Fuel-Powered Compact SMA Actuator

Fuel-Powered Compact SMA Actuator Othon K. Redinioti a, Dimitri C. Lagouda a, Hyoung Y. Jun a, Richard D. Allen a Aeropace Engineering Department Texa A&M Univerity, College Station, TX 77843-3141 ABSTRACT Thi work dicue the numerical analyi, the deign and experimental tet o the uel-powered compact SMA actuator along with it capabilitie and limitation. Convection heating and cooling uing water actuate the SMA element o the actuator. The energy o uel, having a high energy denity, i ued a the energy ource or the SMA actuator in order to increae power and energy denity o the ytem, and thu in order to obviate the need or electrical power upplie uch a batterie. The ytem i compoed o pump, valve, bellow, heater (burner), control unit and a diplacement ampliication device. The experimental tet o the irt deigned SMA actuator ytem reult in 150 M Pa tre (orce: 1560N) with 3 % train and 0.5 Hz actuation requency. The actuation requency i compared with the prediction obtained rom numerical analyi. For the irt deigned uel-powered SMA actuator ytem, the reult o numerical analyi were utilized in determining deign parameter and operating condition. Keyword: Shape Memory Alloy (SMA), actuator, convection heat traner, eiciency, uel, power denity, energy denity. 1. INTRODUCTION The main objective o thi reearch i to deign, abricate and tet a highly compact hape memory alloy baed actuator that utilize the high energy denity o uel, uch a propane. The act that the main element o the actuator, the SMA i a heat engine [1][2][3], i ued to convert the thermal energy o a uel (propane) to mechanical energy. The high energy denity o uel compared to typical electrical batterie, or even uel cell, allow or the energy ource, i.e. the uel, to be incorporated inide the actuator ytem. The energy denity (J/kg) o thee uel i 100 time greater than that o mot advanced batterie. Thi, along with the incorporation o the actuation control hardware and otware inide the unit, can reult in a highly compact actuator. Thu the actuator ytem can be run wirelely by low-power, digital, actuator control ignal. The high-energy denity, high recovery tre and train o SMA will reult in high actuator compactne, orce and troke, repectively. The phae change in a NiTi SMA i achieved by heat exchange with a heat ource and a heat ink. The actuation requency o the SMA actuator i only dependent on the rate o heat traner with it urrounding. Until recently, the heat traner mechanim or mot SMA actuator ha been baed on reitive heating (M A) and cooling with orced convection or natural convection (M A). Thi i a rather ineicient heat exchange mechanim [4] and require the ue o electrical power and thu heavy, low-enegy-denity (at leat a compared to uel) power upplie or batterie. The thermoelectric heat traner mechanim by utilizing emiconductor, which employing the Peltier eect, ha hown high actuation requency [5]. But generally thi kind o device ha very low eiciency. Thu, we propoe orced convection heating and cooling to actuate the SMA actuator. Thi can overcome the low energy denity reitive heating ytem and the low eiciency o the thermoelectric heat traner mechanim, even though it hould need additional device uch a a pump and valve. Alo, the high energy o uel i tranerred eaily to the luid through the combutor and the heat exchanger. Reearcher have worked on developing high eiciency, compact combutor and heat exchanger, uing micro technology [6][7]. Convection heating and cooling o the SMA, can reult in coniderable actuation requencie. In addition, or ytem with uicient paraitic heat, the actuator can utilize the paraitic heat a it energy ource, reulting in a relatively high-eiciency actuating ytem. The actuator deign merge the advantage o SMA and uel, i.e., the high actuation orce, the large power denitie and the ilent actuation characteritic o SMA and the a Further Author inormation (Send correpondence to O.K.R) O.K.R: email: redinioti@aero.tamu.edu, D.C.L: email:lagouda@aero.tamu.edu, H.Y.J: email: hyoungyoll@aero.tamu.edu, R.D.A: email : minitman@aero.tamu.edu

tremendou energy denitie o uel. In thi paper we will dicu the deign o the SMA actuator ytem, recovery tre and train and actuation requency o the SMA actuator by utilizing the high thermal energy o uel. Section 2 o thi paper preent the principle o the uel-powered compact SMA actuator ytem, comparion with other actuator ytem and thermal & Carnot eiciencie o the SMA element o the actuator. Section 3 preent the numerical heat traner analyi o the SMA actuator. Section 4 i or the irt deign o the uel-powered SMA actuator ytem and it component. Section 5 preent experimental reult and comparion with numerical reult and dicue the ytem capabilitie and limitation. 2. DESIGN CONCEPT AND EFFICIENCY OF THE SMA ACTUTATOR SYSTEM 2.1 Energy Denity and Power Denity The ollowing data decribe the deign parameter o the SMA element o actuator ytem o igure 1. Thi data can be modiied accordingly to meet dierent requirement or the actuator. A NiTi SMA trip with a rectangular cro ection meauring about 12 mm x 1 mm (12 mm 2 cro ectional area) wa elected a the SMA element to increae heat traner rate compared to a wire having the ame cro ection area. Four uch trip can be intalled in a rectangular channel with 12 mm x 16 mm cro ection. Our work, [8][9][10], during the lat everal year ha lead to precie actuation control technique or NiTi SMA at tre level o 200 M Pa and actuation train o 3 %. Thee number are ued here in our deign. In order or the SMA trip to produce bi-directional load it will have to be pre-treed. Thi could be achieved a eaily a via a tre-biaing pring. I, or example, a tre bia level o 100M Pa i choen, the SMA trip will be able to produce bi-directional actuation load at the level o ±(100 M pa) x (trip cro ectional area), or, ±1,200 N (or 2,400 N uni-directional). An arrangement o our uch trip in an array will yield a combined bidirectional actuation orce o ±4,800 N. For a SMA trip length o 254 mm the SMA actuation troke (at 3 % train) i 7.62 mm (0.76 cm). The irt ytem in igure 1 reerred to a the SMA-Combutor ytem, comprie o two pump, a combutor, the SMA element, the cooling circuit heat exchanger and uel a energy ource. The SMA trip are embedded in a channel. Heating and cooling luid medium alternatively circulate through the channel to achieve the M-to-A and A-to-M tranormation, repectively. The heating medium (Ethylene Glycol or water) i heated through the burning o the uel in the combutor. The cooling medium, ater it remove the heat rom the SMA goe through a heat exchanger where it dipoe o the energy getting rom the SMA trip. The two pump that circulate the two media are equipped with valve that are properly timed through the heating and cooling cycle. SMA Heating circuit Compact Glycol Boiler Pump Combutor SMA Actuator P SMA Heating circuit Battery SMA Actuator SMA Heating circuit Fuel(methanol) Fuel Proceor O 2 H 2 FuelCell Stack SMA Actuator P Coolant Heat Exchanger SMA Cooling circuit P Pump Coolant Heat Exchanger SMA Cooling circuit P Pump Coolant Heat Exchanger SMA Cooling circuit Figure 1. Compact Actuator Sytem with uel+combutor, battery and uelcell a energy ource.

Alternative energy ource like batterie and uel cell have been conidered a energy ource or the SMA actuator in the other two ytem o igure 1. The reult o the comparion have been tabulated in table 1 and 2 or dierent cae o actuation cycle number (1000, 10000, 100000 actuation cycle). Thee table how that the SMA- Combutor compact actuator ha high energy and power denitie compared to battery and uel cell powered ytem. The uel cell tack itel ha high energy and power denitie, but need additional equipment uch a a reormer or a hydrogen tank, a cleanup unit, a bower, a compreor and cooling device [11][12], epecially or a large number o actuation cycle. The main diadvantage o the uel cell ytem i the large ma o the uel proceing unit or the ma o the uel tank. All thee actor combined give a low energy/power denity or the SMA-Fuel Cell actuator ytem. In the cae o a battery powered SMA (SMA-Battery) actuator ytem, the ma o the battery increae igniicantly with the increae in number o actuation cycle even having the highet eiciency: thi, coupled with the act that batterie have low energy denitie compared to uel cell or uel [13][14], yield the lowet overall energy/power denitie or thi ytem. Table 1. Output Energy Denity (Total Mechanical Work/Ma o Sytem). Actuator Sytem 1,000 Cycle 10,000 Cycle 100,000 Cycle SMA-Combutor 9.887(Wh/kg) 71.04(Wh/kg) 193.6(Wh/kg) SMA-Battery 3.717(Wh/kg) 4.411(Wh/kg) 4.495(Wh/kg) SMA-Fuel Cell 0.7605(Wh/kg) 7.038(Wh/kg) 41.1(Wh/kg) Table 2. Output Power Denity (Total Mechanical Work/(Ma o Sytem Cycle)). Actuator Sytem 1,000 Cycle 10,000 Cycle 100,000 Cycle SMA-Combutor 35.59(W/kg) 25.57(W/kg) 6.971(W/kg) SMA-Battery 13.38(W/kg) 1.588(W/kg) 0.1618(W/kg) SMA-Fuel Cell 2.738(W/kg) 2.534(W/kg) 1.48(W/kg) Table 3. Eiciency o the ytem Actuator Sytem 1,000 Cycle 10,000 Cycle 100,000 Cycle SMA-Combutor 2.253(%) 2.318(%) 2.325(%) SMA-Battery 2.999(%) 3.003(%) 3.003(%) SMA-Fuel Cell 2.499(%) 2.502(%) 2.503(%) 2.2 Eiciency In the deign and development o SMA actuator, the available thermal eiciency and their limit mut be calculated and conidered. Lagouda and Bhattcharyya [15] evaluated the eiciency o the thermoelectric SMA actuator. Jardine[16] and Gil [17] propoed the calorimetric technique and mechanical tet to evaluate the eiciency o hape memory alloy baed on ideal hape memory eect (SME) heat engine cycle[18]. Tranormation temperature o an SMA trip (DSC tet) The phae tranormation temperature and latent heat were irt determined in order to calculate the thermal eiciency o SMA trip. The trip utilized in thi tudy wa provided by Memory Corporation and i a Copper 10 % NiTi alloy. A Perkin-Elmer Pyri 1 Dierential Scanning Calorimeter (DSC) wa ued to determine the phae tranormation temperature and heat o tranormation. H i meaured a the total area under the curve during the heating and cooling cycle. The cooling cycle wa tarted ater holding or 1.0 minute at 120 C. The SMA wa then cooled rom 120 C to 60 C at 5 C/min. The heating cycle began ater holding the SMA or 1.0 minute at 60 C. The SMA wa then heated rom 60 C to 120 C. Figure 2 how the DSC tet reult o the SMA trip.

Figure 2. Tranormation temperature and latent heat (DSC Tet). Carnot Eiciency The proportion o the Carnot eiciency that can be ueully realized in practical engine i typically 60 %. Once the deign o an engine ha been optimized, the only way to get higher perormance i by increaing the Carnot eiciency or by extenion o the dierence in temperature o ource o heat between which the engine operate. The Carnot eiciency o SMA heat engine i [15][19] A ( σ max) M ( σ 0) ηcar = (1) A ( σ ) o da A ( σ ) = A + σ (2) dσ Where da i the lope in tre-temperature diagram o the SMA trip. Hence, dσ M ( σ 0) ηcar = 1 (3) o A ( σ ) A + σ max dσ o Theoretically, M ( σ 0) at zero tre and M hould be the ame value [17]. Hence the Carnot eiciency i like that, o M ηcar = 1 =15.3 % (4) o A ( σ ) A + σ max dσ Subtituting into equation (4) the value o the tranormation temperature getting rom the DSC tet, da / dσ =1/6.7 K/M Pa [20] and 200 M Pa tre. Then the Carnot eiciency i 15.3 %. The energy eiciency i theoretically retricted by the Carnot eiciency ince an SMA actuator i a heat engine operating at low temperature. Thermal Eiciency o Ideal SMA Heat Engine The thermal eiciencie o SME heat engine, which are o interet, have been etimated rom thermodynamic magnitude (enthalpie) and the tranormation temperature o the alloy [16]. The thermal eiciency or the ideal hape memory heat engine cycle i given [16][17] a H To ηth = (5) To ( Cp To + H ( σ )) Where Cp i the peciic heat o the material and To = To(σ ) To. Alo H (latent heat) in the denominator o the eiciency equation i a unction o tre, max

To i etimated a To calculate To (σ ), we take H ( σ ) = H To( σ ) / To (6) o o o o To = 1/ 2 ( M + A ), or,1/ 2 ( M + A ) (7) To ( σ ) = 1/ 2 [ M ( σ ) + A ( σ )] o o [ M + ( dm / dσ ) σ + A + ( da / σ σ ] = 1/ 2 d ) = To + ( dm / dσ ) σ (8) The thermal eiciency can be obtain by utilizing DSC tet data and dσ / dt =6.7 M Pa [20] o the K-alloy. d σ / dt = dσ / dm = dσ / da i aumed to get To. Alo the value o σ i aumed a 200 M Pa. Hence the thermal eiciency or the ideal hape memory heat engine cycle can be obtained rom equation (5) a, H To ηth = = 3.12 % To ( Cp To + H ( σ )) Where the value o Cp i 550 J/kg K [17]. Gil and Planell [17] calculated the thermal eiciency o the NiTiCu hape memory alloy by mean o calorimetric technique and mechanical tet. Thermal eiciencie ranged rom 4.7 % to 5.3 %. Generally eiciency o the SMA alloy i low compared to conventional heat engine. But the ue o Shape Memory Alloy actuator can reduce the ize, weight and complexity o the ytem. It power denity i remarkable high uch a 100 W/kg or more [21]. The conventional actuator produce a igniicant amount o noie, while the SMA actuator i completely ilent. Thu, SMA actuator can be ideal or cae, uch a robotic, micro/miniature and medical application, where power denity, implicity o mechanim and ilent actuation i more important than energy eiciency o the ytem. The thermal eiciencie o three ytem are hown in table 3. Thee value are very reaonable conidering the 3.12 % o the ideal heat engine eiciency. The battery ytem eiciencie how much higher value than thoe o other ytem due to it high eiciency in converting electrical energy to thermal energy and le additional power requirement. In implementation o thee ytem, the eiciency hould be lower than the value o the table 3 due to additional energy loe and additional power requirement. 3. NUMERICAL ANALYSIS OF THE SMA ACTUATOR 3.1 Numerical Heat Traner Analyi o SMA Actuator To etimate the period o the heating and cooling cycle, and thu actuation requency, a numerical heat traner analyi o the SMA actuator wa carried out with commercial otware package. The energy balance equation o the SMA trip can be written a: o T [ ρ Cp T + ρ H ] dv = k dv + q in dain q out daout + q gen dv (9) t z z with temperature T, thermal conductivity k, denity ρ, peciic heat o the SMA trip Cp, time t and latent heat H (J/g). The ubcript mean SMA trip. Where q in i energy input rate rom the hot luid, q out i energy lo rate rom the trip to environment and q gen repreent energy generation rate per volume. The SMA trip i heated by only convection heating. Thereore, there are no q out and q gen term. In the phae tranormation o an SMA, heat i aborbed during the revere tranormation (martenite to autenite) and it i releaed during the orward tranormation (autenite to martenite). Thi heat i called the latent heat o tranormation ( H ). Thi latent heat i expreed by a variation with * temperature o the redeined peciic heat o the SMA Cp. The area under the curve decribed by the peciic heat i the latent heat o tranormation. During phae tranormation the heating and the cooling o the SMA are lowed down due to the latent heat o tranormation. When doing a tranient heat traner analyi it i thereore important to account thi eect. An empirical relation decribing the dependence o the peciic heat with temperature i given in [5]. The certain tranormation temperature, latent heat and peciic heat capacity, uch a A =363.83 K and A = 352.42 K at the value o 150 M Pa tre, were utilized in the ollowing equation. For the orward tranormation it i:

For the revere tranormation it i: Cp Cp * o ln(100) Cp + H 2 ln(100) M + M T M M 2 = e (10) M M M < T < * o ln(100) Cp + H M 2ln(100) A + A T A A 2 = e (11) A A A < T < o The original peciic heat value o the SMA: Cp i 550 J/Kg K [17]. The curve obtained or the variation o the peciic heat o the SMA with temperature during the orward and revere tranormation at 150 M Pa tre are hown on igure 3. A 4500 4000 3500 3000 orward tranormation orward piecewie tranormation revere tranormation revere piecewie tranormation Variation o Cp with temperature Cp(J/kg-K) 2500 2000 1500 1000 500 0 330 335 340 345 350 355 360 365 370 Temperature(K) Figure 3. Variation o Cp with temperature. For the numerical calculation, the FLUENT 5.5 and GAMBIT 1.3.0 commercial package were ued. The initial condition wa T=To (335 K) and the boundary condition or the inlet velocity and temperature o the luid were 1 m/ec and 370 K, repectively. For heating medium and cooling medium, water wa elected. The low i turbulent low (Red = 3,630) and the tandard k-ε model with two-layer zonal model or the near wall region wa ued. The circular channel wall wa aumed to be an adiabatic boundary. Energy loe o both end o channel were ignored. The circular channel and an SMA trip (12 mm x 0.9 mm) embedded in thi channel (D =12 mm, L=200mm) i the computational domain hown in igure 4. The cro ection o the domain i hown in igure 5. Hal o the channel and trip were calculated by uing ymmetric condition. The grid conit o about 200,400 computational cell or the calculation. Thi grid wa generated by GAMBIT. The grid wa dener near the SMA trip in order to capture the boundary layer. * The igure 3 how a maximum increae o the value o the peciic heat (Cp ) o about 8 time o the original value. It wa diicult to get FLUENT to run with the peciic heat varying with temperature uch a igure 3. Intead a piecewie variation ha been adopted. The approximation ued are hown on igure 3. The area under thee approximation i the kept the ame a the original, thereore conerving the value o the latent heat o tranormation. However by conerving the ame maximum value o peciic heat during tranormation, the tranormation tart and inih temperature had to be changed. The tranormation i tarting later and inihing earlier.

Figure 4. The computational domain. Location 3 Location 2 Location 1 Figure 5. The computational grid. 3.2 Reult o Numerical Analyi The temperature calculation o the SMA trip i an unteady heat traner problem. The time tep wa 0.01 ec and maximum number o iteration or each time tep wa et at 200. The autenite inih temperature will be changed to around 360 K rom the 364 K due to piecewie approximation o peciic heat. The y+ value at the wall were le than 5 except entrance region. Thi how that the turbulent calculation wa correctly done. Ater 1.3 ec, the temperature ditribution along the trip wa everywhere higher than 360 K excluding corner o the trip, a hown in igure 6. The Figure 6 (a) how temperature ditribution along the length o the trip at three location. The irt i at the center o the trip (location 1 in igure 5), the third i the end o the trip (location 3), and the econd i the middle line between the irt and the third line (location 2). Figure 6 (b) how temperature ditribution o the whole trip. From the igure 5 and 6 the mall amount o the trip till wa in tranormation (M A), but mot region o the trip wa ully tranormed. Ater 0.9 ec, the temperature ditribution along the trip without conidering latent heat wa above the 364 K except the corner o the trip. The latent heat o trip hould be conidered in numerical calculation epecially or the tranient cae in order to avoid overetimating heat traner rate, thu actuation requency. Cooling cae wa alo calculated uing room temperature water with 1 m/ec o velocity and 360 K initial trip temperature. The cooling wa ater than heating. Thu the actuation requency wa etimated around 0.5 Hz, even under high tre loading condition. Generally higher inlet velocity and higher inlet temperature increae the heat traner rate to the SMA trip. The high boiling point luid uch a Ethylene Glycol can generate higher actuation requency. 370 Static Temperature ditribution along SMA trip Static Temperature(K) 368 366 364 362 Location 3 Location 1 Location 2 360 358 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 Length(m) (a) (b) Figure 6. Temperature ditribution along the SMA trip ater 1.3 ec [FLUENT].

4. FUEL-POWERED COMPACT SMA ACTUATOR SYSTEM 4. Firt Deign o Fuel-Powered Compact SMA Actuator Sytem From the comparion with other actuator ytem in ection 2, the uel powered SMA actuator ytem having the highet energy and power denity wa elected in order to develop compact SMA actuator ytem. To meaure available orce, diplacement and actuation requency, a uel-powered SMA actuator ytem wa deigned. The actuator ytem i compoed o a pump, valve, a combutor, an SMA element, a hot luid tank, bellow and heat exchanger. The bellow ued to prevent mixing between hot and cold luid. Figure 7 how the irt deign o the uel-powered SMA actuator ytem and the correponding, it experimental etup. The load i applied contantly by dead weight or entire heating and cooling cycle. The trip wa put through multiple thermal cycle by cycling it temperature rom the martenite inih temperature (T<M ) to the autenite inih temperature (T>A ) under a contant applied tre o 68 M Pa and 150 M Pa. SMA Heating Circuit Combutor V V SMA Actuator P V V Heat Exchanger 1. SMA Actuator 2. Ampliication device 3. Heater 4. Heat exchanger 5. Bellow 6. Pump SMA Cooling Circuit 1 2 4 5 6 3 Figure 7. Firt Deign o Fuel- Powered SMA Actuator Sytem: Schematic o baic architecture (upper) and correponding, it experimental etup (lower).

Figure 8 how loading path during entire heating and cooling cycle. At the higher actuation requency and under the higher tre condition, the trip can be in partial tranormation due to the deduction o heating duration and the increae o tranormation temperature. The train and diplacement were meaured with a LVDT. σ Platic Deormation M o M o 4.2 Component o the SMA Actuator Sytem A o A o Loading Path 2 Loading Path 1 Figure 8. Heating and cooling cycle under contant applied tre. SMA Actuator The SMA trip wa embedded in a circular ilicone tube, which i able to withtand high temperature and i able to bend and expand. The inide diameter o the tube i 12.7 mm and the wall thickne i 0.8 mm. Two teel connector were utilized to hold the trip, a well a provide upply and return connection or the luid low and to traner the actuation orce to the mechanical troke ampliication device. The ampliication mechanim increae the troke by 86 time o initial train o the trip and reduce the orce by 82 time. Figure 10 how the characteritic o the ampliication device. Figure 9. SMA Actuator with a circular ilicone channel and connector. String Pot (m) 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0-0.1 Diplacement Ratio y = 86.116x - 0.0019 R 2 = 1 0 0.002 0.004 0.006 0.008 LVDT(m) N(Load Cell) 1200 1000 800 600 400 200 0 y = 82.163x - 2.975 R 2 = 0.9989 Force Ratio 0 5 N(Weight) 10 15 Figure 10. Perormance characteritic o the ampliication mechanim.

The SMA trip i K-alloy type, which i a Ni, Ti and Cu alloy and ha 12 mm x 0.9 mm cro ection area and 254 mm long including connection. Thu available length o the actuator i about 210 mm. Out o all the Shape Memory Alloy that have been dicovered o ar, Nickel-Titanium ha proven to be the mot lexible and beneicial in engineering application. Ni-Ti Shape Memory Alloy ha greater ductility, more recoverable motion, excellent corroion reitance and table tranormation temperature [22]. Copper can decreae hyterei and lower the deormation tre (detwinning tre) o the martenite. The trip wa annealed at 450 C or 20 minute. The tranormation temperature o the trip were obtained by DSC tet ater annealing. Thi kind o SMA alloy ha around 70 C active A temperature and how high-energy eiciency compared to other type o SMA alloy [16]. For the irt actuator, one trip wa utilized a the SMA element o the actuator. Combutor and hot luid tank A 185gram compact commercial camping tove wa utilized a the combutor. It lame i adjutable up to 4000 watt and run on a valved butane and propane mix cartridge, which contain 227 g o uel. It can boil 1 liter o water within 3 minute. A 127 mm x 127 mm x 25.4 mm hot luid tank wa abricated by an aluminum block and ha a 635 mm low path inide o it. The aluminum block ealed by crew and RTV and wa heated directly by the camping tove. Two K-type thermocouple were ued to meaure the temperature o hot luid. The eiciency o thi camping tove wa etimated around 40 %. Heat Exchanger The cooling medium, ater it remove the heat rom the SMA trip goe through a heat exchanger where it dipoe o that heat. For the irt actuator ytem, two kind o radiator with a 472 ml/ec an were ued. One i an automobile heater core and the other i a PC cooling radiator. It meaure 152 mm x 178 mm x 5.1mm with a 870 Watt capacity. Pump with Motor The bra Omega FPUGR201 gear pump circulate heating and cooling media. It maximum low rate i 7.95 l/min at low preure and can withtand 0.69 M Pa and 149 C. Thi pump circulate hot luid and cold luid alternatively. One pump ytem wa elected to increae the energy denity o the ytem. However, it uer rom ome energy lo ince hot and cold luid heat and cool the gear pump itel. The pump wa run by a DC motor. The pump run contantly regardle o heating and cooling cycle. The olenoid valve control the heating and cooling medium circulated by the pump. Solenoid valve Four olenoid valve were utilized to control the heat and cooling circuit. The valve located beore the pump in igure 4 were GC direct operated diaphragm 2 way olenoid valve. They can operated in preure range rom 0 Pa to 0.69 M Pa, up to 145 C luid temperature and have a Cv o 3.3. The other two valve were Parker 0.83 Cv direct acting valve. The operating preure and temperature range are rom 0 Pa to 0.138 M Pa and up to 85 C, repectively. The GC valve are operated by on/o ignal coming rom controller while other two valve are controlled by bellow and witche. Bellow The mixing between hot and cold media mut be avoided to reduce energy loe. A bellow wa utilized to prevent mixing between hot and cold luid caued by haring common low path uch a actuator channel and the volume inide the pump. Alo, it can generate ome kind o orce to aid in the luid media circulation. A 152 mm troke, double acting double rod American air cylinder wa ued a a bellow. For the irt actuator ytem, the bellow and witche controlled two olenoid valve, which are located downtream o the SMA actuator (igure 4). An SR latch witch algorithm wa utilized. Control and Data acquiition The data acquiition board ued wa a National Intrument AT-MIO-16XE-50. The board ha 16 bit reolution and eight dierential input. Two digital channel were ued or the witche, three dierential channel or thermocouple and the LVDT. A HP6268B DC power upply provided the power to the ytem. A power plitter board wa controlled through a National Intrument PCI-6704 D/A board. The power plitter board take the ingle input rom the HP 6268B

power upply and plit it into ix individually controlled channel, which are ued to upply controlled power to motor and the olenoid valve. A Lab-window program wa put together to control the hardware and acquire data. 5. EXPERIMENTAL RESULTS AND DISCUSSIONS Water wa elected a heating and cooling medium rom the numerical reult becaue it wa eay to handle and the boiling temperature o water i high enough to heat and cool the SMA trip around 0.5 Hz under high tre (150 M Pa). The inlet velocity and inlet temperature were alo determined baed on numerical analyi. The volume low rate o hot water and cold water wa et a 0.11 l/ec by adjuting the power o the pump, thu the low velocity inide the channel wa around 1 m/ec. The actuator ytem wa teted under 735 N contant load. Figure 11 how the diplacement o the trip under 735 N load or the cloe loop ytem. The hot water, ater heating the SMA trip, returned back to the heater, which added enough energy to the hot water to compenate or it energy lo due to heating o SMA trip and other energy diipation. The cold water, ater cooling the SMA trip, returned back to the radiator where the thermal energy removed rom SMA trip wa diipated to the urrounding by air orced convection. 0.5 Diplacement.v. Time : Heating time = 5 ec 0.25 Diplacement.v. Time : Heating time i 0.5 ec 0.45 0.2 0.4 0.15 Diplacement(inch) 0.35 0.3 Diplacement(inch) 0.1 0.05 0.25 0 0.2-0.05 0.15 0 50 100 150 200 250 300 time(ec) Figure 11. Strain V. Time under 735 N or the cloe loop ytem. -0.1 20 22 24 26 28 30 32 34 36 38 40 time(ec) Figure 12. Strain V. Time under 735 N or the open loop ytem. The 2.5 % train and 0.1 Hz (heating time: 5 ec & cooling time: 5 ec) actuation requency wa obtained rom the cloe loop ytem tet. Thi reult how lower train and lower requency compared to the open loop ytem, in which the hot and cold water wa not re-circulated in the ytem. The igure 12 and 13 how the reult or the open loop ytem. The SMA actuator ytem can generate enough recovery tre and train at 0.5 Hz actuation requency. Figure 12 how higher requency under 735N than that o Figure 13 under 1560 N. The maximum operating preure o the cloe loop ytem i about 69 K Pa in the heating (hot luid) circuit. Preure loe o the ytem and water vapor caued thi preure. In the open loop tet, actuation requency o the SMA actuator (K-alloy, 12 mm x 0.9 mm) can be raied to 1 Hz (heating and cooling cycle) under 68 M Pa load (735 N), and up to 0.5 Hz under 150 M Pa load (1560 N). At leat 3 % train can be obtained by utilizing hot water (370 K) and cold water (295 K). The lower train and lower actuation requency in the cloe loop ytem wa mainly due to the mixing between hot and cold luid in the ytem. A the number o cycle increae, the energy lo caued by the mixing increae. The cylinder and pump are alo heat lo device becaue they contain heating and cooling medium alternatively. Hence the hot luid temperature goe down and the cold luid temperature goe up. Thi degrade the perormance o the heating and cooling circuit. More powerul heater and heat exchanger can overcome thi mixing, but in that cae the ytem might loe compactne and high energy denity. One o the method to prevent mixing i to adjut troke length o the cylinder (bellow), thu making the channel volume and volume o other haring paage equal to the volume o

bellow. I the mixing and energy lo were prevented properly, at leat 0.5 Hz actuation requency and 3 % recovery train could be obtained. Diplacement.v. Time : Heating time = 1 ec 0.15 0.1 0.05 Diplacement(inch) 0-0.05-0.1-0.15-0.2 20 22 24 26 28 30 32 34 36 38 40 time(ec) Figure 13. Strain V. Time under 1560 N or the open loop ytem. The heating period o experimental reult under high tre condition and 3 % recovery train wa around 1.0 ec in the open loop tet. Thi how omewhat ater than that o the numerical reult, which how the heating period around 1.3 ec at 150 M Pa tre conidering latent heat o the tranormation. The partial tranormation o the trip in experimental tet and uncertainty o the material propertie might caue thi dierence. The two-way trained wire can reult in 4 % recovery tre when it i ully tranormed [20]. The experimental reult how 3 % recovery train. The trip might be under partial tranormation due to high tre condition and relatively high actuation requency. Additional tet or the propertie o SMA trip uch a peciic heat need in order to increae the accuracy o the numerical analyi. Thu the numerical reult are reaonable conidering partial tranormation o the trip and uncertainty in propertie o SMA trip. A mentioned above, the 0.5 Hz actuation requency under 150 M Pa tre i obtained rom the experimental tet. Thi value how good agreement with the prediction in ection 3. 6. CONCLUSIONS The uel-powered SMA actuator ytem i imple and compact potentially compared to other actuator ytem. The comparion how that thi ytem ha much higher energy and power denity than that o the battery and the uel cell powered SMA actuator ytem. The orced convection heating and cooling generated relatively high actuation requency compared to reitive heating and air orced convection cooling. The reult o the numerical heat traner analyi are ueul and reaonable compared to the reult o the experimental tet. Thu, in deigning o the thermal induced SMA actuator, the tranient heat traner analyi with latent heat mut be carried out and conirmed numerically to determine the deign parameter and operating condition. The uel powered actuator ytem i being developed and modiied to get high orce actuation under relatively high actuation requency. The irt deigned SMA actuator ytem could actuate the SMA trip (12mm x 0.9mm x 254 mm) at 0.5 Hz under 150 M pa tre with 3 % train. Thi requency i airly high conidering the ize o trip. Fuel uch a butane and propane i relatively cheap, eay to handle and eay to tore in compact module like ga cartridge. Hence the operating cot o thi actuator ytem will be lower compared to batterie and uel cell. Thi reearch alo how the energy aving that SMA preent u with, in ytem where paraitic heat i already preent. Heating o the SMA by utilizing exiting paraitic heat in the vehicle/plant, thi will yield a high energy denity actuator. We are ocuing on optimization and miniaturization o the actuator ytem in order to increae power and energy denity. A pecial combutor, valve, a pump and heat exchanger are being deigned to develop the highly compact actuator ytem. The microchannel technology will be ued to increae heat traner rate, eiciency and compactne o the combutor and heat exchanger. The next deign o compact actuator ytem will be deigned to be modular in order to allow or alternative energy ource uch a batterie, uel cell and paraite heat. The energy denity and power

denity o the SMA actuator ytem will be meaured at the next deign o compact actuator ytem. Actuator atigue tet will be required to determine the lie o the actuator a a unction o actuation tre, train and requency. Tet with high boiling point luid uch a Ethylene Glycol will be perormed to get high actuation requency and to increae energy denity o the ytem. 7. ACKNOWLEDGMENTS The author would like to acknowledge the inancial upport o DARPA through the ARO Grant No DAAD19-01- 0804. The author would alo like to thank Dr. Gary Anderon or hi upport and technical interet in thi reearch. 8. REFERENCES 1. Johnon, A.D, Nitinol Heat Engine, Interociety Energy Converion Engineering Conerence Proceeding, pp530-534, 1975. 2. William S. Ginell, Joeph L. Mcnichol, Jr, and John S.Cary, Nitinol Heat Engine or low-grade thermal engery converion, Mechanical Engineering, pp28-33, 1979. 3. Zurab Saralidze, Deign and Creation o Heat Engine Working on the Bai o Phae Tranormation in Solid and Uing Nontraditional Source o Heat Energy, Intitute o Phyic Georgian Academy o Science, Tbilii Georgia, 2001. 4. Boyd, J. G and Lagouda, D.C. Thermomechanical repone o hape memory compoite, J. Intell. Mater. Struct 5, pp336-346, 1994. 5. A. Bhattcharyya, D.C, Lagouda, Y wang and V.K, Kinra, On the role o thermoelectric heat traner in the deign o SMA actuator; theoretical modeling and experiment,smart Mater. Struct 4, pp252-263, 1995. 6. K.P. Brook, C.J. Call, M. K. Drot, Integrated Microchannel Combutor/Evaporator Development, ASME IMECE Conerence, Nahville, TN, 1999. 7. Chad Harri, Mircea Depa, and Kevin Kelly, Deign and Fabrication o a Cro Flow Micro Heat Exchanger, Journal o Microelectromechanical Sytem, Vol 9, No 4, pp502-508, 2000. 8. Redinioti, O.K., Lagouda, D.C., Garner, L. and Wilon, N. (1998) Experiment and Analyi o an Active Hydrooil with SMA Actuator, AIAA Paper No. 98-0102, 36th AIAA Aeropace Science Meeting, Reno, Nevada, January 1998. 9. Webb, G., Wilon, L., Lagouda, D.C. and Redinioti, O.K., Adaptive Control o Shape Memory Alloy Actuator or Underwater Biomimetic Application, AIAA Journal, Vol. 37, No.12, Dec. 1999. 10. Redinioti, O. K., Lagouda, D. C. and Wilon, L.N. (2000) Development o a Shape Memory Alloy Actuated Underwater Biomimetic Vehicle, AIAA Paper No. 2000-0522, 38th Aeropace Science Meeting and Exhibit, January 2000, Reno, Nevada. 11. Bruce Lin Conceptual deign and modeling o hydrogen uel cell cooter or urban aia, Mater thei, Mechanical & Aeropace Engineering Department o Princeton Univerity, Sept. 1999. 12. Fuel cell power ytem or underwater vehicle, http://www.uelcell.koone.com. 13. Generic battery Technology comparion, http://www.madkatz.com. 14. Energy Denity, http://www.jrc.e/iptreport/vol36/englih/arte5en.gi, In. J. Hydrogen energy, Sept. 1996. 15. D.C. Lagouda and A. Bhattcharyya, Modeling o Thin Layer Extenional Thermoelectric SMA Actuator, In. J.Solid Structure, Vol. 35, pp.331-362, 1998 16. A.P. Jardine, Calorimetric technique or the evaluation o thermal eiciencie o hape memory alloy, Journal o Material Science, 24, pp.2587-2593, 1989. 17. F.J. Gil, J.A. Planell, Thermal eiciencie o NiTiCu hape memory alloy,thermochimica Acta 327, pp.151-154, 1999. 18. A.P. Jardine, Calorimetric meaurement o tranormation thermodynamic and thermal eiciencie o NiTi helice, Journal o Material Science, 23, pp.3314-3320, 1988. 19. B. Cunningham and K.H.G. Ahbee, MARMEM ENGINES, Acta Metallurgica, Vol. 25, pp.1315-1321, 1977. 20. David A. Miller and Dimitri C. Lagouda, Thermo-Mechanical characterization o NiTiCu and NiTi SMA Actuator: Inluence o platic train Smart Material and Structure, 2000. 21. Koji Ikuta, Micro/Miniature hape memory alloy actuator, IEEE, pp. 2156-2161, 1990. 22. Waram. T, Actuator Deign uing Shape Memory Alloy, 2nd Edition.