Energy Harvesting Technologies for Wireless Sensors Andrew S Holmes Optical and Semiconductor Devices Group Department of Electrical and Electronic Engineering Imperial College London 1
Wireless Sensor Applications Wireless sensors very well established in certain market sectors e.g. domestic fuel monitoring Huge opportunity for expansion in other areas such as: Machine/process monitoring Remote monitoring - inaccessible/hostile environments Intelligent buildings - HVAC, lighting, security Medical telemetry - continuous, unobtrusive monitoring Product tracking Ubiquitous computing - ad hoc sensor networks Military surveillance - smart dust concept 1 cc wireless sensor node [IMEC] 2
Power Sources for Wireless Sensors Short term solutions inevitably based on chemical batteries High energy density (~2000 J/cm 3 or ~500 ma.hr/cm 3 at 1V) Limited life before recharging or replacement Disposal/recycling problematic Fuel-burning power sources Very high energy density Technologies still some way from maturity Limited life before refuelling, as for batteries Energy harvesting Long term storage capacity no longer an issue Low power density in most cases e.g. 100 W/cm 2 for solar cell in office environment Intermittent supply in many cases so likely to be used with battery/capacitor back-up MeOH fuel cell [Fraunhofer Inst.] Pico Radio solar cell [UC Berkeley] MEMS gas turbine stage [MIT] Vibration-driven generator 1 mw @ 0.25g rms [Perpetuum Ltd] 3
Energy Harvesting Technologies Energy Source Electromagnetic radiation Ambient light Radio waves (ambient or targeted) Heat Temperature gradients Kinetic energy Movement and vibration Volume flow (of liquids or gases) Conversion Mechanism Photovoltaic cell Antenna / Induction loop Thermoelectric device or Heat engine Electrostatic Magnetic (induction) Piezoelectric Technology of choice will depend strongly on application environment, average power and duty cycle requirements 4
Motion-driven Microgenerators 5
Inertial Energy Harvesters Single point of attachment to moving host e.g. machine, person Peak inertial force on proof mass: F = ma = m 2 Y o where a is the peak acceleration applied by the host Damper force < F or no internal movement Maximum work per transit: W Fz o = m 2 Y o z o Maximum harvested power: P = 2W/T m 3 Y o z o / z o m y=ycos( t) o damper implements energy conversion 6
How Much Power is Available? power (mw) Plot assumes: cubic device with mass occupying half of 100000 volume, the other half allowing movement 10000 1000 100 10 p1 0.1 0.01 0.001 0.01 0.1 1 10 100 1000 volume (cc) const. source acceleration amplitude ( 2 Y 0 ) of 10 m/s 2 (equiv to Y 0 = 25 cm at 1 Hz) proof mass with density 20 g/cc f = 1 Hz f = 10 Hz sensor node * watch cellphone laptop * For the sensor node, we are assuming a simple physical sensor (e.g. temp, pressure or motion) with short-range (e.g. within room) wireless link and low data-rate 7
Comparison of Architectures Normalised axes: c = c excitation frequency resonant tfrequency (resonant devices) Z l /Y 0 = mass travel range excitation amplitude c Z l /Y 0 Power = P (Watts) m 3 Y 0 2 Resonant devices better for large generators / small displacements, operated near resonance Non-resonant good for large displacements, wide input frequency ranges Mitcheson P.D., Green T.C., Yeatman E.M., Holmes A.S., Architectures for vibration-driven micropower generators, IEEE/ASME J. Microelectromechanical Systems 13(3), (2004), 429-440. 8
Machine Powered Applications Resonant vibration-driven generators aimed at machine/process monitoring are the most highly developed Synchronous electrical machines have predictable vibration frequency, making them ideal for resonant energy harvesters Several commercial offerings, e.g. PMG17 from Perpetuum Ltd Resonant generator tuned to 2 nd harmonic of mains frequency 100 or 120 Hz 55 mm diameter x 55 mm length 4.5 mw output power (rectified DC) at 0.1g acceleration 9
Human Powered Applications Excitations are slow, large in amplitude and irregular compared to those generally encountered in machine applications Non-resonant device can win at small generator sizes Data obtained in collaboration with ETH Zurich (T. von Buren) von Büren T., Mitcheson P.D., Green T.C., Yeatman E.M., Holmes A.S., Tröster G., Optimization of inertial micropower generators for human walking motion, IEEE Sensors Journal, 6(1), (2006), 28-38. 10
Non-resonant Device developed at Imperial Model: MEMS parallel plate capacitor implementation: Discharge contact on top plate Moving capacitor plate / mass Fixed capacitor plate on baseplate Pre-charging contact Generation cycle: Capacitor is pre-charged when mass is at bottom (max capacitance) Under sufficiently large (downward) frame acceleration, capacitor plates separate at constant charge, and work is done against electrostatic force stored electrostatic energy and plate voltage increase Charge is transferred (at higher voltage) to external circuit when moving plate reaches position of max displacement 11
Energy Yield per Cycle Input phase Q C input V input Separation Output phase Q C V output output V ouput C C input output V input Generated energy: E 1 C 2 output V 2 ouput 1 C 2 input V 2 input 1 C 2 output V output ( V output V input ) V 1 C 2 output output V input V 2 output 12
Measured Performance generator shaker Vlt Voltage probe has input timpedance >10 12 and dynamically measures voltage on capacitor Net power in this experiment: 2.2 μw voltage probe 13
Motion-driven Harvesters are they any good? Volume Figure of Merit defined as: Useful Power Output FoM V = 1 16 Au Vol 4/3 Y 0 3 Represents ratio of output power to that of idealised generators on slide 7 As of 2008 best devices achieved only about 2% Better devices have emerged since, but there is still a way to go... FoM V 1.8% 1.6% 1.4% 1.2% 1% 0.8% 0.6% 0.4% 0.2% 0 EM ES PZ 2000 2002 2004 Publication Year 2006 2008 Main issues are: (1) damping/transduction need to implement stronger dampers; (2) power conversion electronics difficult to make efficient; (3) adaptive operation Mitcheson P.D., Yeatman E.M., Kondala Rao G., Holmes A.S., Green T.C., Energy harvesting from human and machine motion for wireless electronic devices, Proc. IEEE 96(9), ),(2008), 1457-1486. 14
Flow-driven Microgenerators 15
Basic concept: wind turbines on a smaller scale (cm-scale or smaller) Et Extract tkinetic energy from air flow Energy Scavenging from Air Flow K.E. per unit vol in flow = ½ V 2 K.E. per sec crossing swept area is: 100000 10000 1000 = 2 = 3 100 P avail ½ V xav ½ AV Actual output power is: P = ½ AV 3 C P 0.01 (mw) Output power 10 1 0.1 0.001 where C P = power coefficient 0.0001 For 1 cm-dia disc: Betz limit (C P = 0.59) C P = 0.1 Flight vehicle Land vehicle HVAC duct 0.1 1 10 100 1000 Flow speed (m/sec) 16
2-cm dia. Device developed at Imperial Ducted turbine with integrated axial-flux permanent magnet generator mw output power levels Starts at low flow speeds (~3 m/s) Applications in HVAC duct sensing and gas pipeline monitoring 7 er (mw) Generato or output pow 6 5 4 3 2 1 0 8.0 m/s Tunnel speed 10.0 m/s 9.0 m/s 6.0 m/s 7.0 m/s 0 1000 2000 3000 4000 5000 6000 Rotation speed (RPM) 17
Comparison with other Flow-driven Harvesters Small flow-driven devices are expected to perform relatively poorly because of high viscous losses Small turbines also suffer from relatively large clearances and bearing losses Cm-scale prototype t devices to date have struggled to reach Cp ~ 0.1 Nevertheless, useful (mw) power levels can be generated because available power in flow is significant even at modest flow speeds Duct sensing applications look quite viable even with current devices Power density (mw W/cm^2) 10000 1000 100 10 1 0.1 0.01 0.001 Betz limit Cp = 0.1 Federspiel (2003), A = 81 sq.cm Rancourt (2007), A = 13.9 sq.cm Myers (2007), A ~ 317 sq.cm Holmes (2009), A = 3.14 sq.cm 0.0001 0.1 1 10 100 Flow speed [m/sec] Bansal A., Howey D.A., Holmes A.S., Cm-scale air turbine and generator for energy scavenging from low-speed flows, Proc. Transducers 2009, Denver, Colorado, USA, 21-2525 June 2009, pp. 529-532. 532 18
HVAC Duct Sensor Concept Spider mounted inside id duct Distributed network of wireless sensors with peer-to-peer communication to relay data to control centre Monitoring of: Air flow and temp for HVAC control Air-quality e.g. RH; CO 2, Ammonia, VOCs Sensor array Generator / Transceiver 19
Summary Motion-driven i energy harvesters are still performing at a level l some way below what is theoretically achievable Current performance is adequate for some important applications such as machine monitoring, and commercial solutions are available Improvements in performance will be required before harvesting power from human body motion can become viable Flow-driven devices at cm-scale also have relatively low conversion efficiencies, but the available power in the flow is such that duct sensing applications appear viable
Acknowledgements Motion-driven Generators: Eric Yeatman Paul Mitcheson Tim Green Peng Miao (now with Oxford Instruments) Bernard Stark (now with University of Bristol) Flow-driven Generators: Keith Pullen (now with City University, London) Guodong Hong (now with Microsaic Systems plc) Anshu Bansal David Howey
Contact Andrew S Holmes Professor of Micro Electro Mechanical Systems Optical and Semiconductor Devices Group Department of Electrical and Electronic Engineering Imperial College London Exhibition Road, London SW7 2BT, UK Tel: +44 (0)20 7594 6239 Fax: + 44 (0)20 7594 6308 Email : a.holmes@imperial.ac.uk Web: http://www3.imperial.ac.uk/opticalandsemidev