Nextreme Whitepaper Design Considerations for TEG System Optimization NWP003.1

7/29/2019 Nextreme Whitepaper Design Considerations for TEG System Optimization NWP003.1

http://slidepdf.com/reader/full/nextreme-whitepaper-design-considerations-for-teg-system-optimization-nwp0031 1/14

3908 PATRIOT DR., SUITE 140 DURHAM, NC 27703-8031 919-597-7300 919-597-7301 www.nextreme.com

Nextreme Thermal Solutions, Inc.

Nextreme Thermal Solutions, Inc.Whitepaper

Design Considerations for TEG System Optimization

E. Siivola, R. Mahadevan, P. Crocco, K. von Gunten and D. Koester.Nextreme Thermal Solutions, Inc.



Design Considerations for TEG System Optimization Whitepaper – NWP003.1 2

Introduction

Advances in distributed sensors and sensor networks have led to an increased interestin the use of continuous power sources to replace or augment existing power storagesystems. The use of waste heat is an attractive source of energy for many applicationswhere W-mW power is required. The implementation of a thermoelectric power conversion and energy storage system requires several basic elements in addition to anassumed heat source and electrical load. These elements shown in schematic form inFigure 1 are: 1) a thermoelectric device, 2) a heat sink, 3) a power conditioning circuit, 4)an energy storage device and 5) a power management circuit. The design andoptimization of the system and elements is highly dependent on the thermal boundaryconditions. This whitepaper describes the design issues that must be addressed whendesigning an optimized system for harvesting electricity from waste heat and storing andusing that energy.

Heat

Source

Heat

Source

Charge

Mgmt.(3)

Charge

Mgmt.(3)

BoostConv. &

V. Reg.(3)

BoostConv. &

V. Reg.(3)

LoadMgmt.

(5)

LoadMgmt.

(5)

TEG(1)

TEG(1)

Heat

Reject(2)

Heat

Reject(2)

Battery +

Fault

Protection(4)

Battery +

Fault

Protection

(4)

Cap(4)

Cap(4)

LoadLoad

Heat

Source

Heat

Source

Charge

Mgmt.(3)

Charge

Mgmt.(3)

BoostConv. &

V. Reg.(3)

BoostConv. &

V. Reg.(3)

LoadMgmt.

(5)

LoadMgmt.

(5)

TEG(1)

TEG(1)

Heat

Reject(2)

Heat

Reject(2)

Battery +

Fault

Protection(4)

Battery +

Fault

Protection

(4)

Cap(4)

Cap(4)

LoadLoad

Figure 1: Generic schematic of thermoelectric energy harvesting system.




Thermoelectric Generator (TEG)

The direct conversion of heat energy into electrical

energy can be accomplished through the Seebeck effectin which an induced heat flow through an appropriatelydesigned thermoelectric device produces a voltage andcurrent. Figure 2 shows a single PN couple--the basicbuilding block of a TEG device. The PN couple consistsof a single pellet of P-type and N-type thermoelectricmaterial each that are connected electrically in series.Heat carries the majority carriers from one junction to theother producing a current and voltage. By placing manyPN couples in series electrically and in parallel thermally,a typical TE module can be constructed that generates avoltage proportional to the temperature differential

across the elements (Eq. 1). The maximum power generated by the TEG (when it is matched to theelectrical load) is proportional to the open-circuit voltagesquared (Eq. 2) and also the square of the temperaturedifferential across the elements ( ∆T ). Since the ∆T across the TEG module is related to the heat flow Q through the module and the thermalconductance of the module, Eq. 2 can be written in terms of these parameters as shownin Eq. 3. Figure 3 shows a typical IV curve for a TEG along with the power output. It isevident from this plot that the maximum power output occurs at one half of the open-circuit voltage (½V oc )and one half of the short-circuit current (½I sc ) of the TEG.

kAQL TNV

couple

coupleoc2

Eq. 1

module

222

module

2

44 R

TN

R

VP

coupleocout

Eq. 2

2

22

32NAk

LQP

couple

out

Where:N=number of PN couplescouple=Seebeck potential of the PN couple (V/K) ∆T=temperature difference across PN coupleQ=heat flux through coupleL=length of P and N elements

A=cross sectional area of TE elementk=thermal conductivity of TE elementρ=electrical resistivity of TE elementRmodule=electrical resistance of module

Eq. 3

Heat PumpHeating or Cooling

Heat rejected C u r r ent

Heat absorbed

Hot Junction

Cold Junction

n-type p-type

Electrical Power Input

Figure 2: PN couple operating in Seebeck (power generation) mode.




0.00

100.00

200.00

300.00

400.00

500.00

600.00

0

5000

10000

15000

20000

Voltage(V)

Power(W)

Current (A)

Power (W)

Voltage (V)

Isc/2 Isc

Voc

Voc/2

Pmax

Figure 3: Generic IV and power curve for a TEG. Note that maximum power occurs at 1/2Voc and 1/2Isc.

Two attributes of TEGs at a ∆T (delta temperature)_of 30 K are plotted in Figure 4 for athree different types of TEG devices—1) thin-film, 2) bulk and 3) micro-bulk. Theattributes plotted are sensitivity (Voc/Q) which measures how much voltage is producedby the TEG per watt flowing through it, and power density (P/A) which is the amount of

power generated per unit area (footprint) of the TEG. It can be seen here that thin-filmdevices offer the potential for higher sensitivities and higher power densities thanstandard or micro-bulk devices.




Figure 4: Power Density and Sensitivity plotted for a variety of TEGs at a ∆ T=30K.

The conversion efficiency of a TE device is closely related to the Carnot efficiency of thesystem (Eq. 4), but is adjusted for the intrinsic performance of the TE device itself.

Equation 5 describes the efficiency of the TEG as a function of the Carnot efficiency andthe figure-of-merit, ZT, of the TEG. The figure-of-merit of the device is described inEquation 6. The Carnot efficiency represents the upper bound in conversion efficiency.

h T

T

Carnot Eq. 4

h

c TEG

T

TZT

ZT

11

11Carnot

Eq. 5

RK

TSZT

2

Where:T=mean temperature (absolute)R=module resistanceK=module thermal conductance

Eq. 6

P o w e r D e n s i t y ( W / c m 2 )

Sensitivity (V/W)

Performance Comparison of Various TEG Technologies

Thin-film

Standard Bulk

Micro Bulk

P o w e r D e n s i t y ( W / c m 2 )

Sensitivity (V/W)

Performance Comparison of Various TEG Technologies

Thin-film

Standard Bulk

Micro Bulk




Heat Sinking

In order to sustain a ∆T across a TEG, there must be a constant heat flow through the

TEG. This requires a continuous heat source as well as a heat sink of some typecapable of rejecting the heat passing through the TEG. Heat rejection can beaccomplished in numerous ways--through direct conduction to a large thermal mass, byliquid cooling, spray cooling, phase change or simply through convection to the air usinga traditional heat sink which is the most common and often the lowest cost of theseapproaches. Heat sinks come in a vast array of sizes, dimensions and materials andchoosing the proper one requires a thorough understanding of several key parameters(Fig. 5) [1,2]. Each of these is described in more detail below.

Figure 5: Various heat sink sizes and configurations.

This ability of a heat sink to dissipate heat is often reduced to a single parameter calledthermal resistance, Θ. Thermal resistance, with units of degrees of temperature/heatflow (e.g. °C/watt), is the thermal analog to an electrical resistor. A low thermalresistance means the sink will do a good job of rejecting heat to the environment while ahigh thermal resistance will do a poor job.




Airflow speed The speed of the airflow across the fins or pins of a heat sink play a profound role in theability of the sink to dissipate heat to the environment. Increasing air speed has a non-

linear affect on thermal resistance as shown in Figure 6. Consequently, air speed (andthe use of a fan) has a diminishing effect on heat sink performance and simply using alarger fan to improve thermal rejection can be a losing proposition. Conversely, even asmall amount of air flow across a sink can substantially improve its performance relativeto a case with no forced airflow (known as natural convection). Forced convectivecooling is usually achieved by using a fan. In the case of a stand- alone TEG system, theTEG and thermal system would need to be capable of starting and driving the fan under worst case, natural convection conditions. In addition, the use of a fan will impact theoverall conversion efficiency of the TEG system.

T her m

al R esi st anc e

Air Flow

Heat-sink PerformanceDependence on Air Flow

No airflow = “natural convection”

Figure 6: Thermal resistance of a heat sink shows a non-linear response as a functionof air speed. The special case of no air flow is referred to as “natural convection”.




SizeSize and geometry (such as aspect ratio), like airflow, have a large impact on theperformance of a heat sink [3]. As shown in Figure 7, the performance of a heat sink

improves with size.

Figure 7: Heat sink performance versus volume under natural and forced convection at a fixed condition. Forced convection case shows lines for different air flow rates.




Fin type, dimension and spacing The optimal fin dimension and spacing depends on the fin material, mode of operation of the heat sink (natural or forced convection) and desired operating condition. The fin

width is generally a tradeoff between fin length, fin efficiency, and volumetric efficiency of the heat sink. For natural convection between vertical parallel plates, the optimal spacingbetween plate fins is determined by fin length, H, and operating fin temperature riseabove ambient [4]. For a forced convection heat sink, the optimal fin geometry is usuallydetermined by the flow velocity and pressure drop constraints imposed by the fanspecifications (see [2,5]) and are often improved by the use of pin fins or elliptical fins.

Figure 8: Typical heat transfer coefficient, h0 , as a function of fin spacing, S for a fixed fin height, H.

Thermal spreading resistance in the sink A large mismatch in footprint area between the heat source/TEG and base of the heatsink can lead to additional spreading resistance in the base of the heat sink as the heatas to spread from the smaller source area, through the heat sink base, to the farthestfins on the heat sink [6]. Performance is often improved by usual a more thermallyconductive material (such as copper) &/or varying base plate thickness. Highperformance heat sinks often use heat pipes in the base plate to distribute the heateffectively to the fins.




Orientation of the heat sink (relative to gravity)The performance of a heat sink in natural convection mode (i.e., no forced air movement) depends on the orientation of the fins relative to gravitational direction [7,8]

as shown in Figure 9. Many heat sinks are most effective in natural convection with their fins and base plate vertical; i.e., aligned with gravity (referred to as vertical/verticalorientation). However, heat sinks are also often used on horizontally oriented PCBs withthe base plate horizontal and the fins oriented vertically--referred to as horizontal/verticalconfiguration. In most cases, the heat sink works almost as well in this horizontal/verticalorientation. However, the heat sink performance is considerably poorer in the verticalbase plate, horizontal fins orientation (referred to as vertical/horizontal). Thedependence of performance on orientation can be reduced somewhat by the use of pinfins rather than plate fins in the heat sink. Different pin fin geometry ratios also impactthe orientation dependence as shown below.

Figure 9: Heat sink and fin orientations. Included from left to right are thevertical/vertical, horizontal/vertical and vertical/horizontal configurations which represent good to poor performance in natural convection. (Source: thermalsoft.com)

Voltage Regulation and Power Management Most electronics require more than a volt to operate and many require >3V. Since thevoltage output of a TEG is proportional to the temperature differential ∆T across theTEG, large values of ∆T are desired for most applications. Unfortunately, the ∆Tavailable to many applications is fairly small meaning that the output voltage of thedevice may not exceed the minimum requirement. In such a case, it is necessary toboost the voltage to the minimum usable value. While this may seem like astraightforward and benign task, it should be noted voltage conversion comes at a costsince the converter is not 100% efficient. It is common for designers to focus only on the




power output of the TEG and miss the issue of voltage or vice versa. There are manyways to view power management but we will discuss three basic categories as theyrelate to TEG power management.

Direct to Load In some applications, where the heat flow in the system is consistent and adequate tosustain an adequate ∆T, the output voltage and current of the TEG is adequate to drivethe load directly. This can substantially reduce the complexity of the electronics designmy essentially eliminating any power management. However, as with most heatsources, temperatures and heat flow can vary as can the ambient temperature leading tofluctuations in output. These fluctuations must be tolerated by the load and often somekind of conditioning and/or energy storage is necessary to ensure uninterruptedoperation.

Boost Converters (>400 mV)Many of today’s electronics require a minimum of 1.8V to 3V to operate but often times

the heat source and TEGs combination is unable to generate such a voltage.Fortunately, there are numerous commercially available boost converters that canconvert voltage from as low as 400 mV to >3V. Figure 10 shows an example of conversion efficiency as a function of input voltage for a 3.3 V output for a TI TPS61201converter. With an input voltage of 1V or better, the efficiency of this device easilyexceeds 60% and is greater than 90% under some conditions. As with most converters,efficiency is highly dependent on the voltage and input current. One notable advantageto using a boost converter is that it provides a constant output voltage unlike the “directto load” approach.

Figure 10: Conversion efficiency for a 3.3 V output as a function of input voltage at threedifferent output currents (Source: Datasheet for Texas Instruments TPS61201).




Micro-Power Up Converters (>30 mV)Recent advancements in up-conversion now make it possible to up-convert voltages aslow as 30 mV. This is particularly important in cases where the ∆T available from the

heat sources is very low. One example of this is generating power from the human bodywere reasonable and achievable ∆Ts are in the range of only 5-10 K. For suchapplications, the power being generated is usually quite low and this must be weighedwith the relatively low conversion efficiencies associated with this type of converter.Figure 11 shows the conversion efficiency of a Linear Technologies LTC-3108 converter.In this case the efficiency ranges from 5-40%.

Figure 11: Conversion efficiency and output current as a function of input voltage. Thisconvertor provides high output voltages for relatively low input voltages. The tradeoff for converting such low input voltage is conversion efficiency (source: Linear Technologiesdatasheet LTC-3108).




Optimization Example

Many potential applications in waste heat management and energy harvesting have heat

sources that are not constant or that vary over time.

To employ thermoelectric energy harvesting in this scenario, several key aspects needto be considered:

Time-varying heat source. How to design the system to work with a heat sourcethat is not constant and sometimes simply not available.

Providing power when heat source is inadequate or unavailable. Energy storage. Be able to store energy in order to support the application when

no heat is available.

To illustrate how to manage these design considerations as well as the concepts found

earlier in this whitepaper, we built a example wireless sensor application that includesthermal energy (from sunlight) as the heat source, a thermistor to measure thetemperature of the window frame, an eTEG HV56 Thin-film Thermoelectric Generator asthe power source, a heat sink for heat rejection, a voltage up-converter and a wirelesstransmitter as the load. An energy storage device (capacitor or battery) was notincluded in this demonstration.

The TEG and heat sink assembly was mounted to a window frame that was subjected tosunlight during normal office hours. During daylight hours, the TEG supplied > 40 mV tothe up-converter, which in turn produced 3.3V and approximately 37µW for use by thewireless sensor transmitter. The sensor was programmed to transmit the temperatureand supply voltage to the receiver installed in a laptop computer equipped with a data

logger software application.

Figure 12 shows the temperature and output voltage data obtained over half a day. Theplot illustrates the need to incorporate an energy storage device in the overall solution toaccommodate the variability of the heat source due to weather conditions and nighttimeuse.



Figure 12: Temperature and output voltage during a single day

Energy storage will often be needed during periods where continued operation of theload is required while thermal energy is unavailable.

References

[1] Seri Lee, "How to Select a Heat Sink," Electronics Cooling , June 1995.

http://www.electronics-cooling.com/1995/06/how-to-select-a-heat-sink/[2] Seri Lee, "Optimum Design and Selection of Heat Sinks," Proc. Eleventh IEEE SEMI-THERM Symposium, 1995.[3] "An Introduction to Heatsinks and Cooling," Wakefield Engineering Inc.,http://robots.freehostia.com/Heatsinks/Heatsinks.html[4] Åke Mälhammar, "A Volumetric Approach to Natural Convection," Thermal Design for Electronics, 2003, http://www.frigprim.com/articels2/parallel_pl.html[5] "Optimum Fin Spacing for Fan Cooled Heat Sinks," Thermal Solutions Inc.http://www.thermalsoftware.com/optimum_sink_fan.pdf [6] Seri Lee, et al., "Constriction / Spreading Resistance Model for ElectronicsPackaging," Proc. ASME/JSME Thermal Engineering Conference, Vol 4, ASME 1995.[6] "Comparing Naturally Cooled Horizontal Baseplate Heat Sinks with Vertical

Baseplate Heat Sinks," Thermal Solutions Inc.http://www.thermalsoftware.com/vert_vs_horz_sink.pdf [7] Ren-Tsung Huang etal., "Orientation Effect on Natural Convective Performance of Square Pin Fin Heat Sinks," Intl. Journal of Heat and Mass Transfer 51 (2008), pp. 2368-2376.

Nextreme Whitepaper Design Considerations for TEG System Optimization NWP003.1

Documents

Transcript of Nextreme Whitepaper Design Considerations for TEG System Optimization NWP003.1