Enhancing Heat Recovery for Thermoelectric Devices Jim Szybist and Jim Parks Oak Ridge National Laboratory Essam Ibrahim Alabama A&M University Norman Love University of Texas El Paso September 29, 2010
Enhancing Heat Recovery for Thermoelectric Devices
Jim Szybist and Jim ParksOak Ridge National Laboratory
Essam IbrahimAlabama A&M University
Norman LoveUniversity of Texas El Paso
September 29, 2010
2 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Motivation for exhaust heat recovery efforts
0
200
400
600
Exhaust T (C)
0
20
40
60
80
Spee
d (m
ph)
0
20
40
60
80
0 200 400 600 800 1000 1200 1400
Fuel
Ene
rgy
(MJ)
Time (s)
Exhaust Losses
Brake Work
Coolant, Friction, and Other
1st Law Fuel Energy Distribution
2nd Law Fuel Exergy Distribution
Brake Work10.4%
Exhaust27.7%
Friction, Coolant, and Other
61.9%
Brake Work9.7%
ExhaustExergy8.4%
Irreversibilities, Friction, Coolant, and Other
81.9%
Federal test protocol (FTP) test cycle for a 2007 Saab Biopowershowing speed, Experimental data were collected at the ORNL chassis dynamometer facility.
Availability of energy in exhaustis nearly as high as brake work
3 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Basic heat recovery system for thermoelectrics
• Electrical power generation dependent on temperature gradient across thermoelectric device– Thermoelectric temperature gradient is smaller than temperature gradient
from bulk hot gas to bulk coolant
• Efficiency of thermoelectric heat recover system dependent on both heat exchanger and thermoelectric efficiency
Thermoelectric Generator
QTE_out
QTE_in
QHX_wall
QExhaustExhaust Flow
Exhaust Flow
Exhaust Duct
TE THOT
TE TCOLD
PElectricity
tricThermoelecExchangerHeatExhaust
yElectricitSystem Q
Pηηη *_==
4 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
ORNL Experimental Thermoelectric Apparatus
• Simulate exhaust flow with mass flow controller and intake air heater
• Two rectangular ducts for the hot and cold flows
• Each thermoelectric loaded with 5 Ohm power resistor– Voltage measured to record
power
• Commercially available thermoelectric power generators from Marlow Industries– Thermoelectric material: Bi2Te3
– ZT = 0.73
Gas duct
Gas outlet Surface thermocouples
Coolant Outlet
Thermoelectric element Coolant
inlet Flow thermocouples
Coolant duct
Insulation
Insulation Gas inlet
Gas Inlet
Gas OutletCoolant Inlet
Coolant Outlet
5 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Experiments performed by visiting faculty during summer of 2009 and 2010
• Year 1 study– Parametric study of operating conditions
• Simulated exhaust temperature• Simulated exhaust flow • Coolant temperature
– Packed vs. unpacked duct
• Year 2 study– Multiport heat exchangers– Parametric study of operating conditions– Aluminum vs. stainless steel– Fouled vs. un-fouled heat exchangers
Professor Essam IbrahimAlabama A&M University
Professor Norman LoveUniversity of Texas El Paso
6 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Year 1 Study• Parametric investigation of operating conditions on thermoelectric performance
– Simulated exhaust temperature and flow rate– Coolant temperature
• Enhancement of thermoelectric performance using duct packed with aluminum wool
• Journal article contains complete detailsIbrahim, E.A., Szybist, J.P., Parks, J.E. Enhancement of automotive exhaust heat recovery by thermoelectric devices. Proc. IMechE, Part D: J. Automobile Engineering, 2010, 224(D8), 1097-1111. DOI 10.1243/09544070JAUTO1438
Gas InletGas Outlet
Coolant Inlet
Coolant Outlet
7 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Basic operating characteristics for hollow rectangular duct
• Hot-side bulk temperature and surface temperature decreases
• Temperature drop from bulk gas temperature to hot-side thermoelectric temperature is very significant– > 100 C at most conditions, more than half the total temperature gradient
• Surface temperature gradient dictates thermoelectric power generation
0
50
100
150
200
250
300
350
050100150200250300350
Tem
pera
ture
(o C)
Flow Thermocouple Position (mm from inlet)
Gas Flow
GA
S IN
LET
0
50
100
150
200
050100150200250300
Tem
pera
ture
(o C)
Surface Thermocouple Position (mm from inlet)
Gas Flow
GA
S IN
LET
8 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Heat transfer to the thermoelectric is a major impediment to greater exhaust heat recovery
0
100
200
300
400
500
600
Sim Exhaust TE Hot Side TE Cold Side Coolant
Tem
pera
ture
(C)
Sim
ulat
ed
Exha
ust F
low
Coo
lant
Flo
w
Ther
moe
lect
ricG
ener
ator
Q In Waste Q
Elec
tric
al
Pow
er O
ut
TEG∆T
Aluminum Walls
9 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Packed duct increases heat recovery at all flow ratesLargest percent increase in thermoelectric power occurs at lowest flow
• Hypothesis is that the packing material increased heat transfer by increasing conductive heat transfer to walls and reducing boundary layer effects
• Added backpressure is a concern, but packing material filled only 2.5% of duct interior volume– No backpressure increase was measured in this experiment
• Other published attempts to increase heat transfer, such as fins and diffusers, are typically more effective at highest flows
0
1
2
3
4
5
275 300 325 350 375 400 425 450
Ther
moe
lect
ric P
ower
(W)
Air Temperature (C)
40 SLMP Air Flow63% Improvement
Unpacked Duct
Packed Duct
0
1
2
3
4
5
275 300 325 350 375 400 425 450Th
erm
oele
ctric
Pow
er (W
)
Air Temperature (C)
60 SLMP Air Flow40% Improvement
Unpacked Duct
Packed Duct
0
1
2
3
4
5
275 300 325 350 375 400 425 450
Ther
moe
lect
ric P
ower
(W)
Air Temperature (C)
80 SLMP Air Flow32% Improvement
Unpacked Duct
Packed Duct
10 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Additional year 1 results and conclusions
• Thermoelectric power increased with an increased ΔT across thermoelectric device– Increase in hot-side temperature
• Higher temperature of simulated exhaust• Higher simulated exhaust flow rate• Packed duct rather than hollow duct
– Decrease in cold-side temperature• Practical constraints for a dedicated cooling system for thermoelectrics
• Maximum system efficiency was low, less than 1%– Literature survey shows that this is comparable to thermoelectric system efficiencies during
vehicle demonstrations
tricThermoelecExchangerHeatExhaust
yElectricitSystem Q
Pηηη *_==
11 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Year 2 study: Investigate the effect of heat exchanger material and heat exchanger fouling on heat transfer, thermoelectric performance• Use multi-port heat exchanger design
– Similar to heat exchangers in EGR coolers
• Aluminum and stainless steel heat exchangers– Aluminum k ~ 180 W/m-K– Stainless k ~ 20 W/m-K
• Experimental approach:
• Fabricate duplicate aluminum and stainless steel heat exchangers
• Perform parametric study with un-fouled heat exchangers
• Expose duplicate heat exchangers to engine exhaust at conditions conducive to thermophoresis to rapidly foul the heat exchanger
• Repeat parametric study with fouled heat exchangers
12 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Aluminum heat exchanger performance was superior to stainless steel
• Aluminum heat exchanger provides a 40-60% increase in thermoelectric power output compared to stainless steel– Thermal conductivity of aluminum is approximately 9x higher than stainless steel– High exhaust temperatures may limit use of aluminum heat exchangers
• Although recovered power increases at higher flow rates, system efficiency decreases– Higher heat flux at the higher flow rate
0
1
2
3
4
5
40 60 80 110
Ther
moe
lect
ric
Pow
er (W
)
Simulated Exhaust Flowrate (slpm)
Unfouled AluminumUnfouled Stainless
0
0.2
0.4
0.6
0.8
1
40 60 80 110
Syst
em E
ffic
ienc
y (%
)
Simulated Exhaust Flowrate (slpm)
Unfouled AluminumUnfouled Stainless
Total heat recovery from five thermoelectric generators in-series.Simulated exhaust T= 380 deg C, Coolant T = 40 deg C
13 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Heat Exchanger Fouling ProcedureSingle-cylinder utility diesel engine
Heat exchanger water bath Dynamometer
Back-pressure valve
• Operate engine at 70% load• Engine backpressure set to 1.5 psi• Water bath temperature 50-70 deg C• Conditions conducive to thermophoresis
• 130°C temperature drop across heat exchangers• Experimental conditions held for 7 hours• Approximately 1 g soot deposited per heat
exchanger• 50-60 mg per 12” length of ¼” ID tube,
comparable to EGR cooler fouling studies
14 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Fouled heat exchangers reduced performance
• Performance of thermoelectric devices is degraded 5-10% compared to unfouled heat exchanger
– Heat exchanger on material has a much more significant impact on performance than fouled and unfouled duct
– Result seeming contradict EGR cooler fouling, where heat exchanger effectiveness can be reduced by more than a third with similar soot loading
0
1
2
3
4
5
40 60 80 110
Ther
moe
lect
ric
Pow
er (W
)
Simulated Exhaust Flowrate (slpm)
Unfouled AluminumFouled Aluminum
0
1
2
3
4
5
40 60 80
Ther
moe
lect
ric
Pow
er (W
)
Simulated Exhaust Flowrate (slpm)
Unfouled StainlessFouled Stainless
Total heat recovery from five thermoelectric generators in-series.Simulated exhaust T= 380 deg C, Coolant T = 40 deg C
15 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Why does the heat exchanger material have a larger effect than heat exchanger fouling?• Heat exchanger design was not optimal
• EGR coolers typically have thin walls, creating minimal resistance to heat transfer
• Heat exchanger used in this study had much longer characteristic heat transfer length, and created a substantial resistance to heat transfer
• Proposed improved heat transfer design is multi-layer flat-plate arrangement with thermoelectric devices sandwiched between heat exchanger layers• Additional complexity, system weight, and cost
Thermoelectric Generators
Coolant
16 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Conclusions
• Thermoelectrics can recover part of the large amount of waste heat available in the exhaust systems on gasoline and diesel engines, BUT… capture and conversion to useful work can be difficult
– Exhaust system is sized for full engine load, while the majority of the operating map is spent at part-load conditions
– Result is that heat exchangers must work well over a wide dynamic range
• Packing heat exchanger with aluminum wool enhanced heat transfer, thermoelectric performance
– Use of aluminum may not be practical with high temperatures in automotive exhaust systems, but similar performance expected from stainless steel wool
– Packing density can be low to minimize the exhaust backpressure
• Heat exchanger fouling degrades heat exchanger performance– Fouling of heat exchanger surfaces is a real-world challenge for EGR coolers, and is expected
to be problematic for all exhaust heat exchanger systems on diesel engines– Heat exchanger fouling decreased recovered power by up to 10%– With better heat exchanger designs, fouling layer is expected to degrade performance further
17 Managed by UT-Battellefor the U.S. Department of Energy Szybist_Thermoelectric
Acknowledgements
• The credit for these investigations belongs to the visiting faculty members who came to ORNL through the HBCU/MEI program
– Professor Essam Ibrahim, Alabama A&M University– Professor Norman Love, University of Texas El Paso
• This work was supported by the Laboratory Directed R&D program at Oak Ridge National Laboratory where this work was performed under project L05394.
• Thanks to the researchers at the ORNL vehicle chassis laboratory facility for providing experimental vehicle data, with particular thanks to Dean Edwards, John Thomas, and Brian West.
• Thanks to Scott Sluder for his guidance on heat exchangers and thermophoretic fouling.
Professor Essam IbrahimAlabama A&M University
Professor Norman LoveUniversity of Texas El Paso