460 likes | 584 Views
A Quantitative Investigation of Inertial Power Harvesting for Human-powered Devices. † School of Interactive Computing & GVU Center College of Computing Georgia Institute of Technology Atlanta, GA 30332 USA { jaeseok,abowd }@ cc.gatech.edu. ‡ Computer Science & Engineering
E N D
A Quantitative Investigation of Inertial Power Harvestingfor Human-powered Devices † School of Interactive Computing & GVU Center College of Computing Georgia Institute of Technology Atlanta, GA 30332 USA {jaeseok,abowd}@cc.gatech.edu ‡ Computer Science & Engineering Electrical Engineering University of Washington Seattle, WA 98195 USA shwetak@cs.washington.edu † School of Interactive Computing & GVU Center College of Computing Georgia Institute of Technology Atlanta, GA 30332 USA {jaeseok,abowd}@cc.gatech.edu § Electrical & Computer Engineering Pratt School of Engineering Duke University Durham, NC 27708 USA matt.reynolds@duke.edu JaeseokYun ShwetakPatel MattReynoldsGregory Abowd Presented By: Lulwah Alkwai
Outline • Introduction • Related Work • Understanding of wearable devices • Power Harvester Model • Data Collection and analysis • Experimental Setup • Processing of Acceleration Signals • Data Analysis in Frequency Domain • Simulation and result • Simulation Setup • Power Estimation Procedure • Optimization and Results • Discussion • Limitation and Future Work • Conclusions
Outline • Introduction • Related Work • Understanding of wearable devices • Power Harvester Model • Data Collection and analysis • Experimental Setup • Processing of Acceleration Signals • Data Analysis in Frequency Domain • Simulation and result • Simulation Setup • Power Estimation Procedure • Optimization and Results • Discussion • Limitation and Future Work • Conclusions
Introduction • Power remains a key to unlocking the potential for a sustainable UBIcompreality • Power harvesting from natural and renewable sources is a longstanding research area • Little research into capturing energy from daily human activity • Prior research done in laboratory settings of selected activities • Possibility of powering consumer electronics or self sustaining body sensor networks
Introduction • First principles numerical model • Developed with MATLAB Simulink • Using the most common type of inertial power harvester: Velocity Damped Resonant Generator (VDRG) • Estimate of achievable performance • Available power to devices based on the development model • Used reasonable assumption about the size of the generator that could fit in the devices to develop the model
Outline • Introduction • Related Work • Understanding of wearable devices • Power Harvester Model • Data Collection and analysis • Experimental Setup • Processing of Acceleration Signals • Data Analysis in Frequency Domain • Simulation and result • Simulation Setup • Power Estimation Procedure • Optimization and Results • Discussion • Limitation and Future Work • Conclusions
Related Work • Paradiso and Starner: • Extensive survey of the available energy sources to power mobile devices • Amirtharajah: • Generating power using a model human walking as a vibration source • Mitcheson: • Presented architectures for vibration driven micro power generators • Buren: • Measured acceleration from 9 location on body of human objects and estimated the maximum output power. • The acceleration signals measured from standard walking motion on treadmill • Rome: • A vertical excursion of a load during walking and climbing • Kuo: • A knee brace generator that produced electricity • Troster: • A button sized solar powered node • Possibility of energy harvesting through ordinary exposure to sunlight and indoor light
Outline • Introduction • Related Work • Understanding of wearable devices • Power Harvester Model • Data Collection and analysis • Experimental Setup • Processing of Acceleration Signals • Data Analysis in Frequency Domain • Simulation and result • Simulation Setup • Power Estimation Procedure • Optimization and Results • Discussion • Limitation and Future Work • Conclusions
Understanding of Wearable Devices • Powering devices such as consumer electronics and self sustaining body sensor networks • Alternative to batteries • Monitoring human vital signs • Main result of the paper: • Whether the power garnered from daily human motion can practically power the low power components for wearable electronics
The required power for all electronics Understanding of Wearable Devices
Outline • Introduction • Related Work • Understanding of wearable devices • Power Harvester Model • Data Collection and analysis • Experimental Setup • Processing of Acceleration Signals • Data Analysis in Frequency Domain • Simulation and result • Simulation Setup • Power Estimation Procedure • Optimization and Results • Discussion • Limitation and Future Work • Conclusions
Power Harvester Model • Inertial Generator Model • Works by the body acceleration imparting forces on a proof mass • Three Categories of Inertial Generators • VDRG(Velocity Damped Resonant Generator) • CDRG(Coulomb Damped Resonant Generator) • CFPG(Coulomb Force Parametric Generator) (VDRG is used since the internal displacement travel of a generator exceeds 0.5 mm)
Power Harvester Model • Vibration driven generators represented as a Damped Mass Spring System: • m: proof mass • K: spring constant • D: damping coefficient • y(t): displacement of the generator • z(t): displacement between the proof mass and the generator • t: time
Power Harvester Model Model and operating principle for the VDRG In this Damped mass spring system, the electrical energy generated is represented as the energy dissipated in the mechanical damper
Power Harvester Model • Important parameters during simulation analysis: • m: proof mass • K: spring constant • D: damping coefficient • Zmax: internal travel limit (m and Zmax are limited by size and mass of the object that holds the generator)
Outline • Introduction • Related Work • Understanding of wearable devices • Power Harvester Model • Data Collection and analysis • Experimental Setup • Processing of Acceleration Signals • Data Analysis in Frequency Domain • Simulation and result • Simulation Setup • Power Estimation Procedure • Optimization and Results • Discussion • Limitation and Future Work • Conclusions
1-Experimental Setup • Wearable Data Collection Unit • Two Logomatic Serial SD data loggers • Sampling Rate of 80 Hz • Record each input as a time series file • Six 3 axis Accelerometer • Accelerometer packaged in small container sealed against dust or sweat • Contained in small waist pack • Allow 24 hours of continuous operation
1-Experimental Setup • 4 men, 4 women • 3 days (2 weekdays, 1 weekend) • Participants can take it off when needed ( sleep, shower) • Participants are asked to record their activates and time on diary sheet
2-Processing of Acceleration Signals • High pass filtered measure acceleration signals • 0.05 Hz cutoff frequency • Obtained displacement of the accelerometer through double integrating the acceleration dataset • Feed the resulting displacement time series into the VDRG model
3-Data Analysis Frequency Domain Lower body experience much more acceleration than upper body-> more electrical energy converted from kinetic energy than upper body
3-Data Analysis Frequency Domain Upper body especially the wrist contain energy at low frequencies -> Because has a higher degree of freedom when moving than the lower body
3-Data Analysis Frequency Domain Largest peak in each spectrum occurs at approximately 1 or 2 Hz
3-Data Analysis Frequency Domain The determine the dominant frequency: Top 20 frequency components ranked from highest to lowest from each spectrum
Outline • Introduction • Related Work • Understanding of wearable devices • Power Harvester Model • Data Collection and analysis • Experimental Setup • Processing of Acceleration Signals • Data Analysis in Frequency Domain • Simulation and result • Simulation Setup • Power Estimation Procedure • Optimization and Results • Discussion • Limitation and Future Work • Conclusions
1-Simulation Setup No useful digital object mountable on knee 2 g / 4.2 cm 36 g / 10 cm 100 g / 20 cm Proof mass m and internal travel length Zmax
2-Power Estimation Procedure • Acceleration data split into 10 sec fragments: Ed: the energy dissipated in the damper F=Dzdamping force Z1, Z2: start and end position The equation can be expanded as follows The average power during time interval T=10 sec VDRG built with a single axes with one of the axes of accelerometers. The preferred orientation is chosen on max output power
2-Power Estimation Procedure The following procedure was developed for using the measured acceleration dataset, in combination with the VDRG model to estimate available power
3-Optimization and Result • Search for optimal D which maximums P • All other variables determined previously • After D is chosen the generated electrical power can be estimated
3-Optimization and Result • Typical Efficiency for Mechanical to electrical conversion is %20 • Energy Generated is storable • The direction of the power harvester is continuously aligned with the axis that generates the maximum output • Average Electrical Power Expected:
Outline • Introduction • Related Work • Understanding of wearable devices • Power Harvester Model • Data Collection and analysis • Experimental Setup • Processing of Acceleration Signals • Data Analysis in Frequency Domain • Simulation and result • Simulation Setup • Power Estimation Procedure • Optimization and Results • Discussion • Limitation and Future Work • Conclusions
Discussion • The ratio of the electrical output power of the harvester to the required power for wearable electronics: • a watch hanging on the neck • a watch on the wrist • a phone hanging on the neck • a phone on the waist • a phone on the arm • a shoe on the ankle • The X-index represents the subjects (8 subjects * 3 days = 24 days) • The Y-index represents the wearable electronics
Discussion • Output power is insufficient to continuously run the higher demanding electronics (such as MP3 decoder chip): • Can be used to charge a battery for intermittent operation • Possibly charge a backup battery for when standard recharging options not available • Low power electronics (such as the wristwatches) can be powered continuously
Discussion • From the diary sheet of participants it is possible to find the generated power from certain activities
Discussion Phone scale harvester will generate 15 m W while running -> GPS chip can be powered
Discussion Shoe harvester will generate 20 m W while running -> GPS chip can be powered
Discussion Continuous availability of more than 60 μ W-> Large number of activates taking place
Outline • Introduction • Related Work • Understanding of wearable devices • Power Harvester Model • Data Collection and analysis • Experimental Setup • Processing of Acceleration Signals • Data Analysis in Frequency Domain • Simulation and result • Simulation Setup • Power Estimation Procedure • Optimization and Results • Discussion • Limitation and Future Work • Conclusions
Limitation and Future Work • Possible to increase power output with heavier proof mass • Balanced with increased strain from additional weight • Use all three axis to generate power • Harvester with three mass spring systems • Generalize the K/D measurement across subjects • Adaptive Tuning of VDRG
Outline • Introduction • Related Work • Understanding of wearable devices • Power Harvester Model • Data Collection and analysis • Experimental Setup • Processing of Acceleration Signals • Data Analysis in Frequency Domain • Simulation and result • Simulation Setup • Power Estimation Procedure • Optimization and Results • Discussion • Limitation and Future Work • Conclusions
Conclusion • First 24 hour continuous study of inertial power harvester performance • Analysis of the energy that can be garnered from 6 locations on the body • Shown feasibility to continuously operate motion powered wireless health sensor • Motion generated power can intermittently power devices such as MP3 players or cell phones