BIOLOGICALLY INSPIRED MULTIFUNCTIONAL COMPOSITE PANEL WITH INTEGRATED CIRCULATORY SYSTEM FOR THERMAL CONTROL A. D. Williams, M. E. Lyall, L. E. Underwood, and B. J. Arritt Air Force Researc Laboratory, Space Veicles Directorate 3550 Aberdeen Ave, Kirtland AFB, NM 87117 afrl.rvsv@kirtland.af.mil B. Taft University of Wisconsin, Department of Mecanical Engineering 1513 University Avenue, Madison, WI 53706 SUMMARY Multifunctional systems provide significant benefits at te systems level by increasing capability and decreasing mass. Te focus of tis effort is te development of a multifunctional structural panel tat integrates a circulatory system directly into te face-seet and structural ribs of an isogrid panel for greatly enanced termal control. Keywords: multifunctional structures, isogrid panel, termal control, spacecraft INTRODUCTION Satellite termal control systems can be significantly enanced via strap-on type systems, suc as eat pipes, loop eat pipes, and pumped fluid loops. Tese systems improve termal performance by improving eat transport efficiency and tigt temperature control., but add parasitic mass and reduce te usable volume of te satellite. Te disadvantages are tat tey add parasitic mass, reduce te usable volume of te satellite, are expensive, and tend to be complex. In addition, tey require power for temperature control. Recently, tere as been a surge in interest in biologically inspired or bio-mimetic structures were living organisms are used to provide guidance to improve overall performance of aerospace systems. In particular, te incorporation of a circulatory system to significantly improve te system pysical properties, suc as eat transfer, or enance te system by adding a new capability, suc as self-ealing, is of interest. Te key for tese systems is to improve overall performance or add capability witout increasing te overall mass of te system. Because mass is an important parameter for aerospace systems, a key design factor for aerospace structures is te stiffness-to-weigt ratio of te structure. For tis reason, isogrid arcitectures are widely used in aerospace applications because tey provide ig stiffness and torsional rigidity wile minimizing mass. By taking advantage of isogrid arcitectures, fluidic cannels can be integrated into structural panels wit minimal impact to te overall system structural efficiency and mass. Te use of combined fluidic-structural systems as enormous potential for satellite termal control,
and it will be sown tat isogrid structures provide distinctive properties tat allow for a symbiotic, biologically-inspired system. Tis paper will detail te design of a multifunctional composite isogrid panel tat integrates fluidic cannels into te structure for improved termal control. PANEL CONCEPT Circulatory systems in biological organisms consist of tree primary structures. Te first is te main arterial system, wic provides ig fluid flow rates to te organs. Te second is te capillary system, wic distributes and controls te flow witin te organ. Te final is te eart, wic is essentially a sopisticated pump. A system must consist of all of tese elements and must coexist in a symbiotic relationsip to be a truly biologically-inspired system. Te distinctive properties of isogrid structures enable suc a system; te ribs provide te necessary cross-sectional area for ig flow rates and te face-seets provide te surface area for te capillary distribution system. Te concept is depicted in Fig. 1. Back of Panel Front witout Face seet Primary supply cannels Sub supply cannels Distribution cannels Single pass eat excanger configuration Parallel flow configuration Bio mimetic brancing flow configuration Figure 1: Scematic of te biologically inspired panel concept wic integrates fluid cannels into te rib structure and face-seet Te system is designed suc tat te termal control system can be varied to meet te needs of te components. Te ollow ribs provide te primary fluid flow to te system, and te capillary distribution system provides te eat removal capability. To control te temperature of te subsystem, a passive, termally-coupled valve controls te flow. Wen te component is too ot, te termal valve opens and coolant passes from te arteries to te capillaries to cool te subsystem. Te valves are termally activated by te pase cange expansion of paraffin and are termally coupled to te component. Te coolant flows out of te capillaries back into te main artery system were it flows into a radiator for cooling. Te fluid is ten pumped back into te arterial system.
Wen te subsystems are cool, te valve suts off fluid flow so tat te component can retain its eat tus limiting te eater power requirement for te system. Figure sows a general scematic of te termal subsystem. Heat (Conductive Pat) Subsystem Faceseet Heat Paraffin Valve Capillaries Warm Liquid Flow Pump Radiator Main Arteries Cool Liquid Figure : General termal scematic for te symbiotic isogrid structure Te primary advantage of te isogrid arcitecture is te structural performance of te system after integration of te fluidic cannels into te panel. Wereas te structural efficiency of oter approaces degrades wit te addition of fluidic cannels, te performance of te isogrid structure does not 1. To accommodate te fluid flow in te rib, te cross sectional area is increased, wic increases te moment of inertia of te rib. Te results are reductions in te bending and torsional stresses as well as an increase in te critical buckling load. Additionally, by incorporating smaller fluid cannels under te surface of te panel face-seet, te eat transfer capability of te panel surface can be significantly enanced. Te focus of tis effort was to design a proof-of-concept panel tat integrates te primary supply cannels and te sub-supply cannels into a composite isogrid panel. Future design iterations will focus on improving te adaptability and flexibility of te distribution cannel system in te faceseet. From a termal control standpoint, te most significant advantage of incorporating a circulatory system into te face-seet of a satellite panel is modular, localized termal management using efficient, forced-convection fluid loops. Additional benefit can be obtained by controlling te flow witin te system. Tis can easily be accomplised wit mature, active valve systems, but te requirements for power input, sensing, and control-system ardware for active systems reduce te symbiotic nature of te approac. Instead, a passive-reactive termal valve provides a more symbiotic solution. PANEL DESIGN AND OPTIMIZATION Te primary focus of tis effort was to design and optimize te primary supply cannels and te sub-supply cannels to integrate tem into te ribs of a composite isogrid panel witout increasing te overall mass of te system. Te fabrication of te face-seet distribution cannels is a straigtforward process of macining te cannels into te face-seet and applying a top-seet to seal te cannels. Fig. 3 sows te basic geometric configuration of te panel, wic is approximately 0.09 m.
Front of Panel Front witout Face seet Back of Panel Face seet Primary supply cannels Sub supply cannels Termal Valve Ribs Figure 3: Front, front witout face-seet, and back views of te composite panel sowing only te primary supply cannels and sub-supply cannels Design Optimization Analysis Approac Because of te fabrication approac used for te prototype panel, te basic cannel geometry was set, wic is sown in Fig. 4. However, te size of te cannel could be adjusted by canging te radius of curvature of te brancing arms of rib structure. Te sub-supply cannel size was determined by conducting a parameter study tat examined te eat transfer, pressure drop, and face-seet deflection for various cannel sizes. Figure 4: Scematic view of te face-seet sub-supply cannel cross section Heat transfer: To minimize te pumping power requirements for te system, te system flow rate was cosen to ensure laminar flow under all conditions. An internal flow is considered laminar for Reynolds numbers, Re, in te range of Re < 300 and is determined using Eq. 1: ρud Re = (1) μ were ρ is te density in kg/m 3, u is te velocity in m/s, D is te ydraulic diameter in m, and μ is te dynamic viscosity in N-s/m. For non-circular cross sections, D is given by Eq. :
4A D = () p were A is te area of te cross section in m and p is te perimeter of te cross section. Te independent variable for te analysis was te radius of te cannel fillet, R, and was varied over te range.54 mm R 1.70 mm in te parameter study. Te resulting Reynolds number range was 55 < Re < 45. To model te eat transfer to te cannel, a constant eat flux boundary condition was assumed to correspond to a patc eater placed over te face-seet cannel area. In addition, fully developed flow for te entire lengt of te cannel was assumed because te entry lengt was on te order of.5 cm for te Reynolds number range of interest. It is also important to acknowledge tat secondary flow effects would be present witin te cannel intersection points. Te intersection points affected bot te pressure drop analysis and te termal analysis. For te pressure drop analysis, te flow troug te intersections was treated as flow troug eiter a tee or a 90 abrupt bend. For te termal analysis, te primary effect was on te flow development as te fluid re-enters te cannel. Again, fully developed flow was assumed for te entire lengt. To compare te different cross sections, te eat load and outlet fluid temperature was kept constant for all cases. Assuming a constant eat transfer coefficient and realizing tat te maximum fluid temperature occurs at te cannel outlet, te maximum cannel temperature was calculated based on te convection relation given by Eq. 3: q T w = + Tout (3) A w were q is te eat added to te flow in W, is te eat transfer coefficient in W/m - K, A w is te wetted area of te cannel, and T out is te outlet temperature of te fluid. T w in Eq. 3 represents te wall temperature directly adjacent to te fluid at te outlet of te face-seet cannel. For fully developed flow and a constant eat flux boundary condition, te fluid temperature rises linearly along te cannel, and te temperature difference between te wall and fluid remains constant. Te eat transfer coefficient,, was calculated based on te non-dimensional Nusselt number, Nu, for internal fully developed laminar flow given by Eq. 4: k = Nu (4) D were k is te termal conductivity of te fluid in W/m-K. Te cross section was assumed to be mostly triangular, and a Nu number of 3.11 was used 3. Assuming constant material properties and using te known value of T out and q, te mass flow rate in kg/s was calculated using te first law of termodynamics, given by Eq. 5: q m& = (5) C (T T ) p out were C p is te specific eat in J/kg-K and T in is te fluid temperature at te cannel inlet. Te mass flow rate was used to calculate te fluid velocity, wic in turn was used to compute te pressure loss of te panel. Assuming an incompressible flow and in
utilizing te conservation of mass principle, te fluid velocity was calculated in m/s according to Eq. 6: m& u = (6) ρa Pressure drop: To calculate te overall pressure loss of te panel, te fluid velocity was calculated for te sub-supply cannels. Because of te large cross-sectional area and corresponding low velocity, te losses were ignored in te primary supply cannels. Te pressure loss was computed based on bot major and minor losses. Te major losses accounted for te losses in te long straigt sections of te cannel. Te minor losses were used to account for te pressure loss in te bends and discontinuities of te cannel. Te major losses were calculated according to Eq. 7 4 : 1 L maj = fu (7) D were f is te dimensionless friction factor and L is te lengt of te cannel in m for wic te loss was being computed. Because te flow for all conditions in tis analysis was laminar, te friction factor was independent of surface condition and was computed according to Eq. 8 4 : 53 f = (8) Re Te minor losses due to abrupt discontinuities were calculated according to Eq. 9: 1 min = Ku (9) were K represents a non-dimensional loss factor. Te loss factor K was available in te literature based on experimental data 4. However, te losses in te serpentine faceseet cannel due to gradual bends were computed by Eq. 10: 1 L e min = fu (10) D were L e represents an equivalent lengt of cannel. Te L e /D ratios were also obtained from experimental data in te literature 4. Te overall pressure loss of te panel was calculated according to Eq. 11: Δ P = ρ ( min + maj) were ΔP loss is te overall pressure loss of te panel in Pa. loss (11) Face-seet deflection: Te primary structural requirement of te cannel is to resist deflection caused by te fluid pressure. Because te geometry of te curved walls naturally resist deflection, te primary concern was te deflection of te face-seet. Minimizing face-seet deflection was also important to provide a flus mounting surface for te components. To
calculate te face-seet deflection, te problem was simplified to a simple onedimensional beam bending problem wit uniform load and fixed support boundary conditions. Te maximum deflection at te center of te span was calculated wit Eq. 1 5. 4 wl y max = (1) 384EI were w is te load in N/m, l is te lengt of te span in m, E is modulus of elasticity in Pa, and I is te second moment of area in m 4. Design Optimization Results and Discussion To determine te optimized radius for te cannel geometry, eac of te design parameters was normalized according to Eq. 13 for easier comparison. x x x * value min = (13) x max x min Te results for eac parameter are presented in Fig. 5 for a eat input of 70 W and a flow rate of 50 ml/min. As sown, bot te pressure and termal performance of te panel improved wit increasing cross sectional area on a pat of diminising returns. Te trade-off was an increase in face-seet deflection as te span increased. Normalized Parameter 1 0.8 0.6 0.4 0. 0.5 4.5 6.5 8.5 10.5 1.5 Cannel radius, mm Temperature Pressure Loss Deflection Figure 5: Plot of te normalized temperature, pressure loss, and face-seet deflection of te termal-structural panel To summarize te performance analysis, it was desired to acieve te optimal performance based on te assumed cannel geometry. It is for tis reason tat a cannel radius of 7.6 mm was cosen. However, it is also wort noting tat te purpose of tis study was not to optimize in terms of eat transfer augmentation but to integrate cooling cannels witin a composite isogrid structural arcitecture. Te parameter study discussed in tis section was aimed at selecting te most appropriate cross section, wic can easily be prototyped.
VALVE DESIGN In addition to te panel design, an important aspect of te concept was te design of te passive-reactive termal valve. Te focus of tis design effort was to design for bot functionality and maintainability to facilitate experiments in te lab. Altoug not drawn to scale, Fig. 6 presents a scematic representation of te valve layout in te actuated, or open, position. Te valve utilizes te expansion properties of paraffin wax as it melts to open te valve and te spring force of Bellville wasers to close te valve as te paraffin solidifies. O rings Spacer Teflon Piston Belville Wasers Paraffin Inlet Outlet Heat Flux Figure 6: Scematic cross-section view of te passive-reactive valve Nonadecane (C 19 H 40 ) paraffin was cosen because its pase cange begins at 95 K, wic is consistent wit te normal operating temperature range for most satellite bus components. Nonadecane begins undergoing solid-solid transitions at 95 K and completes te solid-liquid transition at 304 K 6. During tis transitioning process, it expands approximately 11.8% by volume. Altoug not clear in te scematic view, te paraffin camber was cylindrical and limited motion to one axis. Tis resulted in an 11.8% linear actuation distance based on te overall eigt of te paraffin pellet. Te paraffin camber was sized to provide adequate flow troug te valve. Belville wasers, wic provide ig spring constants for small displacements, provided te return force to close te valve. Coil springs were examined, but te available spring constants were too low for te lengts required for tis application. As sown in Fig. 6, two wasers were stacked end to end to acieve te desired displacement and return force. Te spring camber was sized to create a preload on te piston to ensure proper sealing. Teflon was cosen as te piston material to resist sticking. For te overall paraffin pase cange process, te spring force, F s, was designed to vary witin te range of 71 F s 45 N. In designing te paraffin camber, two competing factors were considered. First, to minimize te valve actuation time, te pellet diameter must be small. If te pase cange time was too long, te component could overeat. Second, te cylinder diameter ad to be large enoug to accommodate off-te-self o-ring sizes, wic effectively set te minimum value for te paraffin camber diameter. Based on te latter aspect, te cylinder diameter was set at 5.8 mm. To verify te acceptability of te proposed diameter, a simplified pase-cange analysis was conducted to determine te melt time and resulting interface boundary temperature. Te system was modeled as a one-dimensional infinite solid to provide a conservative estimate. Te actual paraffin pellet will melt muc faster because eat is applied to all
sides of te cylinder as well as te base of te valve. Te melting process of te analysis is depicted in Fig. 7 and assumes te material is initially at te melt temperature, T m. As eat is applied to te boundary, te location of te solid-liquid interface, s(t), moves wit time and te temperature of te liquid rises. Tis process is stated matematically in Eqs. 14a-d 7 : T(x, t) 1 T(x, t) = x α t 0 < x < s(t),t > 0 (14a) T k = q 0 = x, t > 0 (14b) x T = x = s(t), t > 0 (14c) T m T ds(t) k = ρl x = s(t), t > 0 (14d) x dt were α, ρ, and L are te termal diffusivity, density, and latent eat of fusion in m /s, kg/m 3, and J/kg respectively for nonadecane. Note tat te temperature profile defined by tese equations is for te liquid zone. T q Liquid Solid T m 0 s(t) x Figure 7: Diagram of te melting process Equations 14a-d were solved by assuming a linear temperature profile and using te integral metod for approximating partial differential equations 7. Te temperature profile as a function of distance and time for a semi-infinite solid wit a constant surface eat flux condition is given by Eq. 15: q"(x s(t)) T(x, t) = Tm (15) k Te solid-liquid interface location is given by Eq. 16 7 : 1 Lρα s(t) = + 8αt q" 1/ ρl α q" (16)
For a typical component eat flux of 500 W/m, a melt time of 14 minutes was required. However, te more important parameter was te increase in surface temperature at te eat flux boundary. Te resulting temperature difference was 4 K. It is also important to note tat te melt time varies indirectly wit eat flux, but te temperature difference remains essentially constant. As a result, te valve responds in accordance wit te cooling needs of te system. Based on te simplified analysis, it appears tat te response of te valve is compatible wit typical spacecraft termal time constants. Future experimental measurements are planned to verify te numerical predictions and so validate tis symbiotic approac. CONCLUSIONS Te primary focus of te effort was to design a composite, biologically inspired structural panel tat integrated a circulatory, or pumped fluid system, into an isogrid structure to significantly improve cooling performance. Te fluid supply cannels, wic mimic arteries and veins, were integrated into te rib structure of te panel, and te fluid distribution cannels, wic mimic te capillaries witin te organs, were integrated into te face-seet of te panel. A prototype panel and a passive-reactive termal valve were designed, and an optimized cannel configuration was identified tat included termal, pumping, and fabrication issues. Te cannel configuration can best be summarized as an isosceles triangle were te equal lengt sides ave been replaced wit arcs wit a radius of 7.6 mm; tis arc lengt offered te best combination of termal, pressure loss, and structural performance. References 1. Williams, A. D., Arrit, B., Millan, D., Taft, B., and Nieuwdoop, A., Biologically Inspired Termal-Structural Satellite Panels, 48t AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, April 007. Bejan, A., Convection Heat Transfer, 3rd ed., Jon Wiley & Sons, New Jersey, 004, Capter 3. 3. Incropera, F. P., and DeWitt, D. P., Fundamentals of Heat and Mass Transfer, 4t ed., Jon Wiley & Sons, New York, 1996, Capters 5 and 8. 4. Fox, R. W., and McDonald, A. T., Introduction to Fluid Mecanics, 5t ed., Jon Wiley & Sons, New York, 1998, Capter 8. 5. Sigley J. E., and Miscke C. R., Mecanical Engineering Design, 6 t ed., McGraw- Hill, New York, 001, pg 1186. 6. Huang, D., Sindee, L. S., and Mckenna, G. B., Cain Lengt Dependence of te Termodynamic Properties of Linear and Cyclic Alkanes and Polymers, Te Journal of Cemical Pysics [online journal], Vol 1, No. 084907, Paper 1, URL: ttp://link.aip.org/link/?jcpsa6/1/084907/1 7. Özişik, M. N., Heat Conduction, nd ed., Jon Wiley & Sons, New York, 1993, Capter 11.