Rodger W. Dyson

Review of Computational Stirling Analysis Methods

Nuclear thermal to electric power conversion carries the promise of longer duration missions and higher scientific data transmission rates back to Earth for both Mars rovers and deep space missions. A free-piston Stirling convertor is a candidate technology that is considered an efficient and reliable power conversion device for such purposes. While already very efficient, it is believed that better Stirling engines can be developed if the losses inherent its current designs could be better understood. However, they are difficult to instrument and so efforts are underway to simulate a complete Stirling engine numerically. This has only recently been attempted and a review of the methods leading up to and including such computational analysis is presented. And finally it is proposed that the quality and depth of Stirling loss understanding may be improved by utilizing the higher fidelity and efficiency of recently developed numerical methods. One such method, the Ultra HI-Fl technique is presented in detail.

Long-Lived Venus Lander Conceptual Design: How To Keep It Cool

Team (STDT) was commissioned by NASA to study a Venus Flagship Mission potentially launching in the 2020 to 2025 timeframe; the reference lander of this study is designed to survive for only a few hours more than Venera 13 launched back in 1981! This report reviews those studies and recommends a hybrid lander architecture that can survive for at least 1 Venus day (243 Earth days) by incorporating selective Stirling multistage active cooling and hybrid thermoacoustic power.

Fast Whole-Engine Stirling Analysis

An experimentally validated approach is described for fast axisymmetric Stirling engine simulations. These simulations include the entire displacer interior and demonstrate it is possible to model a complete engine cycle in less than an hour. The focus of this eort was to demonstrate it is possible to produce useful Stirling engine performance results in a time-frame short enough to impact design decisions. The combination of utilizing the latest 64-bit Opteron computer processors, ber-optical Myrinet communications, dynamic meshing, and across zone partitioning has enabled solution times at least 240 times faster than previous attempts at simulating the axisymmetric Stirling engine. A comparison of the multidimensional results, calibrated one-dimensional results, and known experimental results is shown. This preliminary comparison demonstrates that axisymmetric simulations can be very accurate, but more work remains to improve the simulations through such means as modifying the thermal equilibrium regenerator models, adding uid-structure interactions, including radiation eects, and incorporating mechanodynamics.

A Computational Aeroacoustic Prediction of Discrete-Frequency Rotor-Stator Interaction Noise - a Linear Theory Analysis

The discrete-frequency noise generated by a rotor-stator interaction is computed by solving the fully nonlinear Euler equations in the time domain in two-dimensions. The acoustic response of the stator is determined simultaneously for the first three harmonics of the convected vertical gust of the rotor. The spatial mode generation, propagation and decay characteristics are predicted by assuming the acoustic field away from the stator can be represented as a uniform flow with small harmonic perturbations superimposed. The computed field is then decomposed using a joint temporal-spatial transform to determine the wave amplitudes as a function of rotor harmonic and spatial mode order. The frequency and spatial mode order of computed acoustic field was consistent with linear theory. Further, the propagation of the generated modes was also correctly predicted. The upstream going waves propagated from the domain without reflection from the inflow boundary. However, reflections from the outflow boundary were noticed. The amplitude of the reflected wave was approximately 5% of the incident wave.

Comparison of Numerical Schemes for a Realistic Computational Aeroacoustics Benchmark Problem

In this work, a nonlinear block-structured CAA solver, the NASA Glenn Research Center BASS code, is tested on a realistic CAA benchmark problem in order to ascertain what effect the high-accuracy solution methods used in CAA have on a realistic test problem. In this test, the nonlinear 2-D compressible Euler equations are solved on a fully curvilinear grid from a commercial grid generator. The solutions are obtained using several finite-difference methods on an identical grid to determine the relative performance of these spatial differencing schemes on this benchmark problem.

Overview 2004 of NASA‐Stirling Convertor CFD Model Development and Regenerator R&D Efforts

This paper reports on accomplishments in 2004 in (1) development of Stirling‐convertor CFD models at NASA GRC and via a NASA grant, (2) a Stirling regenerator‐research effort being conducted via a NASA grant (a follow‐on effort to an earlier DOE contract), and (3) a regenerator‐microfabrication contract for development of a “next‐generation Stirling regenerator.” Cleveland State University is the lead organization for all three grant/contractual efforts, with the University of Minnesota and Gedeon Associates as subcontractors. Also, the Stirling Technology Co. and Sunpower, Inc. are both involved in all three efforts, either as funded or unfunded participants. International Mezzo Technologies of Baton Rouge, LA is the regenerator fabricator for the regenerator‐microfabrication contract. Results of the efforts in these three areas are summarized.

Overview of NASA Electrified Aircraft Propulsion Research for Large Subsonic Transports

NASA is investing in Electrified Aircraft Propulsion (EAP) research as part of the portfolio to improve the fuel efficiency, emissions, and noise levels in commercial transport aircraft. Turboelectric, partially turboelectric, and hybrid electric propulsion systems are the primary EAP configurations being evaluated for regional jet and larger aircraft. The goal is to show that one or more viable EAP concepts exist for narrow-body aircraft and mature tall-pole technologies related to those concepts. A summary of the aircraft system studies, technology development, and facility development is provided. The leading concept for midterm (2035) introduction of EAP for a single-aisle aircraft is a tube and wing, partially turboelectric configuration NASA Single-Aisle Turboelectric Aircraft With Aft Boundary Layer (STARC– ABL); however, other viable configurations exist. Investments are being made to raise the technology readiness level of lightweight, high-efficiency motors, generators, and electrical power distribution systems as well as to define the optimal turbine and boundary-layer ingestion systems for a midterm tube and wing configuration. An electric aircraft power system test facility (NASA Electric Aircraft Testbed (NEAT)) is under construction at NASA Glenn Research Center and an electric aircraft control system test facility (Hybrid Electric Integrated System Testbed (HEIST)) is under construction at NASA Armstrong Flight Research Center. The correct building blocks are in place to have a viable large-plane EAP configuration tested by 2025 leading to entry into service in 2035 if the community chooses to pursue that goal.

Unsteady Validation of a Mean Flow Boundary Condition for Computational Aeroacoustics

In this work, a previously developed mean flow boundary condition will be validated for unsteady flows. The test cases will be several reference benchmark flows consisting of vortical gusts convecting in a uniform mean flow, as well as the more realistic case of a vortical gust impinging on a loaded 2D cascade. The results will verify that the mean flow boundary condition both imposes the desired mean flow as well as having little or no effect on the instantaneous unsteady solution.

On the Need for Multidimensional Stirling Simulations

• identify areas of excessive flow losses due to unintended dead zones, recirculation zones, dissipative turbulence and other losses such as: 5,6, 7 1. Inefficient heat exchange and pressure loss in the regenerator, heater and cooler, 2. Gas spring and working space loss due to hysterisis and turbulence, 3. Appendix gap losses due to pumping and shuttle effects, 4. Mixing gas losses from nonuniform temperature and flow distributions perpendicular to primary engine flow axis, 5. Conduction losses from the hot to cold regions 6. Losses due to combined radiation, conduction and convection in void volumes 7. And in general, inaccurate loss representations due to use of 1-D flow design codes to account for flow and heat transfer through area changes (between components) where phenomena such as flow separation and jetting from tubes or slots into a regenerator may occur. This paper will highlight why multidimensional Stirling analysis is useful and critical for future designs. II. Description of the Problem The dual opposed configuration shown in Fig. 1 8,9, 10 is being developed for multimission use (i.e., for use in atmospheres and space), including providing electric power for potential missions such as unmanned Mars rovers and deep space missions. Figure 1. Dual Opposed Stirling Convertors Reduce Vibration (Schreiber) Only the Stirling engine part of the convertor (Fig. 2) is simulated multi-dimensionally although one could anticipate the entire convertor may one day be prototyped digitally before build-up. • model large, high-power Stirling applications or devices in which the one-dimensional flow assumption breaks down (even some relatively low power, low ∆t, designs approach a ”pancake” shape that could not be well modeled with one-dimensional codes, NASA/TM—2005-213975 2 Figure 2. Stirling Engine Part of Convertor Simulated Only Figure 3. Example Transient Simulation of Stirling ConvertorColored by Temperature An example of the desired full Stirling engine simulation is shown in Fig. 3. It is anticipated that full 3D simulations will provide a level of geometric and flow detail necessary for further design improvements. For example, hardware experiments have shown that large performance gains can be made by varying manifolds and heat exchanger designs to improve flow distributions in the heat exchangers. The kinds of flow and geometry that occur in a Stirling engine which a modeling technique must address are as follows: 1. Oscillating flow which changes the effective conduction, flow friction coefficients and heat diffusion. 2. Low mach number flow (no shocks), 3. Compressible flow due to enclosed varying volumes and heat transfer effects, 4. Laminar, Transitional, and Turbulent flow with Reynolds numbers from 100 to 10000 (based on various length scales pertinent to the flow region), 5. Conjugate heat transfer, thermal dispersion and local thermal nonequilibrium in the porous media regenerator, 6. Micron to Millimeter scale geometry 7. Small separations between sliding solid-fluid interfaces in the seals and appendix gap 8. Deforming Flow Regions Due to Compression and Expansion (i.e., due to moving piston boundaries) III. Current Stirling Cycle Analysis Approaches The techniques of analysis for Stirling engines can be categorized with Martini’s nomenclature. We will use his definition of Third Order and discuss the recent multidimensional category. NASA/TM—2005-213975 3 A. Third Order Analysis Third order analysis uses control volumes or nodes to directly solve the one-dimensional governing equations. Some of the first analysis at this level of fidelity was by Finkelstein, Urieli, and Berchowitz. Some other more recent third order analysis codes are: • The codes by David Gedeon referred to as GLIMPS and Sage are one-dimensional and solve the governing equations implicitly in space and time. The grid includes all time because a periodic solution is assumed/forced. Therefore, it is not possible to model transient startup behavior. • The linearized harmonic analysis code referred to as HFAST solved a steady-state periodic problem in the frequency domain. Again, transient behavior is not modeled. • The code by Martini engineering was never validated but claimed transient modeling capability. It is not clear that rigorous governing equations are being solved since it appears many simplifications based on experimental correlations are used. • Another unvalidated but interesting code by Renfroe attempted a one-dimensional analysis using explicit Runge-Kutta time stepping and a Newton solver to solve the nonlinear equations. • Some other recent one-dimensional solvers are LASER, DeltaE, ARCOPTR, and REGEN3.1. • Finally, the Stirling Dynamic Model (SDM), uses a one-dimensional analogy of an entire Stirling convertor by linking together representative elements within the Simplorer(TM) commercial software package by Ansoft Corp. This tool enables approximate whole convertor dynamic analysis. Recent work attempts to incorporate thermodynamics via David Gedeon’s Sage code described earlier. These approaches will continue to play an important role in Stirling analysis, even as the multidimensional analysis becomes practical to use. A full analysis package incorporating both third and fourth order analysis (defined below) will be presented in Section VI B. Fourth Order (Multi-Dimensional) Analysis At this level of analysis, relatively little has been completed because the third order analysis is faster and for the most part has been an adequate engineering tool. However, to improve efficiency further (to understand and reduce losses) it will likely require a better understanding of the actual flow physics and heat transfer throughout the engine. It is impractical to measure many of these features in practical Stirling devices. Some of these features can be investigated in large scale test modules designed to simulate certain Stirling like processes (such as those being investigated at the University of Minnesota. However, CFD seems best suited for investigation of the details of the fluid and heat transfer physics in a real Stirling device. Some multi-physics analysis tools are listed below. 1. Modified Computer Aided Simulation of Turbulence (CAST) The modified CAST code is based upon the Semi-Implicit Method for Pressure-Linked Equations SIMPLE method but is restricted to two dimensions. It was modified to include oscillatory boundary conditions and conjugate heat transfer. It has been used to model Stirling components but has not been extended to a whole engine simulation tool. A pressure-splitting technique was added to reduce the computational requirements. It was based on separating the thermodynamic and hydrodynamic pressures so that these widely varying scales could be solved with less round-off error and better efficiency. NASA/TM—2005-213975 4 . 2. CFD-ACE This commercial code has been used to model a two-dimensional representative Stirling engine. It is also based upon the SIMPLE technique. The regenerator is not currently modeled correctly since thermal equilibrium is assumed between the gas and solid. This finite volume code can utilize both structured and unstructured grids. 3. Fluent This commercial code is also based upon the SIMPLE method (and PISO method for high speed flows). It currently has similar regenerator modeling limitations in that it is designed for non-oscillating flows. It does however have a sliding interface that could be used for appendix gap modeling on parallel computers. It is being used by several commercial manufacturers for this purpose (references are proprietary). It is also finite volume based and can utilize both structured and unstructured grids. 4. STAR-CD The Simulation of Turbulent Flow in Arbitrary Regions (STAR) code also uses SIMPLE and PISO methods. Its companion product (STAR-HPC) is the parallel computer version. It also has sliding interfaces and deforming mesh capability. It has been used in the related field of internal combustion engine piston modeling, and some Stirling engines have been modeled with it. 5. CFX This code also uses SIMPLE and PISO methods on unstructured grids. It also has sliding interfaces implemented, but no stirling engine modeling with this software has been publicly published. 6. Others While there are other in-house codes, they are usually limited to modeling only specific regions of the Stirling engine such as the regenerator, or the displacer. IV. Recent Whole Engine Modeling The use of two-dimensional CFD models can significantly extend the capabilities, compared to third-order analysis, for the more detailed analysis of the complex heat transfer and gas dynamical processes which occur in the internal gas circuit. More recently, full 3-D calculations have been performed with a commercial code. The temperature results are similar to the second order method results. The multidimensionally computed power, however, was about half of the second order prediction. Moreover, along the axis of the compression space it was found the change of the temperature of the working gas was quite different from harmonic in time. Zhang claims success with modeling a 3D free-piston pseudo-Stirling engine over a 3 month run-time. More recently, a two-dimensional axisymmetric simulation of a full engine has been demonstrated and validated within the observed experimental results taken from two engines. Moreover, it is possible to simulate an entire engine cycle in less than one hour by utilizing modern parallel computer architectures. In short, the capability does now exist to perform whole engine simulations in considerably less time than previously expected. NASA/TM—2005-213975 5 Figure 4. Importance of Properly Modeling the Regenerator for System Studies (Urieli) A very important specific area of modeling difficulty is the regenerator. As shown in Fig. 4, since the regenerator (depending upon one’s definition of effective) has roughly 3 to 40 times more ef

Mean Flow Boundary Conditions for Computational Aeroacoustics

In this work, a new type of boundary condition for time-accurate Computational Aeroacoustics solvers is described. This boundary condition is designed to complement the existing nonreflective boundary conditions while ensuring that the correct mean flow conditions are maintained throughout the flow calculation. Results are shown for a loaded 2D cascade, started with various initial conditions.