of high cycle fatigue at the trailing edge near the end walls. Several design objectives such as reduced trailing edge vortex shedding amplitude, increased airfoil strength were required to be met. Two new airfoil designs were generated using the given objectives and constraints. Here we discuss the design objectives and constraints, and the new designs. An assessment of the flow characteristics obtained for the baseline airfoil and the new designs is also provided.
INTRODUCTION
A sustainable space exploration program requires propulsion systems that are light and compact, have the necessary strength, and possess robust operating characteristics. Repair and maintenance in space can prove extremely expensive, and in some cases impractical. System redundancy as a philosophy, with its attendant cost and weight penalty, is not always a solution to the problem. In particular, if the system is susceptible to a set of operating or external conditions, having a duplicate of the same system on board are not an answer to the problem. The required component and system characteristics need to be built in at the design stage. A design process capable of such a feat will certainly be challenged by high-dimensional search spaces, multiple conflicting objectives, numerous constraints and require high fidelity, compute intensive simulation codes. Nevertheless, superior designs that reduce costs and increase reliability and safety are imperative. The advantages derived from such advanced designs, that require little or no maintenance, will not only benefit space based systems but will also result in substantial reductions in cost and system down time for earth based systems.
Fabricating and operating complex systems involves dealing with uncertainty in the relevant variables. In the case of aircraft and rocket engines, flow conditions are subject to change during operation. Efficiency and engine noise may be different from the expected values because of manufacturing tolerances and normal wear and tear. Engine components may have a shorter life than expected because of manufacturing tolerances. In spite of the important effect of operating and manufacturing uncertainty on the performance and expected life of the component or system, traditional aerodynamic shape optimization has focused on obtaining the best design given a set of deterministic flow conditions. Clearly it is important to both maintain near-optimal performance levels at off-design operating conditions, and, ensure that performance does not degrade appreciably when the component shape differs from the optimal shape due to manufacturing tolerances and normal wear and tear. These requirements naturally lead to the idea of robust optimal design wherein the concept of robustness to various perturbations is built into the design optimization procedure.
Several commonly used approaches such as maximizing the mean value of the performance metric, minimizing the deviation of this metric and, maximizing the probability that the efficiency value is no less than a prescribed value are discussed.
Some of the basic steps involved in both robust optimal design as well as reliability-based optimization such as a) identifying random variables and their associated probability density functions, b) reducing this set of variables to a smaller subset of key random variables, to reduce optimization costs and, c) the effective utilization of Monte Carlo techniques to obtain estimates of performance variability or reliability.
The network-based design optimization codes utilize hybrid neural networks to model the behaviour of any objective in design space. The models are then used to search for optimal designs. The goal here is to reduce the number of simulations required to obtain the optimal design. The network models serve as surrogates to the simulation codes.
The design optimization system based on the method of Differential Evolution (DE) is being developed. The algorithmic goals here are to reduce the population sizes, and the number of generations required to obtain the global optimum. The more practical goal is a significant decrease in cost to solution. More recently these methods have been extended to robust optimal design where performance insensitivity to manufacturing tolerances, normal wear and tear, and random disturbances in operating conditions is an additional objective.
The system under consideration is an axial turbine with liquid oxygen as the working fluid, the Low Pressure Oxidizer Turbo-Pump (LPOTP). Inspection of the first row of vanes in this turbine showed evidence of high cycle fatigue at the trailing edge near the end-walls. CFD analysis of the known sources of HCF indicated vortex shedding as the most probable cause It was found that the shedding frequency range overlapped the vane trailing edge flapping mode natural frequency. At the present time the first vane of the LPOTP is replaced at carefully monitored time intervals thus ensuring the safety of the Shuttle flights. The objectives pursued were increased vane strength, decreased shedding amplitude, decoupling of the shedding and vane natural frequencies, minimal impact on downstream rows and, robustness to manufacturing tolerances. The design assessment indicated that all of these objectives were achieved in substantial measure. Here, we redesign the vane to make it lighter in weight and more compact. All of the original objectives and constraints are retained.
Trailing edge vortex shedding is a complex phenomenon that depends on the Reynolds number, the nature of the suction and pressure side boundary layers, the shape of the trailing edge and other factors. Direct numerical simulations (DNS) and detailed experiments are required to fully understand trailing edge vortex shedding. The Reynolds Averaged Navier Stokes (RANS) equations in conjunction with various turbulence models to provide time accurate simulations of the flow through the LPOTP turbine. RANS simulations using a turbulence model are used for preliminary assessment of the shedding phenomenon.
AIRFOIL REDESIGN AND PRELIMINARY ASSESSMENT
Flow computations are performed using the code ROTOR-2 and the Baldwin-Lomax turbulence model. The flow is assumed to be turbulent on both the pressure and suction sides of the airfoil because of the high disturbance levels in the operating environment.
A comparison of the baseline and O5 airfoils:
The first redesign of the baseline airfoil (baseline is from the LPOTP of the SSME) is discussed below
Increase the thickness of the airfoil, particularly in the trailing edge region, to both strengthen the airfoil and increase its natural frequency corresponding to the trailing edge flapping mode
Reduce trailing edge vortex shedding amplitude
Decrease trailing edge vortex shedding frequency to obtain greater separation of frequencies (shedding and natural flap mode frequencies)
Maintain throat area.
Maintain exit flow angle
Design a trailing edge that eases the manufacturing process (facilitate metal flow in casting).
Reduce pressure fluctuations on downstream airfoil rows.
Desensitize shedding amplitude to manufacturing tolerances and normal wear and tear.
The corresponding increase in safety factors for O5 ranged from 3.5 to 6.3. Unsteady rotor-stator computations including the first row of stators (baseline or O5 airfoils), the downstream rotor row and, the second stator row (downstream of the rotor row) are shown. These computations show that replacing the baseline airfoils in the first row with the O5 airfoils results in a modest improvement or, no change, in the flow downstream.
A comparison of the baseline and O6 airfoil:
O5 is much larger than the baseline airfoil. One objective that was not included earlier was compactness.
The shape of the newly designed airfoil (O6) is compared to the baseline airfoil in Fig. 6. Note that O6 has been masked by a nonlinear transform (pending approval from NASA ARC to publish the actual airfoil shape). It is shorter than the baseline, has approximately the same leading edge dimension but is thicker than the baseline in the last 65% of the axial chord. The interior volume per unit span, and thus the weight, of O6 is about 92.8% of that obtained for the baseline airfoil. The corresponding value for O5 is about 151% of the baseline value.
The time-averaged surface pressure distributions for the baseline and O6 airfoils are shown in Fig. Although the axial chord of O6 is smaller than that of the baseline, the pressure distributions are plotted as a function of the axial position normalized by the axial chord © and, are thus of the same extent in this figure. The leading edge load is much higher for O6. In fact, O6 loading is almost uniform to about 65% axial chord and then diminishes near the trailing edge. This higher loading is required to achieve the necessary turning and flow acceleration over a shorter axial distance. On the other hand, the baseline has almost no load at the leading edge and peak loading at about 75% axial chord. Figure 8 shows the surface pressure amplitudes for O6 and the baseline airfoil. The amplitude distribution obtained with O6 is lower than that obtained with both the baseline airfoil and O5 (compare Figs. 3 and 8) on the entire airfoil surface. In particular, O6 shows a reduction of 90% in peak amplitude (which occurs on the pressure side of the trailing edge). In this aspect O6 is an improvement over both O5 and the baseline airfoil.
The decrease in amplitude obtained with O6 is clearly visible here. It can also be observed that O6 yields a lower shedding frequency. Figure 10 shows results obtained from a spectral analysis of the waveforms in Fig. 9. Again, the substantial reduction in amplitude obtained with O6 is observed. It can also be seen that the baseline airfoil sheds at a frequency of 48.6 Kilohertz and O6 sheds at 40.5 Kilohertz. Thus O6 results in a 17% reduction in shedding frequency. This reduction is a little lower than the obtained for O5 (peak at 37.8 Kilohertz). The natural flapping mode frequency has not yet been determined but is anticipated to be close to that of O5 because of increased airfoil thickness in trailing edge region.
As mentioned earlier, the choice of manufacturing method can have considerable effect onshedding characteristics. Casting, with manufacturing tolerances of ± 0.006 inches (resulting in the trailing edge geometry varying by substantially more than 50%), could easily result in a complete loss of any optimal shedding characteristics that are obtained via design optimization. Both O5 and O6 were designed to maintain low shedding amplitudes even in the presence of manufacturing tolerances as large as ± 0.006 inches.
The pressure amplitude distributions for O6, P1, P2, P3, P4 and the baseline airfoil shown in fig. The amplitude distributions obtained with the perturbed airfoil shapes aredifferent. However, on the scale of the peak value obtained for the baseline, all of the peak amplitude values (P1 – P4) are about the same as that obtained for O6. This is a strong indication that shedding amplitude insensitivity to perturbations in airfoil shape was achieved during design optimization. The four perturbations of O6 are only a small subset of all the possible perturbations that can be generated during the manufacturing process. Thus the data obtained for these four cases serve as an indicator of robust performance but do not confirm it for all possible deviations in shape.
The trailing edge of O6 is larger than the baseline trailing edge and completely encompasses the latter. Thus it is expected to be easier to manufacture than the baseline.
Trailing edge shedding is a complex phenomenon that is dependent on a number of factors. A true understanding of the trailing edge flow can only be obtained via a comprehensive experiment or a Direct Numerical Simulation (DNS) of the flow. A first step in this direction is DNS of flow over a turbine airfoil and the associated numerical methodology. Direct numerical simulations of the flow over the baseline, O5 and O6 airfoils are currently being planned. These simulations should resolve any existing doubts of the validity of the reduction in shedding amplitudes that has been achieved and aid in understanding the physical mechanisms underlying the observed data from the RANS computations.
SUMMARY
Propulsion systems that are light and compact, have the necessary strength and possess robust operating characteristics are essential for space exploration. Here we explore the use of advanced design optimization methods in designing propulsion system components such as turbine airfoils with these characteristics. These design methods are applied to a real world design optimization problem derived from the Space Shuttle Main Engine. The system under consideration is an axial turbine with liquid oxygen as the working fluid, the Low Pressure Oxidizer Turbo-Pump. Inspection of the first row of vanes in this turbine showed evidence of high cycle fatigue at the trailing edge near the end-walls.
Two new airfoil designs were generated using the given objectives and constraints. Here we discuss the design objectives and constraints, and the new designs. An assessment of the flow characteristics obtained for the baseline airfoil and the new designs is also provided.
In addition O6 is lighter and more compact than both the baseline and O5. One significant achievement of both Ref. 10 and the current design effort is the desensitization of the shedding amplitude to large changes in trailing edge shape. Although the redesigned airfoils presented here are of practical interest in their own right, an important contribution of this paper is the demonstration that design processes of the kind utilized in this study can be used to significantly improve component strength, reliability and operational robustness and thus enhance safety and reduce lifetime cost.