15th International Heat Pipe Conference (15th IHPC) Clemson, USA, April 25-30, 2010
DEVELOPMENT OF AN AXIAL GROOVED HEAT PIPE WITH LOW COEFFICIENT OF THERMAL EXPANSION Reinhard Schlitt, Frank Bodendieck OHB System AG Universitätsallee 27-29, 28359 Bremen, Germany T: 0049 421 2020 637, F: 0049 421 2020 900,
[email protected] Christoph Hoffmeister Faserinstitut Bremen e.V. Am Biologischen Garten 2, 28359 Bremen ABSTRACT Due to the excellent stiffness to weight ratio CFRP structures become very popular in aerospace products. Spacecraft structures are generally designed as sandwich panels with CFRP face sheets and aluminium or nonmetallic honeycomb core material. Exceptions are so-called thermal radiators, which are spacecraft panels with embedded or surface mounted heat pipes. These thermal radiators are again sandwich panels, but the face sheets are from an aluminium alloy to avoid any mismatch in the coefficient of thermal expansion (CTE). A low-CTE heat pipe has recently been developed, which overcomes the problem of CTE mismatch of metallic tubing and CFRP face sheets in spacecraft radiators. The design principle features existing aluminium heat pipe profiles, which are wrapped length-wise with high-strength carbon fibres. During an increase of operating temperatures the fibre bundle hinders the metallic heat pipe profile to expand and reduces consequently the thermal expansion of the unit. During testing of heat pipe / carbon fiber prototypes a combined CTE of 6 ppm has been reached, which compares to 23 ppm for the pure metallic piece. The resulting CTE is adaquate for bonding the low-CTE heat pipes to standard CFRP face sheets of radiator panels. Failure tolerance against buckling, fatigue and creep has been verified by analysis. KEY WORDS
Heat Pipe, Carbon Fiber, CTE, Thermal Radiator, Spacecraft
1. INTRODUCTION CFRP sandwich panels have been introduced into spacecraft designs due to their mass saving potential compared to aluminum alloy configurations. Up to date heat pipe radiator panels are generally still from aluminum, since structural problems connected to the large CTE mismatch between heat pipe aluminum profiles and CFRP sandwich face sheets could not be solved. Although the development of low-CTE heat pipes for integration into CFRP panels has been an issue in the last 10 years, reliable solutions for general introduction into spacecraft designs are not available. Mastering low-CTE heat pipe / CFRP radiator technology will not only reduce radiator panel weight (due to the low-density CFRP face sheets). The technology will in addition increase spacecraft pointing accuracy and reduce system mass, since
means to minimize thermo-elastic deformation in a spacecraft with mixed material panels are not necessary (for example separate, decoupled antenna modules). The objective of a in-house develoment program is therefore to introduce an innovative solution, which enables the use of today’s aluminum heat pipe profiles for integration into CFRP sandwich panels. 2. PROBLEM STATEMENT Heat pipes, which are presently used worldwide in virtually all spacecraft, consists of aluminium alloy extrusions with internal axial grooves, externally mounted flanges and ammonia as heat carrier. The drive to use CFRP honeycomb sandwich structures also for heat pipe radiator panels confronts the spacecraft designer with the problem to provide concepts, which solve the large mismatch of the
involved coefficients of thermal expansion (CTE). The CTE of aluminium is about 23 E10^-6 1/K (or 23 ppm) compared to about –1.5 ppm for a typical carbon fibre, i.e. for a related CFRP face sheet. When bonding the two materials with an adhesive, the shear stresses in such a joint are much higher then allowable values for currently known adhesives. In Figure 1 calculated shear and peel stresses in an adhesive bond are compared with allowable values for two adhesives. According to the diagrams the peel stresses are generally not a problem, but the maximum calculated shear stresses cannot be accommodated with such an adhesive joint. This is especially true for higher operating temperatures of the panel.
The first of above-mentioned solutions has the main advantage that existing ammonia / aluminium heat pipes can be used without expensive thermal re- or delta-qualification. The mentioned second solution requires such an expensive, risky and long-duration development program. Due to the missing heritage of any newmaterial heat pipe the acceptance by satellite programs is low and it would be difficult to introduce such a product in running spacecraft projects (especially for communications satellites). The current development effort stresses therefore possible concepts to solve the CTE-mismatch problem with the mentioned first solution, i.e. provide a technology that reduces the CTE of known aluminium / ammonia heat pipes. Such an innovative design is presented in Chapter 4. 3. REVIEW OF CONCEPTS / STATE-OFTHE-ART 3.1 Existing concepts
Figure 1. Comparisons between calculated and allowable shear and peel stresses. There are basically two solutions to solve the problem: 1) Continue to use the known aluminium heat pipes, but modify the profile and heat pipe / radiator integration technique as to master the CTE mismatch 2) Develop a new heat pipe type with a lowCTE material, which would meet the requirements of the low thermal expansion of the CFRP face sheets. Both solutions have been investigated in the past by several engineering companies and are further explained in Chapter 3.1.
The desire to have low-CTE heat pipes to be integrated into CFRP panels is present since 15 to 20 years. The drive to work in that direction is motivated to secure the potential mass saving and to reduce thermo-elastic deformation of CFRP panels when compared to equal performing all aluminium panels. The mass saving is not only due to the increase in stiffness-to-mass performance of carbon fibre structures, but also due to the potential to use lay out configurations with lower heat pipe distance, which is possible, when using thermally high conductive carbon fibres in the face sheets. There are high modulus coal tar pitch-based carbon fibres available with a longitudinal thermal conductivity of > 600 W/mK. For a CFRP, which is a carbon / resin matrix, the resulting thermal conductivity is still > 350 W/mK. This is considerably higher than aluminium alloys with about 150 W/mK, which are currently used for sandwich face sheets. Relevant CFRP / heat pipe development activities have been carried out to a larger extend in Europe, Japan and USA. Also some relevant Russian developments have been reported but not published. The European activities go back to an ESA contract with ERNO Raumfahrtechnik in 1993 (Müller, 1993). The study comprised a CFRP / aluminium honeycomb panel with one surface
mounted heat pipe. The mismatch in CTE was mastered by a use of a highly deformable adhesive as interface filler between heat pipe and face sheet, which is able to compensate the resulting stresses by its elongation capability. During panel testing the interface bonding between heat pipe and face sheet failed at +80°C. The failure was due to a delamination of the bond line between adhesive and heat pipe. The work concluded that it is difficult to compensate the CTE mismatch by applying flexible interface filler between aluminium heat pipe and CFR face sheet. The investigated solution was furthermore only tested on surface-mounted heat pipes and can possibly not be applied to embedded heat pipes. Japanese activities were mainly carried out by the Advanced Technology R&D Centre of the Mitsubishi Electric Corporation. Figure 2 shows a completed CFRP radiator with embedded heat pipes.
find a heat pipe above the Ammonia temperature range. Titanium and Monel with water as heat carrier was finally selected. The heat pipes are designed for integration into a CFRP sandwich structure. Both Ti and Monel have lower CTE than aluminium but still higher than carbon fibre. There has been three different heat manufacturing techniques / wick material concepts investigated (Figure 3).
Felt Wick
Screen Axial Arteries
Figure 2: CFRP panel with embedded aluminum heat pipes
Axial Grooves Figure 3: Different Investigated Ti-Heat Pipes (Anderson, 2006)
The work is published in several conference papers (Ozaki, 1999, 2000, 2005, 2007). The method of embedding heat pipes is furthermore protected by several patents (JP 2000129857A, JP 2001153576A, JP 2004-019990, and others). Most of the referenced patents mention a flexible adhesive between heat pipe and face sheet, which would mean that the process is similar to ERNO / ESA study (Müller, 1993) and would be prone to the same failure mechanism. The Patent JP 2001153576A describes that the heat pipe surface may be configured with a step at the ends with highest relative movement for the parts in order to increase the adhesive thickness in the step zone (or to decrease the adhesive layer in the remaining zone). Increase in adhesive bond line would decrease the shear stresses in the adhesive and a decrease in adhesive bond line in the remainder zone would increase heat transfer across the interface to the heat pipes. The predominant US American work is published in Anderson (2006). The aim of the development is to
From the figure the difficult effort to establish wick structures into a titanium alloy tube can be imagined. The extrusion process, as with Aluminium alloy, is here not possible. The disadvantages of the solution presented in Figure 3 are the following: The felt-wick is a difficult approach for providing reproducible manufacturing processes for the wick. In addition the bonding of the wick to the pipe wall, i.e. the thermal conductivity in this interface, is problematic. The artery wick concept brings back the uncertainties as experienced in similar wick developments some 20 years ago (vapour entrapment in the artery, start-up problems, etc.). Artery wick designs have been practically abandoned for many years. The approach to machine axial grooves into a flat sheet and roll it up to a cylinder, which is then longitudinally welded, is an interesting concept, which has been developed by the author of this paper already in the beginning of
the 1970ties. The solution is, however expensive and does not provide integrated flanges to mount the pipe to flat surfaces. Provision of flanges, which are fabricated in aluminium with the mentioned extrusion process, is generally very difficult when using Titanium or similar high-melting materials. In order to find a solution the pipes are proposed for embedding into the sandwich panel with the help of preformed highconductivity carbon foam (see Figure 4).
Figure 4: Embedding of Ti - heat pipes with carbon foam into sandwich panel (Jaworske et al., 2007) At NASA heat pipes for even higher temperatures where developed, which have Potassium as working fluid and a thin-walled metal liner within a carboncarbon tube re-enforcement as heat pipe wall (Juhasz, 2008). The condenser section is equipped with an integrated carbon fin (see Figure 5).
flanges can be provided. In addition the technique needed to avoid undue shear stresses between liner and carbon re-enforcement is not explained. In case the bond between the two parts can handle these stresses the thin liner will easily undergo plastic deformation. The German patent DE 41 30976 A1 describes different combination of aluminium profile and attached carbon fibres (see Figure 6). The inventor claims that by attaching carbon fibre to the outside of the profile or in pockets or in hollow structures the CTE of the combination is reduced. The invention fails to explain, how the compressive forces of the aluminium are transferred into the fibres. It is therefore questionable, whether the concept will meet the requirements.
Legend: 1 Pipe (thin aluminium) 3 Longitudinal channels 5 Longitudinal wraps (carbon fibre) 7 Hoop wraps (carbon Fibre) Figure 6: Thin Aluminium Pipe wrapped with Carbon Fibre in two directions US 6184,578 B1
1-aluminium, 2-carbon fibre, 10-heat pipe grooves) Figure 7: Aluminium / fibre combination (Patent DE 41 30976 A1) Figure 5: Carbon-Carbon heat pipe with metallic liner (Juhasz, 2008) A similar idea is described in the patent US 6184,578 B1, where a thin Aluminium tube (liner) is wrapped with carbon fibre in two directions (15° / 90°). The idea behind this approach (see Figure 6) is probably that the carbon package hinders the aluminium liner to thermally expand. The disadvantage of the method is that neither integrated axial groves nor integrated
Another US patent describes an additional method to solve the CTE problem. In patent US 6,065,529 a heat pipe profile is proposed, which is encased by a high thermal conductivity liquid gap. The liquid would then enable the aluminium pipe to freely move with respect to the CFRP radiator and the transfer of heat from the pipe to the radiator face sheets is accomplished due to the liquid’s high conductivity.
4. NEW INNOVATIVE CONCEPT 4.1 Concept description During an in-house development program we analyzed and tested an innovative heat pipe concept, which probably meets all the requirements of a lowCTE heat pipe for integrating into a CFRP radiator panel. The approach is based on the well-known state-offthe-art aluminum extrusion heat pipe (Figure 7). Such a heat pipe consists of the actual extrusion profile with a double-flanged configuration, as example. The concept can be applied for any extruded aluminum profile. The core profile may be of circular or rectangular shape and the flanges may be individually machined to suit the specific application. Note: Dual core profiles, as used frequently for embedding heat pipes into sandwich structures are not shown, but can be treated as explained for the double-flanged profiles in Figure 7. The standard heat pipes includes in addition end-caps and a fill tube to charge the pipe, usually with ammonia.
Figure 7 Classical extruded heat pipe profile In order to reduce the CTE of such a heat pipe, we propose to wrap the profile length-wise with carbon fibers. When the aluminum profile is heated up, the high-strength fibers will largely avoid the thermal expansion of the aluminum. In fact, in such a compound the aluminum will be exposed to compression and the fibers to tension loads. Under temperature cycles both the compression and tension loads will change in the same cyclic manner. The carbon fiber / aluminum profile configuration is sketched in Figure 8.
5 5 6 1 3
4
6 6
1a
1b
6 Figure 8: Low-CTE Aluminum / Carbon fiber heat pipe
1c
6
1d
Besides the basic elements of a classical aluminum heat pipe (1, 2, 3, 4) the following features are added: half-circular return part with side-guidance for the carbon bundle (5) carbon fiber bundle (6) The carbon fiber bundle is easily placed between the flanges of the profile (1a, 1b, 1c, 1d of Figure 8). Fibers are wrapped in dry conditions, i.e. the fiber bundle is not saturated with resin. The dominant advantages of the concept are the following: All existing axial grooved aluminum profiles can be used. Thermal performance and material compatibility are known and unchanged from existing heat pipes used in spacecraft. Thermal re-qualification or delta qualification is not necessary. The qualification effort is concentrated on the mechanical performance of the heat pipe / carbon fiber compound and the interface to the CFRP face sheets. It is therefore possible to use the excellent flight heritage of aluminium / ammonia heat pipes for this innovative low-CTE product. Considering a typical aluminum alloy of axial grooved heat pipes (for example 6063 T5) the resulting CTE of the heat pipe compound depends mainly on three characteristics: Youngs modulus of carbon fiber (higher = better) Cross section of aluminum (smaller = better) Cross section of carbon bundle (higher = better) The relationship is given in Figure 9. The diagram shows the resulting CTE of the heat pipe compound over the Youngs modulus of the carbon fiber and the total weight of the specific compound (g/m). The aluminum profile weight with 405 g/m has been assumed constant and the difference to the given total weight is then the weight portion of the carbon bundle. NOTE: The profile weight of 405 g/m is quite high and has been selected as upper worst case. The diagram in Figure 9 reads for example: To obtain a desired CTE of 5 ppm with a fibre Young's modulus of 700 Gpa the total weight of the heat pipe compound would be around 477 g/m or the carbon bundle portion (477 – 405 =) 72 g/m.
Figure 9 Dependence of heat pipe compound CTE from fiber Youngs modulus and carbon bundle weight. 3.2 Low-CTE heat pipe analysis During the development program structural analysis of the compound was performed in order to determine a configuration, which would not violate allowable mechanical properties of the involved materials. Calculations show a strong dependence of the mechanical fiber properties on the resulting CTE of the compound. We selected therefore four different fibers: T800 (Toray), CN-60, CN-80 and CN-90 (Granoc). The second largest influence has the cross section of the heat pipe profile. As smaller the cross-section as smaller the force needed to compress the profile, which results in a corresponding small cross-section of carbon fibre bundle. In a worst case approach the full cross-section (including flanges) has been baselined. In many applications the flanges can be (partly) milled away. Furthermore the wall thickness of the baselined profile is designed for proof / burst pressure factors of 2.5 and 4, which realistically can be reduced to factors of 1.5 and 2.5 at least for surface mounted heat pipes (the factors for
embedded heat pipes depend on the adhesive curing temperature during panel bonding). Based on the properties of the selected materials we calculated a resulting compound CTE between 12.5 ppm using T800 and 5.9 ppm using CN-90 (see Figure 10). The calculated von-Mieses stresses in the aluminium are much lower than the allowable values (see Figure 11). In this calculation we assumed an operating temperature range of the heat pipes of Tmax – Tmin = 100K. The allowable stress values for carbon fibres are much higher than for aluminium alloy and are therefore not critical. It shall be noted that in the described configuration the compression forces in the profile are equal to the tension forces in the carbon bundle.
Table 1 Stress, CTE and specific mass results for variation of profile and fiber bundle cross section Cross Thickness Von CTE of Mass of Section Fibre Mieses Compound Fibre Profile Bundle Stress with CN- Bundle (mm^2) (mm) 3 (Mpa) 90 Fibre (g/m) (ppm)
Specific Total Mass of Heat Profile Pipe (g/m) Weight (g/m)
140 1 1.5 140 5.9 29.8 378 407.8 70 2 1.0 110 6.1 20.1 189 209.1 70 2 1.5 130 4.6 28.2 189 217.2 70 2 2.4 140 3.0 42.3 189 231.3 1 Profile with Proof / Burst Factor 2,5 / 4,0 and flanges of 30 mm width 2 Profile for Proof / Burst Factor 1,5 / 2,5 and partly removed flanges 3 CN-90 fibre bundle thickness are selected to obtain vonMieses stresses in Aluminium between 100 and 150 Mpa
3.3 CFRP Radiator Analysis
Figure 10 Combined CTE compared to the CTE of the involved material (dark red: aluminium, violet: carbon fiber, yellow: profile /fiber compound)
With above analysis results we verified the mechanical behaviour of the low-CTE heat pipe, when bonded into a CFRP panel. For this purpose we assumed: An aluminium -honeycomb of 3/16” cell width A CFRP face sheet from T800 and 0/45/45/0 lay-up (0 = direction of heat pipe) Adhesive 3M Scotch-Weld 2216 and alternatively Hexcel Redux 312 UL between embedded heat pipe and face sheet The analysis results are plotted for Redux in Figure 12. Max. Shear Stress in Adhesive 'HEXCEL Redux 312UL ' 45 40 35 30 τs [MPa]
HEXCEL Redux 312UL (22¡C)
25
HEXCEL Redux 312UL (70¡C)
20
max. Shear Stress
15 10 5 0 T800
CN-60
CN-80
CN-90
Profile Reinforcement
Figure 11 Von-Mieses stress in the aluminum profile during maximum compression (red) and comparison to allowable values (blue). In order to analyze the influence of the cross section for heat pipe profile and fibre bundle we performed calculations where these parameters have been varied. The results are summarized in Table 1.
green and light green: allowable stress at 22 ad 70°C, pink: calculated shear stress
Figure 12 Maximum shear stress in the adhesive Redux 312 UL for the examined fiber bundles The calculated shear stresses in the adhesive bond are compliant in all cases to allowable values. The
values for the T800 fiber bundle (resulting CTE = 12.5 ppm) have the lowest margin. The maximum calculated peel stresses in the adhesive are in all cases very low and therefore not critical. Further mechanical analysis showed positive margins for fatigue of the aluminium profile due to cyclic compression, for creeping of the aluminium profile due to permanent compression load and buckling of the compound.
There are several methods to measure this coefficient. Most processes measure the change in length during warm up of the sample. Known processes are optical, laser, interferometer, dilatometer and others. Our partner, the Fibre Institute Bremen, has specialized to measure the CTE with strain gauges.
3.4 Prototype manufacturing and testing Several short, straight prototypes were manufactured and tested. This hardware related work was carried out in cooperation with the Fibre Institute Bremen, a department of the University Bremen. All tests were carried out with empty profiles (without heat carrier fluid), but end caps and filling tube are included. Both profile ends were equipped with half-circular return parts for the carbon bundle. The return part opposite to the fill tube end is designed to comprise a tightening bolt in order to reach a necessary pre-tension of the fibre bundle. Carbon fibers are guided around the bolt and the filling tube as indicated in Figure 13.
Figure 13 Carbon Fibre Return Part with Tightening Bolt It is important that the carbon fibre bundle is tightened at the lowest required temperature in order to ensure that the aluminium profile will be under compression within the entire specified temperature range. While it was easier for the university institute to follow above described process, we will wrap in future the fibre at the lowest specified temperature in order to avoid the tightening bolt. The completed prototypes are shown in Figure 14. After completion of the profile preparation the resulting CTE of the compound was measured.
Figure 14 Low-CTE heat pipe prototype Table 2 Comparison between predicted and measured CTE Test Object Prototype 2 Prototype 3/4
Fibre Type and EModulus T800 HD 294 GPa UMS S45 430 GPa
Alu Cross Section (mm^2) 190
Fibre Cross Section (mm^2) 40.7
CTE (predicted) (ppm)
CTE (measured) (ppm)
12.56
15.77
149
53.4
6.065
5.83
The results of the CTE measurements are summarized in Table 2. After test with prototype 2 the CTE measurement technique was considerably improved and gives now very good results (Table 2). The test results show the relative influence of the material parameters on a resulting low CTE, i.e. to obtain a low CTE the material parameters must be as follows: High E-modulus of fiber Low cross section of aluminum profile High cross section of carbon fibre bundle 3. CONCLUSIONS The developed low-CTE heat pipe consists of readily available aluminum profiles, which are wrapped lengthwise with high strength carbon fibers. The design features the following advantages: Existing aluminum heat pipes can be used without thermal delta qualification. The heat pipe uses the integrated flanges of the aluminum profile as interface to spacecraft structures.
The resulting CTE of the carbon fiber / heat pipe compound can be tailored to fit different application by selecting an optimized relationship between heat pipe and fiber bundle cross section. The low-CTE heat pipe is suited for radiators with embedded and surfacemounted heat pipes configurations. All mechanical requirements (fatigue, creep, buckling) have been verified with margins. According to current planning the low-CTE heat pipes will be introduced for the first time into an ongoing German telecommunications satellite program. The concept of the presented low-CTE heat pipe has been protected under a German patent. NOMENCLATURE CFRP CTE ppm
Carbon fiber reinforced plastic Coefficient of thermal expansion Parts per million
REFERENCES 1. Müller, R. et al., Design and Test of a Honeycomb Radiator Panel with Carbon Fibre Reinforced Plastic Facesheets and Aluminium Heat Pipes, 1993, SAE Paper 932302 2. T. Ozaki et al., Graphite Faceskin Heat Pipe Embedded Honeycomb Sandwich Panels for ETS-VII Structures, IAF Conference, Amsterdam, 1999 3. T. Ozaki et al., Graphite Facescin Deployable Radiator Panels for ETS-VII Structures, IAF, Rio de Janero, Oct. 2000 4. T. Ozaki and H. Ishii, Development of Low CTE CFRP Faceskin Honeycomb Sandwich Panel with Embedded Heat Pipes, SAMPE 2005, Long Beach, Ca, May 2005 5. T. Ozaki, Advanced Composite Parts and Structures with CFRP Face Sheets and Aluminum Heat Pipes, SAMPE 2007 6. W. G. Anderson et al., Hight Temperature Titanium-Water and Monel-Water Heat Pipes, Int. Energy Conversion Eng. Conf. AIAA, San Diego, Ca, 2006, pp. 828-839 7. D. A. Jaworske et al., Heat Rejection Systems Utilizing Composites and Heat Pipes: Design and Performance Testing, AIOAA 2007-4822
8. A. J. Juhasz, High Conductivity Carbon-Carbon Heat Pipes for Light-Weight Space Power System Radiators, AIAA 2008-5784
