Microscale combustion and power generation Jeongmin Ahn, James Kuo, Lars Sitzki, Craig Eastwood, Paul Ronney Dept. of Aerospace & Mechanical Engineering Univ. of Southern California, Los Angeles, CA (also 5, 6) Supported by DARPA Microsystems Technology Office

Dept. of Aerospace & Mechanical Engineering - University of Southern California Outline n Motivation n Scaling of micro power generation n Some power MEMS concepts n USC design - microFIRE l Overall approach l Engineering development l Experimental and modeling results u Heat recirculating combustors u Effects of wall heat conduction u Catalytic combustion n Related projects n Conclusions

Dept. of Aerospace & Mechanical Engineering - University of Southern California What is microcombustion? n PDR’s definition: microcombustion occurs in small-scale flames whose physics is qualitatively different from conventional flames used in macroscopic power generation devices, specifically l The Reynolds numbers is too small for the flow to be turbulent and thus allow the flame reap the benefits of flame acceleration by turbulence AND l The flame dimension is too small (i.e. smaller than the quenching distance, Pe < 40), thus some additional measure (heat recirculation, catalytic combustion, reactant preheating, etc.) is needed to sustain combustion

Dept. of Aerospace & Mechanical Engineering - University of Southern California Motivation - the seductive lure of chemical fuels

Dept. of Aerospace & Mechanical Engineering - University of Southern California The challenge of microcombustion n Hydrocarbon fuels have numerous advantages over batteries l ≈ 100 X higher energy density l Much higher power / weight & power / volume of engine l Inexpensive l Nearly infinite shelf life l More constant voltage, no memory effect, instant recharge l Environmentally superior to disposable batteries u $31 billion/yr of disposable batteries ends up in landfills u $6 billion/yr market for rechargables

Dept. of Aerospace & Mechanical Engineering - University of Southern California The challenge of microcombustion n … but converting fuel energy to electricity with a small device has not yet proved practical despite numerous applications l Foot soldiers l Portable electronics - laptop computers, cell phones, … l Micro air and space vehicles (enabling technology) n Most approaches use scaled-down macroscopic combustion engines, but may have problems with l Heat losses - flame quenching, unburned fuel & CO emissions l Heat gains before/during compression l Limited fuel choices – need knock-resistant fuels, etc. l Friction losses l Sealing, tolerances, manufacturing, assembly

Dept. of Aerospace & Mechanical Engineering - University of Southern California Scaling of micro power generation n Heat losses vs. heat generation Heat loss / heat generation ≈ 1/  at limit (  = non- dimensional activation energy) Heat loss / heat generation ≈ 1/  at limit (  = non- dimensional activation energy) Premixed flames in tubes or channels: Pe  S L d/  ≈ 40 - as d , need S L  (stronger mixture) to avoid quenching Premixed flames in tubes or channels: Pe  S L d/  ≈ 40 - as d , need S L  (stronger mixture) to avoid quenching S L = 40 cm/s,  = 0.2 cm 2 /s  quenching distance ≈ 2 mm for stoichiometric HC-air S L = 40 cm/s,  = 0.2 cm 2 /s  quenching distance ≈ 2 mm for stoichiometric HC-air Note  ~ P -1, but roughly S L ~ P -0, thus can use weaker mixture at higher P Note  ~ P -1, but roughly S L ~ P -0, thus can use weaker mixture at higher P l Also: Pe = 40 assumes cold walls - less quenching problem with higher wall temperature (obviously)

Dept. of Aerospace & Mechanical Engineering - University of Southern California Scaling of micro power generation n Gas-phase (volumetric) vs. catalytic (surface) heat release rate H (in Watts) l Gas-phase: H = Q R *  *(rate/volume)*volume; rate/volume ~ A gas exp(–E gas /RT), volume ~ d 3  H~  Q R A gas exp(–E gas /RT)d 3 l Catalytic: H =  Q R *(rate/area)*area, area ~ d 2 ; rate/area can be transport limited or kinetically limited u Transport limited (large scales, low flow rates) s Rate/area ~ diffusivity*gradient ~  D(1/d)  Rate/area ~  D/d  H ~  Q R Dd u Kinetically limited (small scales, high flow rates, near extinction) s Rate/area ~ A surf exp(–E surf /RT)  H ~  Q R d 2 A surf exp(–E surf /RT) n Ratio gas/surface reaction l Transport limited: H gas /H surf = A gas exp(–E gas /RT)d 2 /D ~ d 2 l Kinetically limited: H gas /H surf = A gas exp(–E gas /RT)d/(A surf exp(– E surf /RT)) ~ d  Catalytic combustion more may be faster than gas-phase combustion at sufficiently small scales

Dept. of Aerospace & Mechanical Engineering - University of Southern California Scaling of micro power generation n Flame quenching revisited l Heat loss (by conduction) ~ g (Area)  T/d ~ g d 2  T/d ~ d l Heat loss / heat generation ~ d -2 (gas-phase combustion) ~ d -1 (surface, transport limited) ~ d 0 (surface, kinetically limited, relevant to microcombustion)  Catalytic combustion may be necessary at small scales to avoid quenching by heat losses!

Dept. of Aerospace & Mechanical Engineering - University of Southern California Scaling of micro power generation n Turbulence (example: IC engine, bore = stroke = d) Re = U p d/ ≈ (2dN)d/ = 2d 2 N/ Re = U p d/ ≈ (2dN)d/ = 2d 2 N/ U p = piston speed; N = engine rotational speed (rev/min) l Minimum Re ≈ several 1000 for turbulent flow l Need N ~ 1/d 2 or U p ~ 1/d to maintain turbulence (!) l Typical auto engine at idle: Re ≈ (2 x (10 cm) 2 x (600/60s)) / (0.15 cm 2 /s) = high enough for turbulence l Cox Tee Dee: Re ≈ (2 x (0.6 cm) 2 x (30000/60s)) / (0.15 cm 2 /s) = high enough for turbulence (barely) (maybe) l Why need turbulence? Increase burning rate - but how much? u Turbulent burning velocity (S T ) ≈ turbulence intensity (u’) u u’ ≈ 0.5 U piston (Heywood, 1988) ≈ dN u ≈ 67 cm/s > S L (auto engine at idle, much more at higher N) u ≈ 300 cm/s >> S L (Cox Tee Dee)

Dept. of Aerospace & Mechanical Engineering - University of Southern California Scaling of micro power generation n Fricton due to fluid flow in piston/cylinder gap Shear stress (  ) = µ oil (du/dy) = µ oil U p /h Shear stress (  ) = µ oil (du/dy) = µ oil U p /h Friction power =  x area x velocity = 4µ oil U p L 2 /h = 4µ oil Re 2 2 /h Friction power =  x area x velocity = 4µ oil U p L 2 /h = 4µ oil Re 2 2 /h Thermal power = mass flux x C p x  T combustion =  S T d 2 C p  T =  (U p /2)d 2 C p  T =  Re)dC p  T/2 Thermal power = mass flux x C p x  T combustion =  S T d 2 C p  T =  (U p /2)d 2 C p  T =  Re)dC p  T/2 Friction power / thermal power = [8µ oil (Re) ]/[  C p  Thd)] ≈ for macroscale engine Friction power / thermal power = [8µ oil (Re) ]/[  C p  Thd)] ≈ for macroscale engine l Implications u Need Re ≥ Re min to have turbulence  Material properties µ oil,,  C p,  T essentially fixed  For geometrically similar engines (h ~ d), importance of friction losses ~ 1/d 2 ! l What is allowable h? Need to have sufficiently small leakage u Simple fluid mechanics: volumetric leak rate = (  P)h 3 /3µ u Rate of volume sweeping = Ud 2 - must be >> leak rate  Need h << (3   dRe min /  P) 1/3 u Don’t need geometrically similar engine, but still need h ~ d 1/3, thus importance of friction loss ~ 1/d 4/3 !

Dept. of Aerospace & Mechanical Engineering - University of Southern California n Cox Tee Dee.010 Application: model airplanes Weight: 0.49 oz. Bore: 0.237” = 6.02 mm Stroke: 0.226” = 5.74 mm Displacement: cu in (0.163 cm3) RPM: 30,000 Power:5 watts n Poor performance l Low efficiency (4-5%) l Emissions & noise unacceptable for indoor applications n Not “microscale” Re = Ud/ ≈ (2 x 0.6cm x (30000/60s)) (0.6cm) / (0.15 cm 2 /s) = high enough for turbulence (barely) Re = Ud/ ≈ (2 x 0.6cm x (30000/60s)) (0.6cm) / (0.15 cm 2 /s) = high enough for turbulence (barely) l Size > quenching distance even at 1 atm, nowhere near q.d. at post- compression condition Smallest existing combustion engine

Dept. of Aerospace & Mechanical Engineering - University of Southern California Some power MEMS concepts Wankel rotary engine (Berkeley) Free-piston engines (U. Minn, Georgia Tech)

Dept. of Aerospace & Mechanical Engineering - University of Southern California Some power MEMS concepts Pulsed combustion driven turbine (some small school in west LA with a bear for a mascot..)

Dept. of Aerospace & Mechanical Engineering - University of Southern California Some power MEMS concepts Liquid piston magnetohydrodynamic (MHD) engine (Honeywell / U. Minn)

Dept. of Aerospace & Mechanical Engineering - University of Southern California Some power MEMS concepts - gas turbine (MIT) Friction & heat losses heat transfer along casing & rotor, from turbine to compressor Made from silicon - very high thermal conductivity - heat transfer along casing & rotor, from turbine to compressor Very high rotational speed (≈ 2 million RPM) needed for compression (speed of sound doesn’t scale!) Manufacturing tolerances Not microscale according to PDR’s definition: Re ≈ 1000, combustor scale > quenching distance Not microscale according to PDR’s definition: Re ≈ 1000, combustor scale > quenching distance Mixing time vs. chemical time - mixing time scales with combustor size but reaction time does not - need larger relative chamber size as scale decreases … Mixing time vs. chemical time - mixing time scales with combustor size but reaction time does not - need larger relative chamber size as scale decreases …

Dept. of Aerospace & Mechanical Engineering - University of Southern California H 2 PEM fuel cells - CWRU - Savinell et al. n Up to 5 mW/cm 2 demonstrated n Can use borohydride solutions for H 2 storage: ≈ 7 mass % H 2 l NaBH H 2 O  NaBO H 2

Dept. of Aerospace & Mechanical Engineering - University of Southern California Direct methanol fuel cell Methanol is easily stored compared to H 2, but has ≈ 6x lower energy/mass and requires a lot more equipment! (CMU concept shown)

Dept. of Aerospace & Mechanical Engineering - University of Southern California Our approach - microFIRE n Integrated microscale power generation system l Combustion l Heat transfer l Electrical power generation l Fabrication & assembly n “Swiss-roll” heat recirculating burner with toroidal 3D geometry n Direct thermoelectric conversion of heat to electricity n Monolithic fabrication of the entire device with EFAB n Targets l Shirt button size l Weight 500 mg l Volume 0.04 cc l Power 100 mW l Efficiency > 10%

Dept. of Aerospace & Mechanical Engineering - University of Southern California n “Swiss roll” heat recirculating burner - minimizes heat losses Toroidal 3D geometry: further reduces losses - minimizes external temperature on all surfaces microFIRE approach (1) – Combustion 1D counterflow heat exchanger and combustor 2D “Swiss roll” combustor (Weinberg, 1970’s)

Dept. of Aerospace & Mechanical Engineering - University of Southern California n Thermoelectric (TE) elements n Same principal as thermocouple, material optimized for power generation n Imbed in wall between hot (outgoing product) and cold (incoming reactant) streams microFIRE approach (2) - Power generation Overall configuration - Wall itself is electrical conductor Typical thermoelectric configuration - alternating n- and p-type elements Products Reactants Combustion volume K Thermoelectric elements

Dept. of Aerospace & Mechanical Engineering - University of Southern California n Widely used in deep space missions, some commercial applications TE efficiency typically 15% of Carnot with same  T - not bad - but how to get large  T ? TE efficiency typically 15% of Carnot with same  T - not bad - but how to get large  T ? Thermoelectrics

Dept. of Aerospace & Mechanical Engineering - University of Southern California Thermoelectric microgenerator problem n TE wall material: thermal conductivity k ≈ 1 W/m˚C n Gas: k ≈ W/m˚C  Thermal resistance between gas & TE wall (=1/hA ≈ 7.5 d/k air A) >> resistance across TE (=  X/k TE A)  Most  T between gas & TE wall, not across TE  No power generation! n Note with surface catalytic combustion on hot side of TE, thermal resistance can be much lower - but what about cold side? n Macroscale devices - strong turbulence, convective heat transfer, low thermal resistance, but microscale Reynolds # too low! n Need “dirty tricks” ! l Special fin designs l Special TE material processing

Dept. of Aerospace & Mechanical Engineering - University of Southern California microFIRE approach (3) - Fabrication n EFAB (Electrochemical Fabrication) l Enables fabrication of arbitrarily complex 3D structures l JPL (Pasadena, CA) process for electrochemical deposition of TE elements l Enables fabrication using a single monolithic process n Micromachining process developed at USC n Commercialization by MEMGen Inc., Burbank, CA n Analogous to macroscale “rapid prototyping,” “solid freeform fabrication” n Produces arbitrary 3-D structures by stacking layers n No clean room required for device fabrication

Dept. of Aerospace & Mechanical Engineering - University of Southern California “Instant Masking” n Pre-fabricated masks serve as reusable “printing plates” n Polymer mask patterned on anode using conventional photolithography

Dept. of Aerospace & Mechanical Engineering - University of Southern California EFAB process flow Selectively deposit 1st material Using “Instant Masking” + Planarize layer Repeat for all layers Blanket deposit 2nd material Etch sacrificial material Selectively deposited material (usually sacrificial) Blanket deposited 2nd material (usually structural)

Dept. of Aerospace & Mechanical Engineering - University of Southern California EFAB highlights n Minimum isolated feature: 20 x 20 µm, maximum feature size ≈ 5000 µm 90  m Nickel micro-combustor 38 layers, 300 µm tall 12-layer chain, ≈ 290  m wide (world’s narrowest?) 300 µm 2nd generation folded micro-combustor with Pt

Dept. of Aerospace & Mechanical Engineering - University of Southern California Electrodeposited thermoelectrics Deposition of Bi 2 Te 3 elements 13H e - + 2BiO + + 3HTeO 2 +  Bi 2 Te 3 ¯ + 8H 2 O l JPL experience  50  m tall x 10  m diameter TE elements  Up to 11,000 grown with 30  m spacing n Other TE materials also possible; optimize for operating temperature range 100  m

Dept. of Aerospace & Mechanical Engineering - University of Southern California Heat losses  3D Swiss roll Friction losses  no moving parts Sealing  no moving parts Tolerances  no moving parts Manufacturing  EFAB microscale rapid prototyping Assembly  EFAB microscale rapid prototyping microFIRE: challenges met

Dept. of Aerospace & Mechanical Engineering - University of Southern California n Integrated system n Heat losses / flame quenching problems minimized n External T (IR signature, touch-temperature hazards) minimized n No moving parts! n Can use wide variety of fuels without pre-processing n Can use fuel/air mixtures not flammable without heat recirculation - intrinsically safe if combustion stops & unburned mixtures pass through device n Can use diluted fuels not flammable in open air - intrinsically safe fuel storage (e.g. for air travel) n Monolithic fabrication of entire device with minimal assembly n Can electrodeposit materials for catalytic combustion (e.g. Pt) l Lower combustion temperatures l Possibly self-starting (no igniter needed) with some fuels (ethers, alcohols) microFIRE advantages

Dept. of Aerospace & Mechanical Engineering - University of Southern California Engineering approach n Test macroscale and mesoscale versions of microscale burner n Use experiments to calibrate/verify CFD simulations n Demonstrate l Scale down process (macro  meso) l Ability to model (macro, meso) n Use CFD models to optimize microscale device (difficult to use diagnostics in microscale experiments) l Meso  micro

Dept. of Aerospace & Mechanical Engineering - University of Southern California Fabrication of macroscale devices n 2-D facilitates diagnostics, 3-D simulates microscale devices n Soligen™ rapid prototyping process for 2-D and 3-D designs in Al 2 O 3 - SiO 2 ceramic n Soligen devices have porous walls, significant gas leakage! n 2-D inconel designs - no leakage, but high thermal conductivity & thermal expansion coefficient n 2-D titanium - much lower conductivity & expansion than inconel 2D Soligen 3D Soligen 2D titanium

Dept. of Aerospace & Mechanical Engineering - University of Southern California Implementation of macroscale experiments n PC control and data acquisition using LabView n Mass flow controllers for fuel & air n Thermocouples

Dept. of Aerospace & Mechanical Engineering - University of Southern California “Flameless combustion” n Combustion in heat- recirculating burners usually occurs in “flameless” mode - no visible flame! n Also seen in highly preheated air combustion (Wünning and Wünning (1997), Katsuki & Hasegawa (1998), Maruta et al (2000)) n Reaction zone much more distributed than conventional combustion (residence time at high temp. ≈ 50 ms vs. 0.1 ms for conventional flame) n Chemical mechanism different - less CH & C 2 formation - more like “flow reactor” than flame Sitzki et al., 2001

Dept. of Aerospace & Mechanical Engineering - University of Southern California Quenching limits n Dual limits - low-velocity (heat loss) and high-velocity (blow-off) n Out-of-center combustion regime n Very low Re (< 4) possible with catalytic combustion n Lean limit can be richer than stoichiometric (!) (catalytic only) n Weinberg low-Re performance very poor (more heat losses?)

Dept. of Aerospace & Mechanical Engineering - University of Southern California Quenching limits in Swiss roll n Lower Re l Flame always centered - heat recirculation needed to obtain sufficiently high temperature to sustain reaction l Maximum temperatures near stoichiometric n Higher Re l Flame not centered near stoichiometric - less heat recirculation needed to sustain combustion - reaction zone moves toward inlet l Center cool due to heat losses

Dept. of Aerospace & Mechanical Engineering - University of Southern California Quenching limits n Advantage of catalytic combustion NOT mainly due to lower heat loss, but rather higher reaction rate at a given temperature

Dept. of Aerospace & Mechanical Engineering - University of Southern California Quenching limits - continued n Ratio of (estimated) heat loss to heat generation ≈ constant for low Re (indicating heat loss induced extinction) n Ratio decreases at higher Re (indicating “blow-off“ type extinction)

Dept. of Aerospace & Mechanical Engineering - University of Southern California Quenching limits - continued n Temperatures can be reduced dramatically with Pt catalyst - < 500 K even at Re < 4 - reduced materials issues

Dept. of Aerospace & Mechanical Engineering - University of Southern California Thermal behavior of Swiss roll n Peak temperatures correlate well with heat recirculation parameter =  |T i -T i-1 |/(T ad -T ∞ )}

Dept. of Aerospace & Mechanical Engineering - University of Southern California Ignition temperatures n Self-starting fuels & catalysts highly desirable - simpler system n H 2 self-starting at 25˚C on Pt catalyst, dimethyl ether (DME) at < 150˚C, hydrocarbons mostly < 250 ˚C n Even if self-starting not achieved, ignition T is low - can use TE elements as heaters for ignition

Dept. of Aerospace & Mechanical Engineering - University of Southern California Quenching limits - continued  Behavior of different fuels similar when scaled by enthalpy flux ~ (% fuel x flow velocity) & Re  Lower (heat loss) limit: (Heat loss)/(Heat gen) = const  Upper (blowoff) limit: Heat gen ~ 1/(residence time)

Dept. of Aerospace & Mechanical Engineering - University of Southern California Mesoscale experiments n Wire-EDM fabrication n Pt igniter wire / catalyst

Dept. of Aerospace & Mechanical Engineering - University of Southern California Mesoscale experiments n Steady combustion obtained with H 2 even at < 100˚C with Pt catalyst n Sharp transition to lower T at low or high fuel conc., low or high flow velocity - transition from gas-phase to surface reaction? n Can’t reach as low Re as macroscale burner! n Wall thick and has high thermal conductivity - loss mechanism?

Dept. of Aerospace & Mechanical Engineering - University of Southern California Mesoscale experiments n Next generation mesoscale burner - ceramic rapid prototyping using colloidal inks (Prof. Jennifer Lewis, UIUC) n Lower thermal conductivity, thinner walls than EDM parts 1.5 cm tall 2-turn alumina Swiss-roll combustor

Dept. of Aerospace & Mechanical Engineering - University of Southern Californiainletoutlet Numerical model n FLUENT, 2D, 8216 grid points n Conduction (solid & gas), convection (gas), radiation (solid- solid only) n Temperature-dependent gas properties n 1-step & 4-step chemistry (Westbrook & Dryer) n Boundary condition: l Inlet: 300K, plug flow l Outlet: Pressure outlet l Heat loss: volumetric term to simulate heat loss in 3rd dimension n Radiation: discrete ordinates, unit emissivity on all surfaces

Dept. of Aerospace & Mechanical Engineering - University of Southern California Numerical modeling n T flame T adiabatic at lower % fuel High % fuel High % fuel Low % fuel Low % fuel Reaction rates Temperatures

Dept. of Aerospace & Mechanical Engineering - University of Southern California Model results - heat loss effects n Low Re - dominated by heat loss - need ≈ 3x fuel to get similar performance if heat loss considered n Low Re, no heat loss: Thin reaction zone (laminar flame), anchored near inlet (doesn’t need heat recirculation to exist), rest of burner acts as heat sink n No lower limit on Re exists without heat loss! n High Re: little effect of heat loss Reaction rates Temperatures Reaction rates Temperatures (4.35% propane, Re = 58) without heat loss (1.5% propane, Re = 58) With heat loss (4.35% propane, Re = 58) T max ≈ 1540K T max ≈ 1220K T max ≈ 2420K

Dept. of Aerospace & Mechanical Engineering - University of Southern California Model results - radiation effects n Higher temperatures (by ≈150K) without radiation, more nearly isothermal n Radiation transfers heat between walls but not directly to gas - similar effect as increasing wall thermal conductivity n Important for scale-down modeling - radiation will be less significant at smaller scales due to higher gradients for conduction n Boltzman number  T 3 d/ - increases with scale (d) Reaction rates Temperatures No radiation With radiation

Dept. of Aerospace & Mechanical Engineering - University of Southern California n 217,000 cells n Simulation confirms most of heat loss is in axial (z) direction Numerical modeling - 3D

Dept. of Aerospace & Mechanical Engineering - University of Southern California Modeling - 2D vs. 3D n High Re (blowoff) limit well predicted by 2D model n Low Re (heat loss) limit not well predicted because most loss is in 3rd dimension! n 3D prediction for low Re close to experiment n 2D prediction close to 3D experiment – similar heat loss

Dept. of Aerospace & Mechanical Engineering - University of Southern California TemperatureHeat release rate 4-step model (Hauptmann et al.) n 4-step chemical model integrated into FLUENT (1) C 3 H 8  (3/2)C 2 H 4 + H 2 (2) C 2 H 4 + O 2  2CO + 2H 2 (3) CO + (1/2)O 2  CO 2 (4) H 2 + (1/2)O 2  H 2 O n Typical results (V = 20 cm/s, Re = 70, lean propane-air)

Dept. of Aerospace & Mechanical Engineering - University of Southern California Effect of wall thermal conduction n Simple quasi-1D analytical model of counterflow heat-recirculating burners developed including l (1) heat transfer l (2) chemical reaction in well-stirred reactor l (3) heat loss to ambient l (4) streamwise thermal conduction along wall

Dept. of Aerospace & Mechanical Engineering - University of Southern California Effect of wall thermal conduction n Results show low-velocity limit requires heat loss (H > 0) and wall heat conduction (B < ∞) n Very different from burners without heat recirculation! H = dimensionless heat loss B -1 = dimensionless wall conduction effect Da = dimensionless reaction rate

Dept. of Aerospace & Mechanical Engineering - University of Southern California Heat exchanger / combustor modeling n High-velocity limit almost unaffected by wall conduction, but low-velocity limit dominated by wall conduction n Thin wall, low thermal conductivity material (ceramic vs. steel) will maximize performance

Dept. of Aerospace & Mechanical Engineering - University of Southern California Catalytic combustion modeling n Collaboration with Prof. Kaoru Maruta (Tohoku Univ., Sendai, Japan) n Detailed catalytic combustion model (Deutschmann et al.) integrated into FLUENT computational fluid dynamics package n Interactions of chemical reaction, heat loss, fluid flow modeled in simple geometry at microscales l Cylindrical tube reactor, 1 mm dia. x 10 mm length, no wall thermal conduction l Platinum catalyst, CH 4 -air and CH 4 -O 2 -N 2 mixtures n Effects studied l Heat loss coefficient (H) l Flow velocity or Reynolds number ( ) l Fuel/air AND fuel/O 2 ratio

Dept. of Aerospace & Mechanical Engineering - University of Southern California Catalytic combustion modeling n “Dual-limit” behavior similar to experiments observed when heat loss is present

Dept. of Aerospace & Mechanical Engineering - University of Southern California Catalytic combustion modeling n Surface temperature profiles show effects of l Heat loss at low flow velocities l Axial diffusion (broader profile) at low flow velocities

Dept. of Aerospace & Mechanical Engineering - University of Southern California Catalytic combustion modeling n Heat release inhibited by high O(s) coverage (slow O(s) desorption) at low temperatures - need Pt(s) sites for fuel adsorption / oxidation a b Heat release rates and gas-phase CH 4 mole fractionSurface coverage

Dept. of Aerospace & Mechanical Engineering - University of Southern California Catalytic combustion modeling n Computations with fuel:O 2 fixed, N 2 (not air) dilution n Minimum fuel concentration and flame temperatures needed to sustain combustion much lower for even slightly rich mixtures! n Typical strategy to reduce flame temperature: dilute with excess air, but slightly rich mixtures with exhaust gas dilution is a much better operating strategy!

Dept. of Aerospace & Mechanical Engineering - University of Southern California Catalytic combustion modeling n Behavior due to transition from O(s) coverage for lean mixtures (excess O 2 ) to CO(s) coverage for rich mixtures (excess fuel) Lean Rich

Dept. of Aerospace & Mechanical Engineering - University of Southern California Modeling - comparison with experiments n Predictions qualitatively consistent with experiments (propane-O 2 - N 2 ) in 2D Swiss roll (not straight tube) at low Re: sharp decrease in % fuel at limit upon crossing stoichiometric fuel:O 2 ratio n No analogous behavior seen without catalyst - only conventional rapid increase in % fuel at limit for rich fuel:O 2 ratios n Lean mixtures: % fuel at limit lower with no catalyst n Rich mixtures: opposite! n Temperatures at limit always lower with catalyst n Similar results found with methane, but minimum flame temperatures without catalyst exceed burner materials limitation! n Typical strategy to reduce flame temperature: dilute with excess air, but for catalytic combustion at low temperature, slightly rich mixtures with N 2 or exhaust gas dilution to reduce temperature is a much better operating strategy!

Dept. of Aerospace & Mechanical Engineering - University of Southern California Modeling - comparison with experiments

Dept. of Aerospace & Mechanical Engineering - University of Southern California Modeling - comparison with experiments

Dept. of Aerospace & Mechanical Engineering - University of Southern California Alternative - micro-SOFC in a Swiss roll n PI: Sossina Haile, CalTech n Solid Oxide Fuel Cells (SOFCs) can use hydrocarbon fuels directly, but need high temperatures for O = ion conduction n Need rich mixtures for in situ reforming to CO & H 2 n Swiss roll for thermal management n Catalytic after-burner with secondary air to oxidize rich products n Single Chamber Fuel Cell (SCFC) design to minimize cracking/sealing problems

Dept. of Aerospace & Mechanical Engineering - University of Southern California Thermal transpiration for microscale gas pumping n Gas pumping with no moving parts (Knudsen, 1901) n Occurs in narrow channels or pores with applied temperature gradient when Knudsen number ≈ 1 n Kn  [mean free path (≈ 50 nm for air at STP)] / [channel or pore diameter (d)] n Maximum pressure rise  P max for no mass flow (zero Mach # (M)), decreases as M increases (M generally very low) n Length of channel (L) affects M but not  P max

Dept. of Aerospace & Mechanical Engineering - University of Southern California Thermal transpiration n Maximum pumping power ~ Mach # *  P occurs at Kn ≈ 1 n  P higher at low Kn (narrow channels) but flow speed very low n (Results shown are at maximum pumping power (  P/  P max = 0.5))

Dept. of Aerospace & Mechanical Engineering - University of Southern California Microscale gas pump or jet engine with no moving parts n How to use as microscale pump or jet engine? n Vargo et al (1999): can use aerogel (nanoporous material) rather than straight channels - pore dia. (≈ 10 nm) n Aerogel has very low thermal conductivity - low thermal power requirement

Dept. of Aerospace & Mechanical Engineering - University of Southern California Aerogels n Aerogels/ Aerogels/ n Nanoporous (typ. 10 nm) materials with low density (typ. 0.1 g/cm 3 ) n Typically made by supercritical (to avoid surface tension, which would destroy the structure) drying of silica gel using CO 2 solvent n Outstanding insulator, outstanding for thermal transpiration (Kn ≈ 5 for air at STP), but generally fragile

Dept. of Aerospace & Mechanical Engineering - University of Southern California Thermal transpiration n Behavior similar to straight tubes whose length (L) is 1/10 of aerogel thickness! (See code predictions below) n Can stage pumps for higher compression ratios

Dept. of Aerospace & Mechanical Engineering - University of Southern California Microscale jet engine with no moving parts n Can use usual propulsion relations to predict performance based on Vargo et al. model n Non-dimensional performance of silica aerogel ( ≈ W.mK) only 2x worse than theoretical performance predictions for commercial gas turbine engines! n …but how to supply thermal power without electricity???

Dept. of Aerospace & Mechanical Engineering - University of Southern California Microscale gas pump or jet engine with no moving parts n TOP SECRET!

Dept. of Aerospace & Mechanical Engineering - University of Southern California Conclusions n Combustion in microscale devices (low Reynolds numbers) feasible but will likely require l Heat recirculation (e.g. Swiss roll) l Catalytic combustion l Thin walls made of low conductivity materials n Combustion behavior different from “conventional” macroscale systems n Challenges l Surface chemistry submodels l Catalyst degradation & restoration l Heat rejection - 10% efficiency means 10x more heat rejection than battery, 5% = 20x, etc. l Auxiliary components - valves, pumps, fuel tanks

Dept. of Aerospace & Mechanical Engineering - University of Southern California Conclusions - general n Study and applications of microscale thermal/chemical systems is a promising new technology in its infancy n Not as crowded as “traditional” MEMS

Dept. of Aerospace & Mechanical Engineering - University of Southern California Thanks to …… n U. S. Defense Advanced Research Projects Administration (DARPA) Microsystems Technology Office