Q LIBRARY
g o o q MIGHT"? State
University

 

 

 

This is to certify that the
thesis entitled

THERMOELECTRIC CONVERSION OF WASTE HEAT TO
ELECTRICITY IN AN IC ENGINE POWERED VEHICLE:
AN ENGINE MODELING APPROACH

presented by

Andrew T. Hartsig

has been accepted towards fulfillment
of the requirements for the

MS. degree in Mechanical Engineering

 

 

flew a. W

Ma§6r Pr6fessor’s Signature
8//7;/&6

Date

MSU is an affirmative-action, equal-opportunity employer

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K‘IProgIAccsprelelRC/DaleDue indd

THERMOELECTRIC CONVERSION OF WASTE HEAT TO

ELECTRICITY IN AN IC ENGINE POWERED VEHICLE:
AN ENGINE MODELING APPROACH

By

Andrew T. Hartsig

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE
Mechanical Engineering

2008

ABSTRACT
THERMOELECTRIC CONVERSION OF WASTE HEAT To ELECTRICITY
IN AN IC ENGINE POWERED VEHICLE:
AN ENGINE MODELING APPROACH
By

Andrew T. Hartsig

Approximately 60% of fuel energy is lost as heat in a modern internal combustion
engine. This work provides an evaluation Of the potential use Of thermoelectric generators
as a direct energy conversion device to produce electrical energy from the exhaust gases
Of a Cummins ISX 15 liter, turbocharged on-road diesel engine. The generated electricity
is to be used to augment the existing engine power output through a hybrid drive system.
Ricardo WAVE, an engine simulation code, was employed to create a model of the ISX
engine and determine exhaust gas properties as well as heat transfer to the duct walls of
the thermoelectric generator (TEG). The generated data is then used tO calculate a
theoretical improvement in brake specific fuel consumption (BSFC) and also as an aid to
material scientists working to create a thermoelectric material suited for this application.
In the modeling efforts it was established that maintaining power output levels on the
engine was paramount and therefore the TEG must not interfere with turbocharger
Operation. The final model configuration utilized two TEGS with a small heat exchanger,
yielding a BSFC improvement of 4.9% that was calculated at the “cruise” operating point
(1500 rpm and 62% full engine load). This can potentially save over 4200 gallons Of fuel

over the emissions useful life (435,000 miles) Of the engine.

ACKNOWLEDGEMENTS
First of all I would like to express my appreciation to my advisor Dr. Harold Schock for
the Opportunity to work with him at the MSU Engine Lab. I would also like to thank him
for his patience, guidance, and assistance over the past two years. Secondly I would like
to thank Dr. James Novak for bringing me into this project as a summer worker which led
to my enrollment in graduate school, for his assistance with WAVE, the many tangents
we got onto, the reality checks, and helping to keep me on track. I would like to thank Dr.
Giles Brereton for being on my committee as well as always being around for some good
car discussions. Thanks also go out tO Dr. Tim Hogan for being on my committee and for
his assistance with this project. I must thank the people that work at the engine lab Tom,
Ed, Melissa, Mayank, Mulyanto, Cody, Andrew, and everyone else who make it such an
entertaining place to work. I would like to thank the late Dr. James Hannemann and
family for stimulating my interest in science and engineering as a child. I owe thanks to
my Sister Jen for always being there to call me names. Thanks to all my friends for their
continual support and distraction. And finally I would like to thank my parents George
and Sandy Hartsig for raising me to question everything, think for myself and supporting
me in every way. Without their help I doubt I would have been able to come to MSU in

the first place.

iii

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................ vii
LIST OF FIGURES ......................................................................................................... viii
LIST OF SYMBOLS AND ABBREVIATIONS ............................................................... x
1 Introduction .................................................................................................................. 1
1.1 Project Concept and Goals ................................................................................... 1.
1.2 Introduction to Thermoelectrics .......................................................................... 5
1.3 Fuel Costs and Engine Life .................................................................................. 9

2 Introduction to Ricardo WAVE ................................................................................. 12
2.1.1 Heat Transfer Considerations ........................................................................ 15
2.1.2 WAVE Stick Models and Icons ..................................................................... 17

2.2 Cummins ISX Engine Test Data ........................................................................ 19
2.2.1 Engine Operating Points ................................................................................ 19

2.3 Analysis Methods .............................................................................................. 21
2.3.1 WAVEpost Post-Processing .......................................................................... 22
2.3.2 Excel Post Processing .................................................................................... 24

iv

WAVE Studies ........................................................................................................... 25

3.1 Phase I ................................................................................................................ 26
3.1.1 Single TEG Unit per Engine Cylinder ........................................................... 28
3.1.2 One TEG per Three Cylinders ....................................................................... 30
3.1.3 One TEG Unit per Six Engine Cylinders ...................................................... 31
3.1.4 WAVE Model Validation and TEG Effectiveness ........................................ 32
3.1.5 Optimization of TEG Design ......................................................................... 34
3.1.6 Effects of Unsteady Flow Dynamics on TEG Heat Transfer ........................ 38
3.1.7 Overall Results for Phase I ............................................................................ 41
3.1.8 Electrical Energy Utilization ......................................................................... 43
3.1.9 BSFC Calculation .......................................................................................... 45
3.1.10 Conclusions for Phase I ................................................................................. 47

3.2 Phase II .............................................................................................................. 49
3.2.1 Expansion Of Simplified Model ..................................................................... 49
3.2.2 New devices and sub-models ......................................................................... 49
3.2.3 Variable Geometry Turbocharger .................................................................. 50
3.2.4 Intercooler ...................................................................................................... 53
3.2.5 EGR Circuit ................................................................................................... 53
3.2.6 Control System .............................................................................................. 55
3.2.7 Sensors and Actuators .................................................................................... 56
3.2.8 WAVE Model Validation .............................................................................. 57
3.2.9 Results ............................................................................................................ 59
3.2.10 BSFC Calculation for Single TEG/EGR Cooler ........................................... 61
3.2.11 Summary Of Phase II ..................................................................................... 62

3.3 Phase III ............................................................................................................. 63

3.3.1 TEG Placement Concepts .............................................................................. 63
3.3.2 Dual TEG ....................................................................................................... 63
3.3.3 TEG on Each Exhaust Port ............................................................................ 67
3.3.4 TEG / EGR Cooler and Dual TEG Concepts with Heat Exchangers ............ 72
3.3.5 Advanced BSFC Calculation ......................................................................... 77
3.3.6 New Baseline Model ...................................................................................... 80
3.3.7 Updated BSFC Calculations for All Concepts .............................................. 83
3.3.8 Summary of Phase III .................................................................................... 88
4 Overall Summary ....................................................................................................... 89
5 Conclusion ................................................................................................................. 91
6 Future Work ............................................................................................................... 91
APPENDIX: VARIABLE GEOMETRY TURBINE MAPS ............................................ 92
REFERENCES .................................................................................................................. 97

vi

LIST OF TABLES

Table 1: Summary of recent works in thermoelectrics relating to IC engines .................... 3
Table 2: Analysis of "useful life" assuming 10% efficiency increase ............................... 11
Table 3: A sample Of the test data provided by Cummins for the ISX engine ................. 20
Table 4: WAVE heat transfer results for one, three, and six cylinders per TEG

configurations .................................................................................................................... 41
Table 5: Average TEG gas and wall temperatures ............................................................ 42
Table 6: BSFC percent improvement for three Phase I configurations ............................. 46
Table 7: Comparison Of Cummins test data tO WAVE model results ............................... 57
Table 8: Inner and outer wall temperatures for single TEG/EGR cooler concept ............. 60
Table 9: BSFC percent improvement for single TEG/EGR cooler ................................... 61
Table 10: BSFC percent improvement for dual TEG configuration ................................. 65

Table 11: BSFC percent improvement for 300m TEG duct length (un-optimized) .......... 70

Table 12: BSFC percent improvement: Single TEG/EGR cooler with and without heat

exchanger ........................................................................................................................... 74
Table 13: BSFC percent improvement: Dual TEG with and without heat exchanger ...... 76
Table 14: Results of the new BSFC percent improvement calculation for the dual TEG
study ................................................................................................................................... 78
Table 15: Comparison of percent difference between Cummins test data and baseline
WAVE model .................................................................................................................... 82
Table 16: Baseline engine model power output listed for each load point compared to
Cummins test data .............................................................................................................. 83
Table 17: Updated BSFC percent improvement for all concepts ...................................... 84
Table 18: Summary Of BSFC percent improvement ......................................................... 87

vii

LIST OF FIGURES

Figure 1: Sketch of a thermoelectric couple ........................................................................ 5
Figure 2: Single unit TEG with detail of TE module .......................................................... 7
Figure 3: Rectangular cross-section of TEG shown in Figure 2 ........................................ 8
Figure 4: Cummins ISX 6-cylinder diesel engine ............................................................. 10
Figure 5: Equivalent “effective” circular cross-section of TEG ........................................ 14
Figure 6: WAVE icons for a simple single cylinder diesel engine .................................... 17
Figure 7: Basic icons used by WAVE. yjun2 is a complex Y-junction; yjun3 is a simple

Y-junction; duct] is a basic duct with the arrow head defining the right end ................... 18
Figure 8: Six-unit TEG ...................................................................................................... 27
Figure 9: WAVE representation of single-cylinder engine with one TEG ....................... 28
Figure 10: WAVE representation of 3-cy1inder engine with one TEG ............................. 30
Figure 11: WAVE representation of a 6-cylinder engine with one TEG .......................... 31
Figure 12: Effect of TEG effective diameter —- single cylinder case ................................. 35
Figure 13: Exhaust gas temperatures at TEG midpoint ..................................................... 39
Figure 14: Heat flux at TEG midpoint ............................................................................... 40
Figure 15: Gas velocities at TEG midpoint ....................................................................... 40
Figure 16: Architecture of the thermal power B-62 operating point ................................ 43
Figure 17: WAVE engine system layout Of single TEG/EGR cooler configuration ......... 50
Figure 18: Compressor performance map ......................................................................... 51
Figure 19: WAVE icons for the VGT with components labeled ....................................... 52
Figure 20: WAVE schematic Of the intercooler ............................................................... 53
Figure 21: WAVE diagram Of the EGR circuit with labels .............................................. 54

viii

Figure 22: Diagram comparing wall temperatures for the single TEG/EGR cooler
configuration at the B-62 operating point with the Phase I six cylinders per TEG

configuration ...................................................................................................................... 60
Figure 23: WAVE diagram for the dual TEG configuration ............................................. 64
Figure 24: BSFC percent improvement comparison for single TEG/EGR cooler and the

dual TEG configuration ..................................................................................................... 66
Figure 25: WAVE stick diagram Showing the 6 TEGS in the exhaust manifold ............... 67
Figure 26: Plot Of engine horsepower vs. TEG length for all 5 operating points .............. 69

Figure 27: Plot of total TEG heat transfer vs. TEG length for all 5 Operating points ....... 69

Figure 28: WAVE stick diagram of single TEG/EGR cooler (length 150 cm) with
additional heat exchanger (length 50 cm) .......................................................................... 73

Figure 29: WAVE Stick diagram of dual TEG configuration (length 200 cm) with
additional heat exchanger (length 50 cm) .......................................................................... 75

Figure 30: Comparison of the results of the new BSFC percent improvement calculation
to the previous calculations ............................................................................................... 79

Figure 31: WAVE diagram Of baseline model with heat exchanger replacing TEG ........ 80

Figure 32: Comparison of BSFC percent improvement: Single TEG/EGR cooler concept
with and without heat exchanger ....................................................................................... 85

Figure 33: BSFC percent improvement for all five concepts using new BSFC calculation

........................................................................................................................................... 86
Figure 34: Turbine Performance Map VGT Rack Position 1.0 ......................................... 92
Figure 35: Turbine Performance Map VGT Rack Position 0.75 ....................................... 93
Figure 36: Turbine Performance Map VGT Rack Position 0.5 ......................................... 94
Figure 37: Turbine Performance Map VGT Rack Position 0.25 ....................................... 95
Figure 38: Turbine Performance Map VGT Rack Position 0.0 ......................................... 96

Images in this thesis are presented in color.

Symbol

A- 100
A—25
B-lOO
B-62
BHP

BISG
BMEP
BSFC
C-100
CAD
CAE
CCD

EGR
ESC
Ge
HP
IC

NOx

NTEG
CTR

QTEG
RPM
RTGs
Si

TE

Te
TEG
VGT

1IBISG
11le
rITEG

LIST OF SYMBOLS AND ABBREVIATIONS

Description

Operating Point: 1230 RPM @ 100% Engine Load
Operating Point: 1230 RPM @ 25% Engine Load
Operating Point: 1500 RPM @ 100% Engine Load
Operating Point: 1500 RPM @ 62% Engine Load
Brake Horsepower

Bismuth

Belt Integrated Starter/Generator

Brake Mean Effective Pressure

Brake Specific Fuel Consumption

Operating Point: 1800 RPM @ 100% Engine Load
Computer-Aided Design

Computer-Aided Engineering

Charge-coupled Device

Coefficient of Friction

Exhaust Gas Recirculation

European Steady-State Cycle
Germanium

Horsepower

Internal Combustion

N-type Thermoelectric Leg

Nitrogen Oxides

Number of Thermoelectric Generators
Over-the~Road

P-type Thermoelectric Leg

Gas Side Heat Transfer to TEG Duct Wall
Revolutions per Minute

Radioisotope Thermal Generators
Silicon

Thermoelectric

Tellurium

Thermoelectric Generator

Variable Geometry Turbocharger

Belt Integrated Starter/Generator Efficiency
Inverter/Power Electronics Efficiency
Thermoelectric Generator Efficiency

1 Introduction

1.1 Project Concept and Goals

Currently the best internal combustion (IC) engines have brake efficiency of 40%, with
35% of the fuel energy going to exhaust, and 25% to other heat transfer losses such as the
cooling system. Thus in the best IC engines, 60% of the energy content in the fuel is
rejected as heat. As transportation fuel costs continue to rise, new technologies are

needed which will improve fuel efficiency.

This work provides an evaluation for the potential use of thermoelectric materials
implemented in a direct energy conversion device to produce electrical energy from the
exhaust gases of a Cummins on-road diesel vehicle powerplant. The generated electricity
will be used to augment the existing engine power output through a hybrid drive system
and will therefore decrease overall fuel consumption. While the maximum amount of
energy recovery may take place at the peak power output of the engine, road vehicle
engines do not typically operate at peak power for extended periods of time and therefore
more realistic operating conditions must be studied to evaluate the potential energy

recovery.

Recently several groups have investigated the use of thermoelectric devices to capture
waste heat from the exhaust of IC engines. A few relevant findings of these efforts are

discussed below and a summary is listed below in Table 1.

Kushch et al. at Hi-Z Technology Inc., a manufacturer of thermoelectric devices, have
been experimenting with thermoelectric devices on IC engines Since at least the year
2000. Hi-Z has been successful in utilizing 72 of their TE modules to construct a TEG
generating 1 kW of electrical power at 30 V DC. The testing was conducted on a
dynamometer fitted with a Cummins 335 HP Diesel engine. The Hi-Z experiment found
that the TEG power output strongly depended on engine loading and less on the engine

speed. [1]

Ikoma et a1. with Nissan Motor Co. have experimented with integrating TE devices to
capture exhaust heat before the catalytic converter. Using 72 Si-Ge TE modules of their
own design, they were able to generate nearly 36 W of electrical power. The testing took
place on a combustor test rig which approximated the exhaust gas temperature and flow
rate of a 3 L gasoline engine performing a 60 km/hr hill climb. The testing determined
that the generated power was 0.9% of the heat flux from the exhaust gas to cooling water.
They concluded that further improvements are needed to prevent heat transfer around the

TB modules, and the conversion efficiency of the modules also needs improvement. [2]

Work by Matsubara et al. at the Science University of Tokyo in Yamaguchi has also
explored the use of thermoelectric generators to extract energy from exhaust gases of an
IC engine. This group was heavily involved with developing new TE materials for
Operation in the 350—750°C temperature range. Experimental testing of their new
materials took place on a 2 L gasoline engine in a Toyota Estima. The TEG was

constructed of 6 segmented modules (skutterudites / BigTe3) and 4 HZ-14 modules from

Hi-Z Technology, Inc. The testing took place with the engine driving the vehicle at 60
km/hr. The system initially produced electrical power of 141 W (at 22.4 V and 6.29A)
and after modification the system was able to produce 266 W of electricity. They
concluded that the power generated was approximately half of the potential of the TEG
and that poor thermal conductivity between the exhaust gas and the TE modules was the

cause.[3]

A review paper produced by Vazquez et a1. (Universidad Pontificia Comillas in Madrid,
Spain) researched works over the previous three decades concerned with thermoelectric
devices utilizing exhaust heat energy from [C engines. This work provided a useful
breakdown of other works into multiple categories and providing comparisons between
works. This work concludes that the transfer of heat from the exhaust gas to the TB

modules is the primary obstacle to high performance in TEGS. [4]

Table 1: Summary of recent works in thermoelectrics relating to IC engines

 

Recent work I Description Of work Assessment

 

’ HiZ-lkW Generator for Diesel, 1kW,
5% efficiency, ~600C exhaust gas, but ~600C Exhaust gas and 250-275C

 

 

HI‘Z [I] only 250-275C hot side, water cooled hot Slde suggests poor
cold performance of heat exchanger
They concluded poor
Nissan [2] 36W generator, 1% efficiency, 595C performance of heat exchanger
‘ Exhaust gas, Si-Ge and significant parasitic heat loss
from module
They concluded poor heat
266 Watts from 2.0 Liter engine, exchanger performance- ~200C
Skutterudite-BiTe, ZT~1 at 800C, drop from exhaust temp to hot

Matsubara et

al [3] Exhaust gas ~650C, but hot side only side and ~125C drop on cold side.

450C, and cold side 150C despite water TE generator performance 50% of
cooling at 25C what was expected as a result of
poor heat exchanger.

 

Critical component of waste heat
recovery is heat exchanger
efficiency.

Vazquez et al. Significant experience with TE
[4] generators

 

 

 

 

 

The work presented within is a component of a greater DOE Project under way at MSU
in conjunction with Iowa State University, NASA Jet Propulsion Laboratory, Cummins
Engine Company and Tellurex Corporation. Previous works were primarily focused on
experimental testing of TE devices. This work makes use of the engine simulation
software WAVE developed by Ricardo to predict gas and duct wall temperatures and
heat transfer rates at various locations on the engine. The temperature and heat transfer
data generated by WAVE is used in conjunction with a simple thermoelectric generator
(TEG) model to estimate the improvement to the brake specific fuel consumption (BSFC)

of the powertrain.

1.2 Introduction to Thermoelectrics

Currently, thermoelectric devices are commonly used in a variety of cooling and power
generation applications. They are fully reversible heat pumps, such that when electrical
current is supplied to the device, a temperature gradient is established, and when a
temperature gradient is supplied across the device, electrical current will flow (if a load

resistance is attached).

Thermoelectric generators (TEG) are constructed of a large number of thermoelectric
(TE) couples joined together thermally in parallel and electrically in series. A diagram of
a TB couple can be seen in Figure 1. Traditional uses include cooling electronics such as
infrared detectors, laser diodes for fiber optics, and CCD arrays, as well as power
generation applications such as radioisotope thermal generators (RTGs) used on Apollo

12, 14, 15, and 17, Voyager I and H, Galileo, etc. [5]

Hot

 

 

 

 

+ Cold -

’WV) L0...

Figure 1: Sketch of a thermoelectric couple

 

 

There are a number of advantages for thermoelectric devices over competing

technologies, including:

High reliability (> 250,000 hrs)

- Silent and no vibration

- Small electromagnetic signature

- Temperature control to fractions of a degree

- Not position—dependent

- Function in environments too severe, or too sensitive to conventional refrigeration
- Small and lightweight

- NO chlorofluorocarbons, chemicals, or compressed gases (nothing to replenish)

- Environmentally “green”

- Direction of heat pumping is fully reversible

The most significant disadvantage of thermoelectrics is their relatively low efficiency. A
broader related study at MSU focuses on increasing this efficiency by using longer legs,
higher temperature gradients, segmented legs, and new materials and fabrication

techiques.

As noted above, consideration of the possible sources of “waste heat” from diesel engines
led to the conclusion that the exhaust gases are the highest in energy and most easily
accessible by TEGS. Their use offers the further benefit Of replacing the existing exhaust
gas recirculation (EGR) cooler with the TEG, thus enabling effective NOx emissions

control.

Figure 2 shows one proposed design of what will be referred to here as a single TEG
which could be used in the exhaust system. The hot exhaust gas would flow through the
center passage and engine coolant would flow through the outer two passages similar to
counterflow heat exchangers. The thermoelectric modules are sandwiched between the
hot exhaust passage and the coolant passage on each side of the device. This initial
overall TEG design resulted from CAD “packaging” studies with the intent of fitting a
sufficient number of P-N modules into a single TEG unit to produce approximately 1 kW
of power. This was based on certain “best estimates” of P-N junction dimensions and

efficiencies.

Coolant In

   
  
 

Exhaust
Gas Out

Exhaust
Gas In

Figure 2: Single unit TEG with detail of TE module

As illustrated in Figure 2, exhaust gases from the engine exhaust ports, or the exhaust

manifold outlet (depending on the type of application, as discussed below), enter the left

end of the TEG unit and exit on the opposite end. Similar to a counterflow heat
exchanger, liquid coolant enters on the right, through separate ducting, and exits on the
left. Also shown is an enlarged picture of a P-N “Module” which includes 54 P-N
junctions. In the figure, 40 of these modules (20 each on the top and bottom) make up
the entire TEG “Unit” shown. Note that the overall length of the TEG Unit, which is a

key design variable as shown below, will detemline the total number of modules.

Figure 3 below shows a cross-section of the Single Unit TEG from Figure 2. The TEG
assembly is sandwiched with aluminum heat exchangers through which engine coolant
flows. The central duct, where the exhaust gas will flow, is constructed of stainless steel.
The P and N legs of thermoelectric material are electrically isolated from the heat
exchangers with aluminum nitride pads which also provide thermal conductivity between

the components.

Alumlnum
Heat Exchanger
(Cold Sldo)

  
 
  
 
  

Alumlnum
Nitrldo

Tungsten Pad

Stalnloss Steel
Hoat Exchan gar
(Hot Side)

Figure 3: Rectangular cross-section of TEG shown in Figure 2

1.3 Fuel Costs and Engine Life
There were two Cummins engines originally considered for use in this project. Both are
inline 6—cylinder, direct injected, turbocharged, and utilize cooled EGR for emissions

control.

The first is the Cummins ISB engine. This engine has a displacement of 5.88 L, and the
bore and stroke are 102 mm and 120 mm respectively. This engine also employs a
computer-controlled high-pressure common rail fuel injection system. This base engine,
in various configurations, is utilized in many applications. These applications include
marine, construction and agricultural, recreational vehicles, medium duty trucks and

busses, electrical power generators, and the Dodge Ram pickup truck.

The second is the Cummins ISX engine. This engine has a displacement of 14.95 L with
a bore of 137 mm and a stroke of 169 mm. The primary application of this engine is
Class 8 over-the-road (OTR) trucks. Other applications include construction and
agricultural equipment, electrical power generation, and recreational vehicles. This
engine employs many new technologies found in modem diesel engines such as computer
controlled unit injectors, variable geometry turbocharger (VGT), and cooled EGR. The
EGR system is of particular interest for this study because the ISX engine uses an EGR
cooler to lower the temperature of the exhaust gas before it is mixed into the intake air
charge. This offers a prime location where waste heat energy could be captured and

reintroduced into the powertrain by use Of thermoelectrics.

 

Figure 4: Cummins ISX 6-cylinder diesel engine

The aim of this project is to improve fuel economy in an on-road application. Two
common applications considered were the Dodge Ram pickup truck with an ISB engine
and a typical Class 8 OTR truck with an ISX engine. A comparative analysis was
conducted to determine which engine (ISB or ISX) would be the most economically

viable to implement the thermoelectric generator device.

An engine / powertrain efficiency gain of about 10% in either the ISX or ISB engines
would be considered a very significant improvement. According to Cummins, the engines
have an emissions “useful life” as defined by emission regulations. The emissions
equipment must function properly for the duration of the “useful life”. The “useful life”
of the ISB and ISX engines is 185,000 and 435,000 miles respectively. Table 2 shows
simple analysis of the “useful life” based upon emissions regulations, estimated cost of
fuel, some typical fuel economy expectations, fuel savings, and associated monetary

savings.

Table 2: Analysis of "useful life" assuming 10% efficiency increase

 

 

 

 

 

 

 

 

ISB Dodge Pickup ISX Class 8 Truck
Emissions — Useful Life 185,000 miles 435,000 miles
Typical Fuel Consumption 16 mpg 5 mpg
Fuel Consumed During Useful Life 11,500 Gallons 87,000 Gallons
Fuel Consumed with
10% Improve d Efficiency 10,500 Gallons 79,100 Gallons
Fuel Saved 1,000 Gallons 7,900 Gallons
Money Saved
($4.00/Gal April 2008) $490000 331’600'00

 

 

 

 

It is clear from the analysis that the ISX engine will see a greater overall fuel and cost
savings from an increase in fuel efficiency. The greater savings will be needed to help
offset the additional cost of the TEG device. If the fuel savings will not be able to offset

the cost Of the device it would be very difficult to market such a device successfully.

ll

2 Introduction to Ricardo WAVE

Early in the proposal phase of this project, it was determined that an engine simulation
code would be essential to the success of the project as a whole. Through the use of the
simulation code, gas and wall temperatures as well as heat flux can be predicted. This
data is useful to the TE materials group in developing and selecting materials which
perform optimally at the operating conditions of the engine. WAVE is one such engine
performance simulation code. It is a commercially available CAE tool which has been
utilized in industry for over twenty years in the development of internal combustion
engines. One of the primary uses of this code is to aid in the design of intake and exhaust
system components at various engine loads and throttle conditions. WAVE was chosen
because it provides a reasonable Simulation of the complex interactions found in an IC
engine. Wave dynamics can have a significant effect on engine operation. Wave
dynamics are pressure waves propagating through gases as they move through ducts.
These pressure waves can be reflected back in the duct by abrupt changes in cross-
sectional area (Duct to plenum junctions, opening/closing of valves, etc.). WAVE also
has the capability to simulate many of the complex components found on the ISX engine
(turbochargers, intercoolers, heat exchangers, etc.). The following section covers the

theory of operation and use of this software.

12

2.1 Theory of operation

WAVE utilizes a one-dimensional approach to solving the partial differential equations
governing the gas flow throughout the intake and exhaust system ducting of the engine.
While the governing equations are mathematically one-dimensional (with the spatial axis
along the centerline of the duct) it has many parameters which take into account three-
dimensional effects such as changes in duct cross-sectional area, bends and other flow
losses. In addition, substantial conductive and convective heat transfer models are
employed in WAVE. The primary reason behind the quasi-1D approach is that in a fully
3D code the computer calculation time can be prohibitively long. By using a quasi-1D
formulation, WAVE is able to provide results on a cycle-averaged basis, which is quite
useful when the extremely detailed thermo-fluid flow characteristics of a conventional
3D simulation are not required. Therefore, this type of quasi-1D code is the primary type
of CAE tool utilized in industry for studying the effects on engine performance and fuel

consumption from various engine design changes.

13

Although the preferred TEG construction shown in Figure 3 cannot be directly modeled
with the Quasi-1D approach, WAVE does allow one to account for most design features
of interest, such as the non-circular duct cross-section shape and its attendant increased
wall wetted surface area for heat transfer. Figure 5 illustrates how the rectangular TEG
cross-section (perpendicular to the gas and coolant flow directions) would be represented
within WAVE as a circular cross-section of the same “effective” flow and wetted wall
areas (Figure 3). The circular duct is surrounded by TEG modules which are then

surrounded by the engine coolant heat exchangers.

 

Figure 5: Equivalent “effective” circular cross-section of TEG

2.1.1 Heat Transfer Considerations

Noted previously, one of the major challenges in integrating a TEG into an IC engine
exhaust system is the effectiveness of the heat exchanger. In order to meet requirements
for TEG performance, the convective heat transfer from exhaust gas to TEG duct wall
must be increased. The following excerpt from the WAVE Basic User Manual [7]

discusses heat transfer multipliers applied to the convective heat transfer to duct walls.

The user may optionally multiply the calculated values of Cfdepending on

wall roughness. The multiplier should be set to unity for smooth walls and
to larger values for rough walls. The WAVE heat transfer model assumes
the Colbum analogy; that the transport of scalar quantities (such as heat)
follows a similar law to that for vector quantities (such as momentum).
Accordingly, the heat transfer multiplier should typically have the same
value as the skin friction multiplier. One exception is in the modeling of

intake and exhaust engine ports where C f Should be set to 0.0 because wall

friction is already included in the valve flow coefficients.

Normally WAVE suggests a value of 2.0 for the heat transfer multiplier for exhaust
ducts. This assumes that most exhaust ducts will have a rough inner surface which will

increase the convective heat transfer due to turbulence.

Work conducted by Shih et al. at Iowa State University, a partner in the DOE project, has
shown that the use of pins, ribs, dimples or a combination of the three will significantly

enhance the heat transfer over that of a smooth wall. They predict that in our model we

15

can assume that by using additional heat transfer enhancements we can increase the heat
transfer by up to 5 times over a simple rough wall. Combining the 5 times enhancement
with the 2 times multiplier from the rough surface, we assumed that ideally we will have
a 10 times heat transfer multiplier which will be used in the convective heat transfer

calculation for the TEG duct walls. [6]

For further detail on heat transfer in the WAVE model refer to the WAVE Basic, Engine,

and Conduction Manuals [7],[8],[9].

l6

2.1.2 WAVE Stick Models and Icons

This section is a brief introduction to how WAVE displays the layout Of the engine in its
preprocessor. WAVE uses icons to represent the various components found in the engine,
and this brief introduction will help familiarize the reader with the icon system which
appear later in this document in multiple figures. Every component has an associated
dialog box where parameters of that component are specified.

Figure 6 below contains icons for a special type of junction which represents an engine
cylinder, two ambients, two orifices and several ducts. Ducts are represented by solid
black lines with an arrow head on one end (labeled as IntDuct, IntPort, Ethort and
Ethuct in Figure 6). The arrow head on the duct is used to designate the right end of the
duct, not the flow direction. This left to right convention is used regardless of the
orientation of the duct on the canvas and is primarily used when ducts have a taper along
their length. The blue circle on the left side of the EngineCylinder junction in Figure 6

represents the intake valve. The red circle on the right represents the exhaust valve.

 
         
 
    

IntDuct

   

IntPort» Ethuct

 

   

IntakeAmbient Orifice1 Engine ylinder Orifice2 ExhaustAmbient

Figure 6: WAVE icons for a simple single cylinder diesel engine
Ducts representing ports (IntPort and EthOIt) are used to connect the valves to the
orifices (Orificel and Orifice2) which then have ducts (IntDuct and Ethuct) connecting
out to the ambients (IntakeAmbient and ExhaustAmbient). By default the diameter of an

orifice automatically matches that of the ducts which connect to it. The diameter of an

17

orifice can also be specified to any diameter from 0 to the diameter of the ducts which
connect to it. This variable orifice diameter is often used to model a simple valve. At the
top of the cylinder junction is an icon which represents a fuel injector, this may be placed
directly into cylinder junctions, ducts, or Y-junctions depending on the engine
configuration.

Figure 7 shows a basic duct (ductl) connecting a complex Y-junction (yjun2) to a simple

Y-junction (yjun3).

 

yjun2 yjun3

Figure 7: Basic icons used by WAVE. yjun2 is a complex Y-junction; yjun3 is a simple Y-junction;
duct] is a basic duct with the arrow head defining the right end

Simple Y-junctions allow multiple ducts to connect together without needing to specify
the exact orientation of each duct. It is assumed that the ducts are evenly separated on a
virtual spherical volume. Complex Y-junctions allow for much greater control over how
ducts connect together including angles between ducts and volume of junction, in
addition to allowing a special type of duct called a multiple duct which consists of several
small ducts all connected in parallel which is useful for heat exchanger modeling. Special
icons are introduced later as the associated components are integrated into the WAVE

models.

18

2.2 Cummins ISX Engine Test Data

Cummins Inc. supplied experimental and design data on their ISX engine. Cummins tests
their heavy duty OTR diesel engines at 5 operating points as defined by the ESC
(European Steady-State Cycle). These operating points represent the 5 of the more typical
driving situations that the engine will experience. Experimental data are collected at these
5 operating points on an engine dynamometer. This includes engine output (torque) as
well as data gathered from sensors in various locations around the engine yielding gas
temperatures and pressures. Additionally, critical values such as fuel rate, air flow rate
and EGR mass flow have been measured. Cummins also supplied a great deal of design
information about the ISX engine including, basic geometry of the engine (bore and
stroke), intake and exhaust manifold geometry, fuel injection profiles, valve lift profiles

and other important factors to the modeling effort.

2.2.1 Engine Operating Points

The operating points are designated by a letter and a number. The letter corresponds to
the engine speed and the number is the percentage load on the engine. Five operating
points are of interest consisting of three engine speeds of: (A) 1230 rpm, (B) 1500 rpm,
and (C) 1800 rpm. Each of the three engine speeds is tested at full (100%) load. Two
operating points are considered to be “part load”, 25% of full load at 1230 rpm (A-25),
and a 62% of full load at 1500 rpm (B-62). The B~62 operating point is the focus of much
interest as this is designated as the “cruise point” by Cummins. This is the operating point
where a truck has a full trailer and is cruising at a steady highway speed. Cummins has
gone to great lengths to make the B-62 point as fuel-efficient as possible for this is how

the engine will Spend the majority of its useful life. In the WAVE simulation, engine load

19

is regulated by controlling the fuel rate. The fuel rate for the five operating points was
provided by Cummins, and the fuel injectors in WAVE were set to deliver the specified
fuel rate regardless of how the rest of the engine is performing. Therefore, if the power
output of the model is similar to that of the experimental engine, it can be assumed that
the model is performing accurately. Table 3 below displays a sample of the test data

provided by Cummins. The five columns on the right Show the data for each of the five

Operating points.

Table 3: A sample of the test data provided by Cummins for the ISX engine

A-IOO B-62 B-100 0-100
Units
Engine Crankshaft Speed rpm 1230.00 1230.00 1500.00 1500.00 1800.00
Torque ft-Ib 472.15 1886.80 1170.20 1887.30 1577.70
BMEP psi 78.05 31 1.92 193.45 312.00 260.82
Power HP 1 10.58 441 .88 334.22 539.02 540.72
kW 82.46 329.52 249.23 401.96 403.22
Fuel Rate lb/hr 40.30 147.76 111.88 183.56 196.64
kg/hr 18.28 67.02 50.75 83.26 89.20
Air Flow Rate lb/min 19.81 47.78 46.71 63.55 68.42
Fuel-Air Ratio 0.034 0.052 0.040 0.048 0.048
EGR Mass Flow lb/min 7.18 9.87 12.80 15.96 13.22
EGR Fraction of Intake Charge 0.27 0.17 0.22 0.20 0.16
Barometer Hg abs 29.43 29.43 29.43 29.43 29.44
Intake Manifold Pressure Hg abs 13.90 61.21 46.19 73.57 62.38
Intake Manifold Temperature °F 120.07 121.78 114.42 130.47 135.12
Exhaust Manifold Pressure Hg abs 18.37 70.19 54.14 85.95 79.61
Exhaust Manifold Temperature °F 733.27 1170.90 945.27 1159.60 1233.60

A-25

20

2.3 Analysis Methods

WAVE has some features which are very useful in experimenting with model changes:
constants, cases and subcase sweeps. Constants allow the user to specify a name for a
particular numerical value. These names are then linked to a constants table which can
contain several cases. Cases are normally used to switch the state of the model to
investigate some change to the model. WAVE re-runs the simulation for each case
specified. Most commonly, cases are used to specify engine speeds, and in the case of

this project, the five operating points supplied by Cummins.

The constants table is arranged much like a spreadsheet and will have the various
constants set to each row and the different cases assigned to the columns. In addition to
the engine speed the fueling rate, boost pressure, injection timing and many other values
supplied by the Cummins test data are stored in the constants table. Subcase sweeps let
the user specify multiple values for a constant, usually by specifying the number of
subcases and the upper and lower bounds of the range and letting the software evenly
divide the range of values. When WAVE runs subcases, it will run each combination of
case and subcase. For example, if there are 5 cases and a subcase sweep cOntains 10
different values, WAVE will run the simulation a total of 50 times in order to generate
data for each combination. WAVE will allow subcases to be specified for up to two
constant names. This can be quite useful when trying to find a combination of two values
which yield a desired result. Running subcase sweeps will generate a large amount of raw

data, particularly when two constant subcase sweeps are run concurrently. This can be

21

particularly cumbersome to analyze because WAVE stores much of the data in a text

format.

Like most computer simulation work the post-processing of data takes as much if not
more time than the setup of the model. Fortunately WAVE is bundled with a post-
processor called WAVEpost which is quite effective at generating 2D and 3D plots for
any number of variables. Unfortunately WAVEpost is not ideally suited to certain aspects
of analysis particularly when a single cycle-averaged number is needed. For these

situations, a combination of text editors and Microsoft Excel work quite well.

2.3.1 WAVEpost Post-Processing

WAVEpost 2D plots are generated in two basic formats; time/engine cycle based and
sweep plots. WAVEpost will automatically generate several default time/engine cycle
plots covering a wide variety of engine performance and operation. If any time plots were
specified in the preprocessor, WAVEpost will also automatically construct those plots. It
is quite easy to create plots for case and subcase sweep studies. It is also possible to add a
secondary ordinate to plots in order to plot two different sets of data against a common x-
axis, for example torque and horsepower vs. engine speed.

Engine cycle plots are useful for studying phenomena such as wave dynamics or
temperature variation over a finite time period. Time-based plots are good for exploring
transient behavior like duct wall temperature variation. Sweep plots are used in

conjunction with either case or subcase constant sweeps, where the value being varied

22

(RPM, duct diameter, etc.) would be used as the x-axis and the dependent variable on the
y-axis.

Additionally, WAVEpost allows for 3D plots that are very useful in analyzing data
obtained from subcase sweeps of two constants (a double sweep). When a double sweep
is conducted, it is similar to the multiplying effect discussed above on cases and
subcases; but in many situations a much higher number of simulations will result. For
example, if two constants each have 10 subcases, that will require 100 simulations for a
single case. Now if that same setting is simulated for 5 cases that will require 500
simulations to generate data for each possible combination. Double sweeps are usually
limited to a single case to reduce calculation time and to make the analysis much easier.
A 3D plot can be generated, for example, with the two constants of a subcase sweep
assigned to the x and y-axes and some value like horsepower on the z—axis.

WAVEpost also allows for complete customization of the plot by changing the axis
scales, line type, colors, line size, and many other parameters in order to display the data
conveniently. WAVEpost can, in addition to outputting directly to a printer, save the plot

in a variety of digital formats.

23

2.3.2 Excel Post Processing

While WAVEpost is quite effective for plotting, it can be lacking when singular averaged
values are desired. A useful alternative was found in a combination of a text editor, in this
case Microsoft Notepad, and Microsoft Excel. The data is structured in the .sum text file
such that it displays all output data for each case before continuing on to the next case.
This case by case text block organization allowed the use of the find feature in Notepad
to search for the name of a particular value, such as TORQBR (Torque in ft/lbs), and
when the search was continued on from that point the next case’s value would be
highlighted. When values were identified they were copied into a spreadsheet organized
such that the values were easily compared case to case. In addition, functions were
written in Excel which automatically calculates the percent difference between two
values. This was used extensively when trying to match the model’s output variables to

those values in the experimental test data.

24

3 WAVE Studies

This chapter focuses on the modeling efforts conducted in WAVE. The chapter is divided
into three phases, ranging from a simplified model used for preliminary studies to
advanced models exploring alternative configurations with the goal of improving the
BSFC of the Cummins ISX engine. Phase 1 contains a section of work previously
conducted by Dr. James Novak who is a visiting professor at Michigan State University.
The author became involved in Phase I at the section titled Optimization of TEG Design.
This previous work section of Phase I has been included to provide an introduction to the

advanced modeling efforts of Phases II and HI conducted by the author.

25

3.1 Phasel

Phase I of the WAVE studies were conducted with a simplified model approach to
determine early in the project the amount of heat energy available. Critical information
obtained included the gas and wall temperatures in the TEG which will be used in
conjunction with the thermoelectric group to determine the best thermoelectric material
for this application. The Phase 1 modeling effort was focused only on the Bi-62 load point
(1500 RPM, 62% load), which is where this engine will be operated for the majority of its
useful life. Therefore this operating point is the most important in terms of fuel
consumption.

There are three basic options to consider in determining how best to configure one or
more of the single TEG units, shown in Figure 2, into the exhaust system of the ISX 6-
cylinder engine (see Figure 4):

1. One single unit TEG per each cylinder — six units for the 6-cylinder ISX
engine. This configuration might be best for packaging the largest number of
P-N modules per engine; hence, generating the most electrical power. Figure
8 illustrates one such configuration.

2. One TEG unit per 3 cylinders (two TEG units per engine). This case seems
to be a very reasonable alternative, because it would provide for three, equally
spaced (in time) exhaust gas “pulses” to each TEG unit per engine cycle (two
engine crankshaft revolutions). This follows from a consideration of the firing
order of the ISX engine and the fact that the interval between the opening

events of each cylinder’s exhaust valves is 120 crankangle degrees.

26

 

Figure 8: Six-unit TEG

3. One TEG unit per six engine cylinders. This configuration would entail only
one TEG unit for the entire engine and would provide the highest frequency of
exhaust pulses to the TEG per engine cycle — three pulses per revolution; six
per engine cycle. Although the heat loading to the TEG would be the highest
of any configuration, the number of P-N modules that would be exposed to

this energy input would be the least due to simple packaging considerations.

27

3.1.1 Single T EG Unit per Engine Cylinder

For this configuration, in which each cylinder exhausts into its own TEG, experiments
have shown and WAVE modeling results confirm that at engine speeds below 2000 rpm
a 6-cylinder engine behaves as six independent single cylinder engines. Therefore, the
case of six TEGS (one per cylinder) was effectively modeled as a single cylinder engine
and, as necessary, displacement-specific results (e.g., torque and horsepower) multiplied
by a factor of six. Figure 9 shows the simple single cylinder stick model used. This
assumption decreases the computer processing times by over a factor of six and allows

more extensive design optimization studies to be carried out in a practical way.

Intake Side J Exhaust Slde

 

Figure 9: WAVE representation of single-cylinder engine with one TEG
In this figure, flow through the engine is from the left inlet ambient (Shown as a blue
“cloud”) to the right exhaust ambient. Note that an ambient corresponds to an infinite
volume plenum for which thermodynamic properties, such as pressure, temperature, and
gas composition, remain fixed at specified values during the entire simulation. These
conditions can be those of the true ambient (i.e., atmospheric conditions), or they can

represent other locations in the engine’s intake and exhaust systems where conditions are

28

approximately constant with time and are known; e.g., from experimental data for the
operating conditions of interest. The latter case can be useful to simplify the model and
avoid having to calibrate and “carry along” computations for parts of the intake and/or

exhaust systems that do not have a significant effect on the results of interest.

In this study, we make use of this approximation to avoid the complexities of modeling
the variable geometry turbocharger of the Cummins ISX engine. Instead, we use the
measured values of pressures and temperatures: (1) at the outlet of the turbocharger
compressor and intercooler (intake manifold pressure and temperature in Table 3); and
(2) the inlet to the turbocharger turbine (exhaust manifold pressure and temperature in

Table 3); to represent the inlet and exhaust ambient conditions, respectively.

Another important assumption used throughout this initial study of TEG behavior in IC
engine exhaust systems, is that all of the exhaust gas expelled from the cylinders passes
through the TEG(s) before entering the exhaust ambient (turbine inlet). As a
consequence, the results to follow represent an ideal “upper bound” on the amount of
energy that can be extracted by the TEGS. In the actual case of the ISX engine, up to 50%
of the exhaust gas will be bypassed through the TEG/EGR cooler and the rest will enter
the turbine inlet. The turbine efficiency ultimately depends on the total enthalpy of its
inlet gases. This delicate balance of energy distribution within the ISX exhaust system
will depend strongly on the location and design of the TEG and of the

turbocharger/intercooler.

29

3.1.2 One TEG per Three Cylinders

Similar to the assumptions discussed above and confirmed by modeling studies, this
configuration effectively behaves as two independent three cylinder engines coupled
mechanically, but independent in terms of the flow dynamics in the exhaust system. It
can, therefore, be modeled as one 3-cylinder engine to facilitate shorter computer
processing times and, hence, more practical design optimization studies. Figure 10

shows the model used in the analysis below.

 

 

Figure 10: WAVE representation of 3-cylinder engine with one TEG

30

3.1.3 One TEG Unit per Six Engine Cylinders

Figure 11 shows the WAVE representation of this last configuration of interest. All six
cylinders merge on the exhaust side into a single TEG duct. This case required the
greatest CPU times, since all cylinders had to be modeled to predict the heat loading on

the TEG and the output of the engine accurately.

 

 

Figure 11: WAVE representation of a 6-cylinder engine with one TEG

31

3.1.4 WAVE Model Validation and TEG Effectiveness

Validation for Base Engine
The WAVE code was first used to simulate the baseline ISX production engine at the
B-62 operating point to validate model results for engine output. The detailed input
included:
0 Intake and exhaust valve lifts at one crankangle degree intervals pad

0 Intake and exhaust valve/port flow coefficients at 1mm valve lift intervals

F314? ”44.4

o Combustion energy release profile
0 Base engine design parameters (bore, stroke, connecting rod length, valve
diameters, compression ratio)
0 Centerline lengths and cross-sectional diameters of all intake and exhaust ducting
(ambient to valve seat)
0 Operating conditions (B-62 “cruise point.”)
The predicted value of engine BMEP was 210 psi. This is approximately 8.5% higher
than the experimental value of 193.45 psi listed in Table 1. A value within 10 percent of
experimental data was considered acceptable, especially in light of the (somewhat
idealized intake and exhaust system designs, which did not include the details of the
turbocharger compressor and turbine. Flow and heat transfer losses throughout the actual
engine’s extensive induction and exhaust system components will typically reduce shaft
output approximately 10 percent. As such, this baseline level of engine output was

considered well within the accuracy of engine simulation codes.

32

Approach Used to Evaluate TEG Effectiveness

Once it was determined that the code was Simulating the base engine output in a

reasonable manner, the single, three, and six cylinder TEG configurations were evaluated

to determine which is the most effective in extracting exhaust heat energy. To simulate a

“best case” scenario, several assumptions were incorporated into the engine-TEG system:

1.

2.

3.

4.

The engine piston, cylinder liners, and cylinder head combustion chamber
surfaces were near adiabatic. This was accomplished by reducing the gas side
heat transfer coefficients by 99%. This would be similar to using ceramic or other
insulating coatings on these surfaces.

The exhaust ports and manifold were similarly assumed to be beat insulated to
maximize the heat input into the TEG duct.

The TEG coolant temperature was assumed to be a constant value of 325 degrees
Kelvin along its entire length. In actuality the coolant temperature will increase
along its length, but this is of secondary importance in the resulting total exhaust
heat extraction through the TEG structure.

The walls of the TEG are made up of two homogeneous layers. The inner gas-
side layer is carbon steel, 3.175 mm (1/8 inch) thick, with a thermal conductivity
of 48 W/m/K; and the outside layer represents the P-N modules with a thickness

of 11mm and thermal conductivity of 3 W/m/K.

These assumptions allow for the maximum exhaust heat loading of the TEG, and the

following results thereby represent an upper bound to the energy that would be available

for conversion to electrical power.

33

3.1.5 Optimization of TEG Design

Figure 3 above contains the key dimensions for the baseline geometry of the single unit
TEG. This configuration was used as the starting point to determine an “optimized”
design for the single, three, and six cylinder cases, by simulating the engine operation at
the B-62 point for a range of TEG cross-section diameters and lengths. The primary
output variable of WAVE, used as a measure of the TEG effectiveness in these studies,
was the total heat transferred per unit time through the walls of the TEG duct during an
engine cycle, denoted “DHTEGI” (units of kilowatts) by the code. This variable is
calculated within WAVE by integrating the gas-side heat flux, which depends on the
instantaneous values of wall and gas temperatures and the instantaneous heat transfer
coefficient, over the duct’s internal surface area at each instant. The final number
reported by WAVE represents a cycle-averaged value. The optimizations that follow are
based on finding the maximum value of DHTEGI as a function of key design parameters
and subject to certain constraints, such as maintaining engine output (BMEP), TEG
packaging limitations, etc. BMEP is used as the measure of engine shaft output, instead
of brake torque, because it normalizes the output with respect to engine displacement;

i.e., it is independent of the number of engine cylinders.

The interior (exhaust gas-side) cross-section diameter was found to be the most important
design variable for the TEG, because it has a direct effect on the engine output, as well as
the amount of heat transferred from the exhaust gases. It was considered important in
these studies to avoid any significant degradation of the ISX engine output by the

introduction of the TEG in its exhaust stream; e.g., by increasing the exhaust flow

34

restriction losses. Such a reduction in BMEP would ultimately have a detrimental impact
on the engine’s fuel efficiency and emissions by necessitating an increase in engine

displacement to meet load requirements.

For the single cylinder case (one TEG per cylinder), Figure 12 shows the effect of
changing the TEG interior (exhaust gas-side) cross-section diameter on both the engine
BMEP (right-hand scale) and the total heat energy lost from the exhaust gases to the
surrounding TEG wall surfaces per unit time during an engine cycle, DHTEGI (left-hand

scale).

---- DHTEG1 — BMEP

a)
5

 

236

5

0)
(Q
I

.
b,—e

E

M
00
ha
[Isd] eInsseId aagoege ueaui axeia

—
u—T'
‘—
a
r
v
a
a
a
r
.
‘l
a
p
v
’a
v
,r
’v
’r

    

 

 

 

Heat transfer from gas to inside surface of duct TEG1 [kW]

31 -
— 230
30— ,’
r"
29 I V I f + 228
5.0 5.5 6.0 6.5 7.0 7.5 8.0

TEG Diameter [cm]

Figure 12: Effect of TEG effective diameter — single cylinder case

35

As TEG diameter is reduced, the BMEP does not change substantially until the diameter
is less than 6.5 centimeters. This general behavior is expected, since reductions in the
effective flow area of the engine exhaust system eventually lead to a significant
restriction of the exhaust gas flow and an accompanying increase in the engine pumping
work. The figure also shows that DHTEGI increases with larger cross-section diameters.
This follows from the concomitant increase in interior surface area within the TEG
available for heat transfer from the exhaust gases. Although the heat transfer coefficient
also increases as diameter is decreased because of the associated increase in gas
velocities, the dependence of the overall heat transfer in the TEG on surface area proves

to be the dominant factor.

Based on these results, a diameter of 6.5 cm was chosen for the single cylinder case.
Although larger diameters would increase the value of DHTEG1, the average velocities
through the TEG would decrease substantially, adversely affecting the performance of the
turbocharger turbine located immediately downstream from the TEG duct. Therefore, the
“optimum” TEG diameter was chosen to be the smallest value that does not diminish the
engine output (BMEP) at the B-62 operating point. Phase II studies will further refine
these design parameter tradeoffs by modeling the full turbocharger and EGR bypass

systems at both part and full load conditions.

Similar TEG diameter optimization studies were carried out for the three and six cylinder

configurations, resulting in diameters of 7.5 cm and 8.5 cm, respectively.

36

The other primary TEG design parameter is its axial length. As expected, the modeling
results indicate a nearly linear increase in DHTEG1 for lengths ranging from 0.5 meters
to 2.5 meters. Within this range, the BMEP was not significantly affected. A final value
of 1.5 meters was chosen for all three configurations due to estimated packaging limits
for the ISX engine. Again, Phase II studies will further define this parameter. In the
current feasibility study, we want to focus on the number and overall integration of the

TEG(s) within the engine’s exhaust system.

37

3.1.6 Effects of Unsteady Flow Dynamics on TEG Heat Transfer

It is important to note that the basic characteristics of the gas dynamics within the
exhaust ducting, and the TEG duct in particular, are very different for the three
configurations of interest, primarily because of the differing number of cylinders
connected to the TEG. Each cylinder produces one exhaust “blowdown pulse” per
engine cycle during the initial stages of the exhaust valve opening. Each of these pulses
contain high pressure and temperature exhaust gases from the cylinder that will traverse
each exhaust duct and be transmitted or reflected at duct transitions (e.g., changes in
cross-sectional area). This in turn results in dynamic variations in the gas-side heat
transfer coefficient. Figures 8, 9, and 10 illustrate this by showing the variation in the
exhaust gas temperature, heat transfer coefficient, and gas velocity, respectively, in the
TEG at its midpoint in length. The three curves in each figure are the results for the

optimized geometries of the one, three, and six cylinder cases.

Figure 13 shows distinct peaks in the gas temperatures for each of the cylinder blowdown
events for the single cylinder (one broad peak) and the six cylinder (six peaks) cases.
The three cylinder case has three primary peaks and three secondary peaks of lesser
amplitude. The secondary peaks represent reflections of the primary waves at the
ambient exit boundary. The abscissa is the engine crankangle for cylinder 1 in which
zero crankangle degrees represents the TDC (top dead center) piston position at the
beginning of the expansion stroke for cylinder 1. For reference, the exhaust valve begins

opening at 134 crank degrees.

38

 

 

 

 

 

 

 

..... 1 Cylinder - Midpt TEG
_-_ 3 Cylinder - MidptTEG
--- 6 Cylinder - Midpt TEG
900
850~ ,
I, \“ {‘x If“
I, I I‘\ '1 I /h\ I I II“
800 _ \ I ‘ I \ ' ‘ ’I \‘ I | I
I—0 \ I I I \ I I I \ ‘ I
x I \ I \ I ‘J ‘ i ‘v’
- I r
g 750- \ I \” \
E
E 7004 ‘~
2 ‘- ..’ - a ‘s‘
E 5""
650~ 5
600~ _W .. , i. i I. ''''' ‘
550200 0 200 400 600

CRANK ANGLE [deg]

Figure 13: Exhaust gas temperatures at TEG midpoint

Figure .14 shows the corresponding heat fluxes to the TEG walls at its midpoint location.
(A negative heat flux indicates heat transferred from the exhaust gases to the adjacent
walls.) Again, distinct primary peaks are observed, depending on the number of

cylinders “feeding” the TEG. The single cylinder case has the higher amplitude heat flux

because of the higher gas velocities, shown in Figure 15. The higher velocities, in turn,

are a result of the smaller cross-sectional area needed to expel the exhaust gases from a

lesser number of cylinders.

39

 

 

 

\
IIIII

l

.1:
llllll

lllllll
ll
lllll

 

1 Cylinder - Midpt TEG
—-- 3 Cylinder - Midpt TEG
_ 6 Cylinder - Midpt TEG

460

 

I I: .-
IIIIIIIIW-IIHUHIIII uuuuuuuuuuu

200
CRANK ANGLE [deg]
Figure 14: Heat flux at TEG midpoint

 

 

 

1 Cylinder - Midpt TEG
—-- 3 Cylinder - Midpt TEG

6 Cylinder - Midpt TEG

 

 

 

o
o

4
r532 x3“. cm:

-600 -

 

 

4309200
200
150 -
100 «

5

Es. Eood>

 

360

1éo
CRANK ANGLE [deg]

540

Figure 15: Gas velocities at TEG midpoint

40

3.1.7 Overall Results for Phase I

Table 4 summarizes the overall results of this Phase I analytical study of the potential
viability and benefits of the application of TEG technology to the ISX engine.

All results are for the B-62 operating point. The second column displays the resulting
value of DHTEGl, the average heat transfer rate from the exhaust gases to the walls of
the single TEG in kW, for the one, three, and six cylinder configurations. The third
column breaks this number down to a per cylinder value. The final column gives the
amount of energy transfer rate from the complete engine/TEG system in which there are a

total of six, two, or one TEG, respectively, for each of the configurations.

Table 4: WAVE heat transfer results for one, three, and six cylinders per TEG configurations

 

 

 

 

 

. . TEG Energy Energy Input Total TEG
(O tfrgiggguéaegret ) Input Rate per Cylinder Energy Input
p ry [kW] [kW/cylinder] [kW/engine]
One Cylinder per TEG 31.9 31.9 191.4
Three Cylinders per TEG 50.2 16.7 100.4
Six Cylinders per TEG 64.5 10.8 64.5

 

 

 

 

 

The overall results indicate that more energy is extracted per cylinder per engine cycle for
the single cylinder configuration. This follows primarily from the fact that the total
wetted gas-side surface heat transfer area of the TEG per cylinder is three and six times
greater, respectively, than for the three and six cylinder configurations. Also, the lower
frequency of exhaust “events” per engine cycle for the single cylinder case, allows the
interior walls more time to cool off between pulses and, thereby, provide a larger average
temperature gradient to “drive” the transfer of energy from the exhaust gases to the TEG
structure. Table 5 shows the time—average wall and exhaust gas temperatures at the inlet
and exit of the TEG for each configuration. Both temperature values increase as the

number of cylinders feeding the TEG increases.

41

Table 5: Average TEG gas and wall temperatures

 

 

 

 

 

 

Wall Temp Gas Temp Wall Temp Gas Temp
Configuration (T EG Inlet) (T EG Inlet) (T EG Outlet) (T EG Outlet)
[K] [K] [K] [K]
One Cylinder per TEG 708 856 597 634
Three Cylinders per TEG 798 884 662 711
Six Cylinders per TEG 831 898 731 785

 

 

 

 

 

The result above is that one TEG unit per cylinder is more effective and preferred over
the cases of one TEG per three or six cylinders. This result has profound implications for
the development and testing of TEGS for application to internal combustion engines.
Most importantly, a single cylinder test engine, rather than a multi-cylinder (three or six)
engine, is all that is necessary in the first development and test phase of the TEG. This in
turn substantially lowers the time and cost to fabricate the prototype TEG designs, as well
as that of the associated testing. Moreover, the WAVE engine simulation can be used in
conjunction with this hardware design and testing to further optimize the TEG for
packaging, as well as function. Following tests on a single cylinder engine, WAVE can
be calibrated to the experimental data and then used to project the TEG and engine

performance on the full Cummins ISX six-cylinder engine.

42

3.1.8 Electrical Energy Utilization

The electrical systems and modeling team envisions the recovery of engine waste heat in
large OTR trucks for the primary purpose of redirecting this energy to the vehicle’s
wheels. Recovery of waste heat and its immediate delivery to the vehicle wheels in effect
turns this concept into a thermal power split hybrid. This is a micro hybrid to be precise.
The overall architecture of the thermal power split hybrid is shown as Figure 16.

Engine Coolant Induction

Air

    
  
 
  
 
       

Pe@62%=12kW i

PC @ 100%: l SkW
Compressor

     
     

   
  

 

 

   
         
     

 

:5 <2
i Z mm Assume 50% of
:5 3 i ,x Q ' exhaust to EGR,
5 cl: } >< 50% of exhaust to
‘3- r 1 \/ turbine.
/ ‘ KW
\/ Turbine
_ /\
nle- 0996/ ................ 1 KJ
= ‘ WV T , Cool Exhaust
”BISG 0' 3 >:; h qh
"mi: 0.89 ; K j .. if” ’ .- "'7
Excess l‘_/WP\‘_W
Electrical A (fig/‘1“ :1 ":74
( ( a . .
Power p m \ ( Additional
g [3 Trans. 1 energy recovery
' Powere , J opportunity
Ancillaric FT

9\ Wheels
T , if FD ,L W

\__,// Pn];249+16 @ 100%, 10-5 @ 62%

 

Figure 16: Architecture of the thermal power B-62 operating point

In the system of Figure 16, a fraction of the heat rejected via the engine exhaust will be
returned to the engine induction port via an EGR loop. Opportunity for energy
scavenging exists in the EGR 100p because a conventional EGR system will incorporate
an EGR cooler to cool the exhaust gas to near ambient conditions. Rather than reject this

heat through the radiator, the presence of a thermoelectric generator, TEG, will use the

43

presence of a thermal gradient to convert a fraction of this rejected heat directly into
electrical power as shown. The remainder of the engine-rejected heat is delivered in the
conventional manner to the turbine side of the turbo charger and from there to the exhaust

system.

The recovered waste heat from the EGR loop is matched to the hybrid drive system at a
nominal potential of 144 VDC. At this voltage level two important attributes are
realized: 1) it is economical using today’s power electronics and electric machine
technology to process up to 20 kW or more, and 2) the required number of total TEG
couples at 144 V is less than if a 42 V nominal electrical distribution bus voltage is
selected. The net effect of both benefits is that overall warranty is improved because
fewer couples are series connected and moreover, the need for excessive parallel paths is

minimized.

Electricity generated by the TE system is converted into mechanical work by the belt
integrated starter/generator (BISG). The BISG will reduce the load on the engine and
therefore improve efficiency of the overall powertrain. Some additional benefits of a
BISG include providing a start/stop capability of the engine to reduce idling time and

some degree of regenerative braking while the truck is coasting.

44

3.1.9 BSF C Calculation

Improvement of Brake Specific Fuel Consumption (BSFC) is the overall goal of this
project. That being the case, a method of calculating the improvement was necessary.

BSFC is calculated as follows in Equation 3.1:

FuelRate _ kg/hr

BSF C = —
Power Output kW

 

(3.1)

As previously mentioned, the WAVE model fuel rate (kg of fuel per hour) was calibrated
to match the Cummins test data. Since the fuel rate cannot change, the BSFC will
improve (numerically iower) as the total power output increases. The calculation to

determine the percent improvement in BSFC is:

% imp.in BSFC ___ QTEG 'NTEG '77TEG '773/30 'n/Nv X100

0.746 . BHP (3'2)

Where:
QTEG = Gas side heat transfer to TEG duct wall (kW)
NTEG = Number of TEGS being evaluated

Theo = Efficiency of TEG (Dependent on temperature gradient)
113.50 = Efficiency of Belt Integrated Starter Generator (96 %)

1’]le = Efficiency of Inverter and Power Electronics (93 %)

0.746 * BHP = Brake Horsepower converted to kW
(334.1 hp @ B-62 per Cummins data)

In the percent improvement in BSFC equation above, the numerator calculates the
amount of additional power which is introduced to the engine belt drive from the
thermoelectric and electrical drive systems. In Equation 3.2 the efficiency of the TEG

was estimated based on the temperature gradient reported by each TEG configuration.

45

The TEG efficiencies used in the calculation were 9.1%, 11%, and 12.3% for the one,
three, and six cylinders per TEG configurations respectively. The denominator of
Equation 3.2 is the crankshaft power produced by the WAVE model. Therefore this
equation yields the percentage of power increased over the crankshaft power output. As
the BSFC is dictated by the total power in the system, this percentage directly translates

to an improvement in BSFC.

A calculation for the improvement of BSFC was calculated using the values from Table 4
and the 334.1 horsepower from the Cummins test data (Table 3) and compiled in Table 6
below. The horsepower from the Cummins test data was used in the BSFC calculation
rather than the WAVE reported horsepower. This was done in order to draw a better

comparison between the three different TEG configurations.

Table 6: BSFC percent improvement for three Phase I configurations

 

 

 

 

 

. . TEG Energy BSFC
Configuration
. . Input Rate Improvement
(Optimized Geometry) [kW] [7°]
One Cylinder per TEG 191.4 6.2
Three Cylinders per TEG 100.4 4.0
Six Cylinders per TEG 64.5 2.8

 

 

 

 

46

3.1.10 Conclusions for Phase I

The objective of this Phase I analytical study was to quantify the potential benefits of
thermoelectric generator (TEG) design alternatives in converting waste exhaust heat from
the Cummins ISX, 15 liter, 6-cylinder, diesel engine into electrical energy. The studies
are carried out for engine operating conditions most representative of those for an on-
road, eighteen-wheeled, Class 8 truck. To determine the maximum possible energy
available for conversion, the TEG was located in the exhaust system immediately
downstream from the engine’s exhaust ports and manifold where it could serve as a
possible replacement for the current engine’s EGR (Exhaust Gas Recirculation) cooler. It
was assumed that 100% of the exhaust gases pass through the TEG.

The study indicated several important results:

1) A greater amount of exhaust energy can be extracted through the TEG when each
cylinder of the 6-cylinder engine has its own separate TEG passage, as opposed to
alternate designs having three or six cylinders “feeding” a single TEG unit.
Comparing the most practical designs of one versus three cylinders per TEG unit,
simulation results for the B-62 (1500 rpm and 62% of full load) highway cruise
operating point predict that the single cylinder configuration allows over 90%
more energy extraction through the TEG unit per cylinder than the three-into-one
design. For the full 6-cylinder ISX engine, this indicates that 191.4 total kilowatts
of exhaust energy could be extracted from the exhaust gases and is, therefore,
available for conversion into electrical energy. A significant consequence of this
finding is that the next stage of TEG design, development, and testing can be

carried out using a single cylinder engine test rig, as opposed to the full 15 liter, 6-

47

2)

cylinder engine. This would lead to substantial cost and time savings as TEG
prototypes would be one-sixth smaller and therefore require fewer P-N modules.
Also, the WAVE engine simulation code can then be calibrated to the single
cylinder test data and used to project the impact of each design iteration on the
six-cylinder ISX engine.

Using liquid coolant, equivalent to that used in the engine cooling system (and
assumed to be at a constant 325 degrees K), to cool the outer walls of the TEG

results in average exhaust gas-side wall temperatures in the range of about 600 —

700 degrees K.

48

3.2 Phase II

3.2.1 Expansion of Simplified Model

The second phase of this study involves modeling the more realistic case of the full six-
cylinder ISX engine with the TEG unit(s) fully integrated into the engine induction and
exhaust systems, such that the engine exhaust gases divide between the TEG unit and the
variable-geometry turbine. This configuration is similar to the actual ISX engine and is
necessary to understanding the balance of the exhaust energy flow split between the TEG
and the variable geometry turbocharger in order to meet the engine’s power output, fuel
consumption, and NOx emissions requirements. Figure 17 shows the new system model
illustrating the manner in which the TEG now connects to one outlet of the exhaust
manifold, while another exhaust outlet connects to the turbine inlet. Part of the
experimental data supplied by Cummins detailed some of the intake/exhaust manifold
geometry and was used where applicable. In the case that geometry data was not
provided, the dimensions were approximated from pictures of the actual engine and

intake/exhaust manifold volumes which were provided by Cummins.

3.2.2 New devices and sub-models

On the left side of Figure 17, the cloud represents the intake air inlet which feeds the
compressor side of the variable geometry turbocharger (VGT). The compressed air then
passes through an intercooler before entering the intake manifold plenum. The cloud on
the right-hand side of the figure represents the outlet of a 1m long exhaust pipe connected
to the outlet of the turbocharger turbine. The heavy black line connecting the compressor
and turbine represents the mechanical shaft linkage and is not a gas flow path. New

control system elements shown in Figure 17 include control actuators for the EGR valve,

49

which controls the outlet flow of the TEG/EGR—Cooler, and the “rack position” of the

variable geometry turbine, which controls the inlet area size of the turbine.

  

 

TEG located
in EGR
Circuit

 

   
 
   
 
 

VGT
Turbocharger

Exhaust

 

 

4*
46.

Figure 17: WAVE engine system layout of single TEG/EGR cooler configuration

3.2.3 Variable Geometry Turbocharger

The Cummins ISX engine utilizes a Variable Geometry Turbocharger (VGT). One piece
of data which was not available from Cummins is the turbocharger maps for the VGT.
This problem was overcome by using a combination of map data which was already in
the WAVE code. The compressor map used was from a 10 L truck engine example which
was scaled to mimic the required flow for the ISX engine, shown below in Figure 18. The
map data was scaled by adjusting the diameter multiplier of the compressor or turbine

which then multiplies the mass flow rate values listed in the map data.

50

 

 

 

 

 

 

 

 

 

Nncqolnv-NNQ omoovgo
a 883238889a§83mw¢es
g vnnnnnanNN—Fv—v—oo‘g
1% guanom'nuvgg
a E:
o 'O
i 96
T: g;
E z o
r N
O
8
O
D
g
g a
e 1
g o
.. Q
to,
O
.928
g!
o
0 ,
a:
E
E
8a
28 .
a :
o 5
V 2 3
Ln :
’3 g 2
08 , F 8
l l fI‘ .
to o, m 0. Ln. 6 in. Q°
V V 0° U) N N ‘- v-
omemsseid

 

 

 

Figure 18: Compressor performance map
The VGT turbine maps (for each rack setting) came from another example engine which
was a 4 L high-speed twin turbo diesel. The turbine maps are shown in the appendix. The
“rack position” physically translates to adjusting the vanes or cross sectional area of the

turbine to adjust performance. In the code the “rack position” determines which turbine

51

map is used in the calculations. If the “rack position” falls between maps the map data is
interpolated from the adjacent “rack position” maps. These turbine maps were also scaled
to represent performance similar to the actual turbocharger.

Figure 19 shows the WAVE icons which represent the VGT. Included are the
intake/exhaust ambients, compressor/turbine, actuator to control the VGT and associated

ducts and physical shaft connections.

1. Intake ambient

2. Compressor side of turbocharger

3. Duct to intercooler

4. Shaft connecting compressor to turbine
5. Duct from exhaust manifold

6. Variable geometry turbine

7. Actuator which adjusts VGT

8. Exhaust ambient

 

Figure 19: WAVE icons for the VGT with components labeled

In real world applications a VGT system is most commonly used to reduce turbo-lag and
also prevent damage caused by excessive boost pressure during operation. In this WAVE
model the VGT is used to adjust the boost pressure automatically to match that supplied
in the experimental data. This allows the code to adjust the turbocharger performance
automatically to supply the boost necessary for the overall engine output to match that of
the experimental data for each load point as various aspects of the model are modified.
This turbocharger does not perform exactly like the physical device, but this is acceptable
since no transient effects are being studied. If the model in the future were used to test

transient effects, the turbocharger maps would need to be reevaluated.

52

3.2.4 Intercooler

Exact dimensions of the intercooler were not supplied by Cummins. However, inlet/outlet
pressures and temperatures were supplied. Therefore an intercooler was copied from the
10 L truck example engine built into the WAVE code and the parameters were changed
until the cooling and flow performance were approximately that of the experimental data.
In WAVE the intercooler is treated as several small ducts which have ambient
temperature air flowing over the outside approximating a heat exchanger. Figure 20
shows the WAVE icons for the intercooler with the components labeled. Newer versions
of the WAVE code have a heat exchanger sub-model, which may provide a closer

approximation of the experimental results if it should become necessary in the future.

1. Inlet duct from compressor

2. Complex Y-junctions
representing the plenums of the

p , . intercooler

———————:——_—.‘ 3. Special duct comprised of several

hundred small ducts with

ambient air temperature on the

outer surface

® 4. Outlet duct to intake manifold

     
 

Figure 20: WAVE schematic of the intercooler

3.2.5 EGR Circuit

Exhaust gas recirculation (EGR) is used in most modern IC engines to meet NOx
emission requirements. When exhaust gas is mixed with the fresh air entering an engine
the peak temperature reached during combustion is reduced. This reduces engine
efficiency but the lower temperature stops the formation of NOx. The EGR circuit

consists of the ducting, TEG/EGR cooler and the valve required for the introduction of

53

exhaust gas into the intake manifold. The EGR circuit presented a similar challenge to
that of the intercooler, in that there was little information provided on the dimensions of
the EGR system. As in the case of the intercooler, experimental data was presented for
the inlet/outlet pressures and temperatures of the EGR cooler. Some problems
encountered were keeping the exhaust gas flowing in the correct direction and
maintaining the correct EGR fraction in the intake charge. On the ISX engine the intake
boost pressures can be great enough to reverse the EGR flow if the correct pressure
gradient between the intake and exhaust manifolds is not maintained. With much trial and
error, eventually a system was devised which utilized a single TEG duct acting as the
EGR cooler and a valve located between the TEG/EGR cooler and the intake manifold.

Figure 21 illustrates the EGR circuit and the associated components.

 

 

 

 

 

1. EGR Fraction 1 Boost Sensor at ®
Outlet to Intake Manifold

2. EGR Valve with Actuator
TEG l EGR Cooler

 

fig 4. Inlet from Exhaust Manifold

Figure 21: WAVE diagram of the EGR circuit with labels

54

At this stage the cooled exhaust gas is approximately 100 K higher than the experimental
data. This is primarily because the TEG/EGR cooler was not modeled as a heat
exchanger, as the intercooler had been. The thermoelectric material on the duct is acting
somewhat like an insulator and prevents exhaust heat from escaping. It is believed that in
actual operation the TEG will perform in this fashion. Even though the temperature is
higher than it should be, the overall model engine performance is quite close to that of the

experimental data and was therefore accepted as reasonable.

3.2.6 Control System

Achieving an accurate simulation of this complex engine system presented several
challenges. Most important, controlling the EGR fraction (i.e., the required fraction of
the total intake gases that are exhaust gases circulated through the TEG/EGR cooler) and
the VGT rack position, which determined the turbocharger boost pressure and
correspondingly the specified engine power output, required use of the WAVE software
feedback control system feature. The controller was given target EGR fraction values
derived from test data which is shown in Table 3. By using this approach the software
itself would adjust to changes elsewhere in the model automatically. The controller is a
PID which has automatically calibrated gains. In order to calibrate the controller
automatically, WAVE will move the actuator slightly and determine the response time of
the sensor. In order to calibrate effectively, the auto-calibrate routine must be conducted
several hundred times for each case in the model. The auto-calibration of this model can

take over a day to complete.

55

3.2.7 Sensors and Actuators

Sensors and actuators are two components which must be added to the model in order for
the control system to function. The sensors consisted of an EGR fraction sensor and a
pressure sensor which take their readings from where the ducts from the intercooler and
EGR circuits join the intake manifold as shown as #1 in Figure 21. Since the EGR
fraction of the intake air is commonly controlled by an EGR valve, an actuator was
employed to control the diameter of an orifice between the intake manifold and the
TEG/EGR cooler. The EGR valve actuator is shown in Figure 21 as #2. The other
actuator used was for VGT rack position which controls the angle of the inlet vanes in the
turbine. By adjusting the rack position the turbine wheel will speed up or slow down and
therefore increase or decrease the amount of boost pressure created by the compressor.

This VGT actuator is shown in Figure 3.2.2 as #7.

56

3.2.8 WAVE Model Validation

The WAVE code was used to simulate the baseline ISX production engine at each of the
five operating points to validate model results against dynamometer data provided by
Cummins engine. These model validation results are shown in Table 7 below. The
majorloutput parameters for all but one case match the Cummins data within 10%, which
is considered acceptable for this type of modeling. Table 7 shows the correlation
between the WAVE modeling predictions for brake torque (ft-lb), brake horsepower,

airflow (kg/hr), EGR fraction, and compressor boost pressure (bar).

Table 7: Comparison of Cummins test data to WAVE model results

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A-25 A-100

Cummins WAVE %Diff Cummins WAVE %Diff
TORQBR 472.2 485.2 2.8 1886.8 1879.8 -0.4
BHP 110.6 113.7 2.8 441.8 440.3 -0.3
AIRKGH 539.1 529.6 -1.8 1300.4 1305.8 0.4
EGR 0.265 0.274 3.4 0.171 0.171 0.3
BOOST 1.467 1.369 -6.6 3.069 3.083 0.5

B-62 B-100

Cummins WAVE °/oDiff Cummins WAVE %Ditf
TORQBR 1170.2 1149.3 -1 8 1887.3 1752.6 -7.1
BHP 334.1 328.3 -1.7 538.8 500.6 -7.1
AIRKGH 1271.4 1226.9 -3.5 1729.4 1496.2 -13.5
EGR 0.214 0.218 1.8 0.201 0.201 0.1
BOOST 2.561 2.560 0.0 3.488 3.286 -5.8

C-100

Cummins WAVE %Diff
TORQBR J577.7 14613 -7.4
BHP 540.8 500.8 -7.4
AIRKGH 1862.0 1901.7 2.1
EGR 0.161 0.158 -2.2
BOOST 3.109 3.351 7.8

 

 

 

 

 

57

 

For the B-lOO load point (1500 rpm/full load), the total airflow prediction is —13.5%
different than the measured result. While this value is outside the +/- 10% range it is still
considered to be adequate considering that the brake torque, which is dictated by airflow,
is still in the range. Because the VGT rack was at its fully closed (max boost) position at
this load point, it has been assumed that the airflow would match better if the actual
turbocharger map data was used. These results provided some confidence in the overall
modeling approach accuracy, allowing us to consider results for TEG output and

therefore the BSFC improvement.

58

3.2.9 Results

Once the model had been validated as being a good approximation of the physical engine,
various parameters of the model can be examined. The inner and outer wall temperatures
of the TEG are of particular interest. These values can be given to the material science
engineers in charge of developing the thermoelectric material. Then they can use the
temperature data to select an appropriate material and predict the thermoelectric
efficiency, which is fraction of heat input energy that will be converted to electricity.
Figure 22 below illustrates that the inner (T2) and outer (T3) wall temperatures reported
by WAVE are calculated radially in the middle of each material thickness. The
temperatures T2’ and T3’ are of interest to the TE materials group as these are the
temperatures at the interfaces. The interface temperatures are important because these
temperatures establish the thermal gradient which the TEG relies upon to function. The
temperatures for T2 and T3 under EGR Cooler in Figure 22 are from the B-62 operating
point. The temperatures are recorded axially at the inlet, midpoint, and outlet of the TEG

duct

59

 

 

 

 

 

  
 

Temperature

Heat flux

wall

 

   

Inner
wall

 

Figure 22: Diagram comparing wall temperatures for the single TEG/EGR cooler configuration at
the B-62 operating point with the Phase I six cylinders per TEG configuration

The inner and outer wall temperatures of the TEG/EGR cooler duct for all five operating

points are compiled below in Table 8.

Table 8: Inner and outer wall temperatures for single TEG/EGR cooler concept

 

 

 

 

 

 

 

 

 

 

 

 

 

Inner Wall Temperature(K)
Axial Pos A-25 A-100 B-62 B-100 C-100
Inlet 546 815 710 936 862
Midpoint 429 600 560 732 670
Outlet 375 472 461 571 513
Outer Wall Temperature(K)
Axial Pos A-25 A-100 B-62 B-100 C-100
Inlet 433 564 513 622 587
Midpoint 375 457 437 520 490
Outlet 350 397 391 445 41 7

 

 

 

 

60

 

3.2.10 BSFC Calculation for Single TEG/EGR Cooler

Utilizing the equation for the improvement in BSFC from Phase I, the percent
improvement was calculated for all five operating points using the Phase 11 model, shown
below in Table 9. At this point a TEG efficiency of 9.1% was adopted for the remaining
BSFC calculations. It was assumed that this efficiency could be achieved for the typical
temperature gradients reported by WAVE regardless of TEG location. The first
conclusion that follows from the table is that the predicted improvements in BSFC are
substantially less than those in Phase I, even for the worst case of six cylinders into a
single TEG at the B—62 load point of 2.8%. This can be attributed in the calculation to the
reduction in heat flux through the thermoelectric material and slightly lower crank
horsepower. In Phase I all of the exhaust was flowing through the TEG, and as a result
the heat flux for the B-62 load point was calculated to be 64.5 kW which is

approximately twice the heat flux of Phase II.

Table 9: BSF C percent improvement for single TEG/EGR cooler

 

 

 

 

 

 

 

 

 

 

 

Load Point TEG Heat WAVE BSFC
Transfer Rate (kW) BHP Improvement (%)

A-25 14.64 1 13.65 1.4

A-100 39.83 440.27 1.0

B-62 32.34 328.28 1.1

B-100 53.28 500.59 1.2

C-100 47.63 500.84 1.0

 

However, one cannot make a direct comparison of the TEG output between that of Phase
I and Phase II for two reasons. First, in Phase I the TEG received the full exhaust flow. In
Phase II, for the B—62 load point, the TEG/EGR cooler has 20% of the exhaust flow

diverted into the duct. Second, in Phase II the internal diameter of the TEG duct was

61

increased from 6.5 cm to 8.5 cm. This was done to approximate the cross-sectional flow
area of the physical EGR cooler. This resulted in an increase of internal surface area of
the TEG/EGR cooler by approximately 30% which is used in the calculation of the heat

flux.

3.2.11 Summary of Phase II

In Phase II a more realistic model has been created in WAVE which includes many
features which were omitted in the Phase I efforts. The TEG has been placed in the EGR
circuit and no longer receives the full exhaust flow. A variable geometry turbocharger
was added to the model as well as an intercooler to lower the intake air temperature. A
control system was implemented to adjust the EGR valve automatically and the VGT
rack position during simulation. The model was validated to the Cummins test data and
was accepted as providing a reasonable approximation of the physical ISX engine. Data
was collected from the model including TEG wall temperatures for use by the
thermoelectric development group. An estimated BSFC percent improvement was

calculated for all five operating points and ranged from 1.0 to 1.4% improvement.

62

3.3 Phase III

3.3.1 TEG Placement Concepts

The WAVE engine system modeling has been expanded to explore different TEG
placement opportunities with emphasis on obtaining higher temperature gradients across
the TEG and thereby achieve further improvements in BSFC. Several placement
possibilities were explored including a secondary TEG after the turbocharger, a TEG on
each exhaust port similar to the Phase I study, and the integration of an additional heat
exchanger into the EGR circuit after the TEG. A new strategy in calculating BSFC is

developed using an improved baseline model.

3.3.2 Dual TEG

As a first attempt to achieve a higher percent improvement in BSFC, an additional TEG
was included in the exhaust stream after the turbocharger as shown in Figure 23 below.
This concept is based upon the relationship of inner surface area to TEG output discussed
in Phase I. In order to obtain the maximum surface area possible, the lengths of both
TEGS were also increased from the 150 cm (in Phase I and the initial Phase 2 results
above) to 200 cm. Increasing the length of the TEG/EGR cooler by 50 cm increased the
inner surface area by 33% which will subsequently increase the amount of heat transfer

into the duct wall.

63

The second TEG was simply a copy of the TEG/EGR cooler from Phase H, which was
attached to the outlet of the turbine in place of the 100 cm exhaust duct leading to the
exhaust ambient. The only modification to any of the second TEG’s parameters was
increasing the inner diameter from 8.5 to 10.16 cm to match the duct it is replacing. The
secondary TEG has an inner surface area 20% larger than the single TEG/EGR cooler.
The combination of the two TEGs increases the total inner surface area by 193% over

that of the single TEG/EGR cooler configuration.

 

— EGR Flow

TEG located N w
in EGR ’ ll Exhaust

 

 

 

 

 

 

 

 

   
   

 
   
 
  

 
    

Circuit
i ”I 1 9|
EGR Valve 5 'm " " Additional
3 “a g n *1 TEG added
1’ a . after
Q m ' Turbine
K n /
3 g n-
: VGT ,,, .
Turbocharger m n E
‘3
l

 

 

 

 

 

1:: am

Figure 23: WAVE diagram for the dual TEG configuration

64

The percent improvement in all five points was again calculated and compiled below in
Table 10. Here we see that the total TEG energy input rate is more than double that of the
single TEG/EGR cooler studied in Phase II. Also the percent improvement in BSFC is

nearly the 2.8% which was predicted in Phase I for the B-62 load point.

Table 10: BSFC percent improvement for dual TEG configuration

 

 

 

 

 

 

 

 

 

 

. Total TEG Energy BSFC Improvement
Load Pomt Input Rate (kW) BHP (% )
AQS 37.03 114.28 3.5
A-100 98.01 444.76 2.4
B-62 74.85 329.82 2.5
B-100 127.34 510.51 2.7
C-100 117.83 516.59 2.5

 

 

By extracting exhaust heat downstream from the turbine there were no negative effects
on engine power output. All of the operating points reported an increase in BHP as well
as greatly increasing the total heat flux in the TEGS.

Figure 24 below compares the BSFC percent improvement for the dual TEG and single
TEG/EGR cooler configurations for all five operating points. By adding the secondary
TEG the total surface area was increased more than three times over the single TEG/EGR
cooler. This larger surface area increased the total TEG energy input rate by l30-150%.
Comparing the two configurations the dual TEG configuration shows more than double
the percent improvement in BSFC than the single TEG/EGR cooler for all operating
points. The higher crankshaft horsepower for all five operating points also contributed to

the increase in BSFC percent improvement.

65

 

 

 

I Dual TEG: EGR Cooler & Alter Turbine. Single EGH Cooler TEE
4.0 _ . .

    

3.5

3.0

2.5

% Improvement
M
O

A-25 A-100 B-62 B-100 C-100
Operating Point

 

 

 

Figure 24: BSFC percent improvement comparison for single TEG/EGR cooler and the dual TEG
configuration

66

3.3.3 TEG on Each Exhaust Port

Similar to the most promising configuration of Phase I, a more comprehensive engine
system model with the one TEG per cylinder concept was studied. This was
accomplished by replacing the ducts which connect each exhaust port to the exhaust
plenum with a TEG unit for a total of six TEGs. Previous work with the full engine
model, which includes the turbocharger and EGR cooler subsystems, considered one and
two TEGs per engine in order to minimize the loss of exhaust gas heat energy and,
therefore, a potential loss of engine power. The updated WAVE stick model is shown in

Figure 25. Each TEG duct is basically a copy of the TEGs that were modeled in previous

 

    

 

 

 

 

 

 

work.
name: I. "' '
f6, i an“ . TEG ducts In
a, «m -xhaust manifold
D; 1'
‘ k .41 ll
0 5): 0| ‘ c“l ,,
I. Q:
’* a k .. 3 RI on. 2 YE W
m an ‘m h“
I ' .. ' i
Q 1 ~ MI
W n m
. .‘
_ a: «L w'
i " an
J no:
.‘
a: 8“ 41¢ -—
* mucou- * su § m ‘5
, , W I
(in: MAB!" “w“ fi
13 an a.“ m

Figure 25: WAVE stick diagram showing the 6 TEGs in the exhaust manifold

67

For each speed/load point, if engine power decreased as a result of reductions in turbine
inlet energy, attempts were made to recover the loss through changes in turbocharger
geometry, specifically the multipliers for compressor and turbine diameters. These
multipliers scale the mass flow rates from the turbo map data and therefore effectively
scale the turbochargers’ performance.

Each of the TEGs attached to the exhaust ports is a copy of the TEG/EGR cooler from
the previous iteration of the model. The inner diameter and the length of the TEG duct
were the same as the ducts which were replaced. It should be noted at this time that the
TEG/EGR cooler duct is still in place even though it is labeled as EGR_COOLER in
Figure 25. The heat transferred to the TEG/EGR cooler was not included in the following
calculations. This was done in order to obtain a better understanding of how effectively
the six TEGs perform in a configuration similar to Phase I.

Once the TEG ducts were in place, a parametric study was undertaken to determine the
effect of the TEG duct lengths on the engine performance. The lengths of the ducts were
varied from 100m to 150cm with calculations being performed at 10cm increments. The
engine horsepower output is compared with the TEG duct lengths in Figure 26. At the
initial length of 10cm, the horsepower output is essentially the same as the previous
model, which was operating at the power levels provided by the Cummins test data. As
the TEG duct length is increased, the horsepower output of the engine decreases, except
for the A-25 load point of lowest engine speed and load. For the most important B-62
point, which best represents highway cruise conditions, there is no loss in BHP until TEG

lengths are greater than 30 cm.

68

600

 

— A-25 -x-— A-100 -—+—— B-62 -B- B-100 --- C—100

ES
5’

300 ~

 

Brake power produced by engine [hp]

 

 

 

 

1oo . . . ; i
o 25 50 75 100 125 150

Exhaust Manifold TEG Length [cm]

Figure 26: Plot of engine horsepower vs. TEG length for all 5 operating points

35x105 — A-25 ~x— A-100 -+- B-62 -B— B-100 --- C-‘lOO

 

3.0x105 .

2.5,(105.

Boost [Pa]

2.0x105 ~

15x105~~7~

 

 

 

 

 

 

mm5 + . i 1 i
o 25 50 75 100 125 150

Exhaust Manifold TEG Length [cm]

Figure 27: Plot of total TEG heat transfer vs. TEG length for all 5 operating points

69

Figure 27 shows the corresponding total heat flux produced by the 6 TEGs versus the
TEG duct lengths. As seen in previous Phase I and Phase H studies, the heat flux is a
strong, generally increasing, function of the TEG duct length. The exception to this trend
is the load point A-100 at TEG lengths greater than 90 cm. This drop in TEG heat
transfer can be attributed to the significant drop in engine power output.

While the heat transfer at the longer lengths is increased significantly, the engine
performance depicted in Figure 27 decreases significantly. These results represent the
best possible engine output achievable by the aforementioned attempts to re-optimize

turbocharger turbine parameters.

It was determined that BHP losses for TEG lengths greater than 30 centimeters would be
unacceptable by the manufacturer. Therefore, calculations were conducted at the length
of 30cm to determine if there is potential for improvement in BSFC. Table 11 shows the
results of the BSFC calculation. The highest improvement was 2.7% better than that of
the base engine itself. The BSFC is dependent on the horsepower produced by the engine
and when the fueling rate is held fixed, the BSFC will fluctuate with the engine power

output.

Table 11: BSF C percent improvement for 30cm TEG duct length (an-optimized)

 

 

 

 

 

 

 

 

BSFC

Load Total heat flux Brake

Point in TEGs (kW) Horsepower Imprqy/3ment
A-25 20.27 112.51 2.7
A-100 64.31 398.24 2.4
B-62 39.67 329.19 1.8
B-100 67.99 491.93 2.0
C-1OO 65.00 518.56 1.8

 

 

 

 

70

 

The horsepower output of the engine was down by over 10% for the 1230rpm (A-lOO)
and 1500rpm (B—100) cases. The data in Table 11 above was compiled without changing
the compressor and turbine multipliers which were set to 1.0 and 1.5 respectively. It
appears that the engine model is operating quite efficiently at its current configuration for
this fuel rate. Therefore, it does not appear to be possible to increase the power level
further without increasing the fuel consumption. It is then concluded that while the pre-
turbo exhaust manifold area is a good source of high temperatures and energy to drive a
TEG, this energy cannot be extracted in large quantity without negatively impacting the

overall engine performance.

71

3.3.4 TEG / EGR Cooler and Dual TEG Concepts with Heat Exchangers

As the studies progressed, discussions With thermoelectric experts suggested that the
TEG material acts more like an insulator than previously thought. Therefore, when used
as a replacement for the EGR cooler, the TEG does not lower the EGR temperature as
much as the original cooler would have. In many cases the EGR was more than 100 K
higher than the Cummins test data. Since such a high EGR fraction is required to meet
NOx emissions requirements, this was found to impact the overall performance of the
engine negatively because the hotter EGR was heating the intake air charge and therefore
reducing the power output of the engine. A simple solution to this was to integrate a
small heat exchanger into the EGR circuit after the TEG to reduce the temperature of the
EGR before being introduced to the intake manifold. This is somewhat counter to the
original goal of eliminating the EGR cooler completely and replacing with a TEG.
However, the performance increase of the engine power and therefore improving BSFC

more than outweighs the cost and complexity of the additional components.

72

 
    
  

@TEG/ EGR Cooler
®Heat Exchanger

 

 

i" ‘ menu 5
.7 _y ‘1’,
:m as:

 

Figure 28: WAVE stick diagram of single TEG/EGR cooler (length 150 cm) with additional heat
exchanger (length 50 cm)

The heat exchanger model is a special type of duct that was included in WAVE version
7.2. This model consists of a duct 50 cm long with an inner diameter of 8.5 cm between
two complex Y—junctions which represent the tube bundle of a physical heat exchanger.
The stick diagram showing how the ducting is connected is shown in Figure 28. The
model uses controllers for effectiveness and pressure drop. Effectiveness is based upon
the temperature drop across the heat exchanger. Fortunately, pressure and temperature
drop as well as mass flow rate data for the EGR cooler were provided in the Cummins
test data. This data was used to construct maps with which the controller would look up

the flow rate and predict how much cooling and pressure drop should occur. The heat

73

transfer and friction coefficients of the ducts are automatically varied to match the

temperature and pressure drops between the inlet and outlet complex Y-junctions.

Table 12: BSF C percent improvement: Single TEG/EGR cooler with and without heat exchanger

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Load TEG Energy BHP BSFC
Point Input Rate (kW) Improvement (%)
Phase II Single TEG / EGR Cooler (No Heat Exchanger)
A-25 14.64 113.65 1.4
A-100 39.83 440.27 1 .0
B-62 32.34 328.28 1.1
B-100 53.28 500.59 1.2
C-100 47.63 500.84 1.0
Single TEG I EGR Cooler with Heat Exchanger
A-25 14.64 114.86 1.4
A-100 42.80 431 .44 1 .1
B-62 30.00 332.33 1 .0
B-100 49.87 522.45 1.0
C-100 42.27 519.27 0.9

 

 

 

 

 

Table 3.3.3 makes a comparison between the single TEG/EGR cooler from Phase II and
the modified version with the heat exchanger. In almost all cases the addition of the heat
exchanger reduced the TEG energy input rate while increasing the BHP of the model. As
shown in the table the addition of the heat exchanger did not improve the BSFC. The
only exception was for the A-IOO case. The A-100 operating point exhibited a total net
power loss (crankshaft BHP + TEG power) of 6.45 kW. However, the BSFC percent
improvement calculation reported an increase of 0.1% which is impossible according to

the definition of BSFC. This discrepancy is explored further in the next section.

74

This additional heat exchanger concept was expanded and studied on the dual TEG
concept discussed in Section 3.3.1.1 as well. The TEG lengths were both kept at 200 cm
as in the earlier study and a 50 cm heat exchanger was added after the TEG/EGR cooler.
The WAVE stick model for the dual TEG with heat exchanger concept is shown below in

Figure 29.

r - -.IIIIIIIIIIIII.

G) TEG/EGR cooler
® Secondary TEG

C!) Heat Exchanger

 

 

 

l3 :0 I:
(331
.. 5. .i

     

 

Figure 29: WAVE stick diagram of dual TEG configuration (length 200 cm) with additional heat
exchanger (length 50 cm)

The results of this model with the heat exchanger are compared to the dual TEG model
without the heat exchanger in Table 13 below. It can be seen from Table 12 and Table 13
that the addition of the heat exchanger made no appreciable improvement to the BSFC

percentage. This appears to originate in the lower energy input rate in all but the A-100

75

operating point. This is a bit contradictory because the engine horsepower output has
increased in all but the A-100 operating point. This implies that there is a problem with
the BSFC improvement calculation. Each kW gained in the TEG input energy rate will
contribute about 0.1 kW to improving the BSFC. On the other hand, each horsepower
which is gained in the model output is adding 0.75 kW to the BSFC. Therefore it can be
concluded that crankshaft BHP has a greater effect on BSFC than the power generated
via the TEG system. Any decrease in crankshaft BHP must be less than the additional
power introduced by the TEG system for the BSFC to improve. This leads to the next

section where the BSFC calculation is reevaluated.

Table 13: BSFC percent improvement: Dual TEG with and without heat exchanger

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Load TEG Energy BHP BSFC Improvement
Pomt Input Rate (kW) (%)
Dual TEG without Heat Exchanger
A-25 37.03 1 14.28 3.5
A-100 98.01 444.76 2.4
B-62 74.85 329.82 2.5
B-100 127.34 510.51 2.7
C-100 117.83 516.59 2.5
Dual TEG with Heat Exchan er
A-25 35.26 114.89 3.3
A-100 103.00 434.31 2.6
B-62 67.76 334.72 2.2
B-100 111.68 522.76 2.3
C-100 113.31 521.23 2.4

 

 

 

 

 

76

3.3.5 Advanced BSFC Calculation

As the WAVE studies progressed, the BSFC Improvement calculation was found to be
inadequate. While the earlier calculation (Equation 3.2) proved sufficient for the Phase I
studies, it lacked the ability to compensate for power increases and decreases in the
WAVE model. The revised calculation is shown below as Equation 3.3. The primary
change made between the old and new equations is the ability to specify a baseline power
number to serve as a point of reference. By implementing this we are able to compensate
for situations where the WAVE model engine power output increases or decreases due to

changes in TEG geometry and placement.

Powerma, — PowerBase

 

BSFC IMP.(%) = X100

PowerBase

_ (3.3)
PowerTotal " QTEG X77750 X77350 XnINV + P ower WAVE

With the use of the new calculation it was decided to revisit the dual TEG study of

Section 3.3.1.1 to observe the change on the BSFC Improvement numbers. 4

The single TEG/EGR cooler WAVE model was used as the baseline for this calculation.
The single TEG/EGR cooler model was considered to be the most accurate model of the
actual ISX engine created to date. The results of the calculation are shown in Table 14

below. It should be noted that the WAVE Power and the TEG heat flux are the same as

77

previously reported. The only change has been the inclusion of a baseline for comparison

purposes. Results of the new calculation are compared to the previous results in Table 14.

Table 14: Results of the new BSFC percent improvement calculation for the dual TEG study

 

 

 

 

 

 

 

Total heat flux Single TEG/EGR BSFC
Load in TEGs WAVE Cooler Baseline Improvement
Point (kW) Power (kW) (kW) (%)
A-25 37.03 85.22 84.75 4.1
A-100 98.01 331.66 328.31 3.4
B-62 74.85 245.95 244.80 3.0
B-1 00 127.34 380.69 373.29 4.8
C-1 00 1 17.83 385.22 373.48 5.7

 

 

 

 

 

 

 

As shown in Figure 30, the BSFC percent improvement for the dual TEG case is actually
better than previously reported. This is due to the fact that the WAVE engine power
output actually increased over that of the baseline single TEG/EGR cooler model. Now it
can be seen that there is a measurable improvement in BSFC in all cases, and in the case

of C- 100 it more than doubles the previous value.

78

 

I Single EGR TEG Old Calc I Dual TEGs Old Calc n Dual TEGS New Calc
6.0

 

5.0

4.0

3.0

BSFC Improvement (%)

 

A-25 A-100 B-62 B-100 C-100
Operating Point

 

 

 

Figure 30: Comparison of the results of the new BSFC percent improvement calculation to the
previous calculations

These results illustrated the need for a baseline model which can serve as a benchmark
for model power output that does not include a TEG anywhere in the system. The next

section covers this new baseline model that was developed.

79

3.3.6 New Baseline Model

Up until now all the BSFC calculations which used a baseline horsepower level for
comparison had used the single TEG/EGR cooler model. This model was the most
accurate to the Cummins test data that had been developed up to this point. Therefore this
model was used as the starting point for our new baseline model. The TEG duct was
removed from the EGR circuit and replaced with a heat exchanger duct, similar to what
was used in Section 3.3.4. The function of the heat exchanger model is the same as the
previous attempts, but in this case the duct used an inner diameter of 8.5 cm and a length

of 150 cm. Figure 31 below shows the WAVE stick diagram of the baseline model.

I 1
6(3le

é’ MIIIIIIIIIIIGIIIIIIIII.“‘ ’90
‘\ ‘7

® Heat Exchanger , / .

 

 

l "drew.

>A
‘ ‘Wnr
rd»

1:: all

Figure 31: WAVE diagram of baseline model with heat exchanger replacing TEG

80

Great lengths were taken to ensure that this model mimics the performance of the
Cummins ISX engine. Nearly every parameter that was supplied in the Cummins test data
was compared to the baseline model. The percent difference of this comparison has been
compiled into Table 15 below. The percent difference listed in Table 15 shows how close
the baseline WAVE model is to the Cummins test data. If the percent difference is greater
than 10% then the cell is colored red or blue depending on if the value is greater or less
than 110%. The bulk of the values out of the 110% range are related to the EGR valve
and cooler. It will be very difficult to get all of the temperatures and pressures to match
exactly for this type of system, especially since little is known about the exact dimensions

of the EGR circuit.

81

Table 15: Comparison of percent difference between Cummins test data and baseline WAVE model
Load Point

-2.42 -2.43
-3.42 3.99
0.01 0.02
0.09 -8.76

-4.75 5.09
-3.40
-3.33
-0.01

0.97

-6.18

1.07

In

Gas

Cooler

 

82

Considering how well the data matches, this baseline model is accepted as a good
representation of the production Cummins ISX engine. The baseline model power output

which will be used in the BSFC calculations is listed below in Table 16.

Table 16: Baseline engine model power output listed for each load point compared to Cummins test
data

 

 

 

 

 

 

 

 

 

 

Load Cummins Test Baseline Model Difference

Point Data (hp) (hp) (%)
A-25 “057 114.99 4.00
A-1 00 441 -78 426.78 -3.40
3432 33406 326.28 -2.33
B-100 53877 522.46 -3.03
c-1oo 540.79 519.50 -3.94

 

 

The baseline model power levels. are within 5% for all five operating points. For the
B-100 and C-100 operating points, the horsepower levels of the baseline model are
approximately 20 horsepower higher than the single TEG/EGR cooler model. It was
decided to implement this baseline model for the BSFC calculations rather than use the
Cummins test data because the WAVE model will not be able to exactly duplicate the
performance of the physical engine. By using another WAVE model as the baseline, the

comparisons drawn will be more consistent.

3.3.7 Updated BSFC Calculations for All Concepts

Now that there exists a reasonable baseline model without a TEG, the BSFC percent
improvement can be recalculated using the new equation. These percent improvements
are listed below in Table 17 along with the percent difference between the baseline model

and each concept’s horsepower.

83

Table 17: Updated BSF C percent improvement for all concepts

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

L TEG Energy BHP Difference BSFC BSFC
oad WAVE .
Point Input Rate BHP from Baseline Improvement Improvement
(kW) (%) Old Calc. (%) New Calc. (%)
Single TEG I EGR Cooler without Heat Exchanger
A-25 14.64 113.65 -1.17 1.4 0.2
A-1 00 39.83 440.27 3.16 1.0 4.2
B-62 32.34 328.28 0.61 1 .1 1 .7
B-100 53.28 500.59 —4.1 8 1.2 -3.1
C-100 47.63 500.84 -3.59 1.0 -2.6
Ie TEG I EGR Cooler with Heat Exchanger
A-25 14.64 114.86 -0.11 1.4 1.3
A-1OO 42.80 431.44 1.09 1.1 2.2 .
B-62 30.00 332.33 1.85 1.0 2.9
B-1OO 49.87 522.45 0.00 1.0 1.0
C-100 42.27 519.27 -0.04 0.9 0.8
Dual TEG without Heat Exchanger
A-25 37.03 114.28 -0.62 3.5 2.9
A-100 98.01 444.76 4.21 2.4 6.7
B-62 74.85 329.82 1 .09 2.5 3.6
B-100 127.34 510.51 -2.29 2.7 0.4
C-1 00 117.83 516.59 -0.56 2.5 1.9
Dual TEG with Heat Exchanger
A-25 35.26 1 14.89 -0.10 3.3 3.2
A-100 103.00 434.31 1.77 2.6 4.4
B-62 67.76 334.72 2.59 2.2 4.9
B-100 111.68 522.76 0.06 2.3 2.4
C-100 113.31 521.23 0.33 2.4 2.7
6 TEGs on Exhaust Ports (30 cm Leggth)
A-25 20.27 112.51 -2.16 2.7 -0.2
A-100 64.31 398.24 -6.69 2.4 -5.0
B-62 39.67 329.19 0.89 1.8 2.2
B-100 67.99 491 .93 -5.84 2.0 -4.4
C-100 65.00 518.56 -0.18 1.8 1.2

 

 

 

 

 

 

84

 

Table 17 lists the TEG energy input rate, WAVE BHP for each configuration, the percent
difference between each configuration’s BHP and the baseline model, and the BSFC
improvement calculated with both the old and new equations. Immediately obvious is the
possibility to have a negative BSFC improvement percentage. This happens when the
power from the thermoelectric conversion to mechanical work is insufficient to offset any
engine power loss due to the addition of the TEG. Above all this is seen in the case of the
6 TEGs concept where only two of the five operating points report an improvement in

BSFC.

1- srngr’fig. (0%:an Sinsfiséfiéw Eqnlgsmsveje w/ Heat EXFh- (WEE
5.0

 

4.2

1“
o

9"
o

.N
o

.5
O

 

P
O

 

B-62

BSFC Improvement (%)
2';
O
M

I
I"
O

I
9°
0

 

 

 

 

-4.0 1122,, ,,_,,_,,,,2-,,,, --- ~——-

Figure 32: Comparison of BSFC percent improvement: Single TEG/EGR cooler concept with and
without heat exchanger

Figure 32 above graphically illustrates the BSFC improvement for the single TEG/EGR

cooler concept, including the new and old BSFC equation and the inclusion of the heat

85

exchanger after the TEG in the EGR circuit. It is quite obvious from this graph that the
lower horsepower output of the single TEG/EGR cooler model at the B-lOO and 0100
operating points noticeably impacts the BSFC improvement percentage. It can also be
seen in Figure 32 that when the heat exchanger is added after the TEG in the EGR circuit,
the engine horsepower is increased and even though there is less TEG input energy the

BSFC improves.

 

ifingle TEG II finglefiTEGfiw/ilil D Dual TEG D DuaLT'EGLIHX LGTEGI

0.0 4

 

BSFC Improvement (%)

-2.0 '

4.0

 

 

-s.o
Figure 33: BSFC percent improvement for all five concepts using new BSFC calculation
All of the concepts discussed previously are compared in Figure 33 using the new BSFC
improvement equation. This graph demonstrates dramatically how the 6 TEGs on exhaust
ports concept does very little to improve BSFC. This is further reinforced when looking
at the average BSFC improvement listed below in Table 18. The 6 TEGS on exhaust ports
concept shows a -1.3% improvement in BSFC which indicates that overall this concept
would not be a good fit for this project. Also, from this table, the dual TEG with the heat

exchanger concept decidedly is the best overall concept for the average increase in BSFC.

86

In addition this concept also has the highest improvement in BSFC at the B-62 load point,
which is the most critical load point for real world engine operation, with nearly 5%

improvement in BSFC.

Table 18: Summary of BSF C percent improvement

 

 

 

 

 

 

 

 

Average Operating Max BSFC
Concept Point BSFC Improvement
Improvement (%) (%)

Single TEG I EGR Cooler 0.1 4.2 @ A-100
Single TEG / EGR Cooler 1 6 2.9 @ 8'62
with Heat Exchanger '
Dual TEG 3.1 6.7 O A-100
Dual TEG wrth Heat 35 4.9 Q B -62
Exchanger
6 TEGs on Exhaust Ports -1.3 2.2 @ B-62

 

 

 

 

It is quite significant that the dual TEG with heat exchanger concept reportedly increases
the BSFC by 4.9% at the B-62 “cruise” operating point. If this can be realized on an ISX
engine in a class 8 truck over the emissions useful life of the engine (435,000 miles) it
will potentially reduce the fuel usage by 4200 gallons. It can be assumed that fuel will
continue to be at least $4.00/gallon and the reduced fuel consumption will save the

operator of the truck $17,000 or potentially more as fuel costs increase.

87

3.3.8 Summary of Phase III

Phase III explored multiple TEG placement possibilities and went on to compare the
percent BSFC improvement of these various concepts. In the process of comparing these
concepts, it was realized that the previous equation for calculating BSFC improvement
proved insufficient for making comparisons between concepts with differing engine
power levels. A new equation was developed which incorporates a baseline power level
that compensates for increases and decreases in crankshaft horsepower in the model. A
new baseline model was developed which used a heat exchanger model in place of the
TEG in the single TEG/EGR cooler concept which was developed in Phase H. This new
baseline model was a very good match for the Cummins test data. The BSFC percent
improvement was recalculated for all concepts investigated. It was realized that the dual
TEG with heat exchanger concept provided the greatest improvement in BSFC, and
moreover the highest BSFC improvement was seen at the B-62 load point where the

engine will operate most of its useful life.

88

4 Overall Summary

It has been demonstrated that the use of WAVE can be a useful tool in predicting the
performance of thermoelectric generators and assisting with their design.

Phase 1 demonstrates the potential benefits of thermoelectric generators to capture and
convert waste heat energy to electricity. Phase I focused on the B-62 operating point,
which simulates the typical highway cruising condition. Several assumptions were made
to obtain a “best case scenario” prediction of the improvement to BSFC. Extensive
studies determined the optimal combination of inner diameter and length of the TEG for
performance and packaging with the Cummins ISX engine. The predicted improvement
for the one, three, and six cylinders per TEG configurations were 6.2, 4.0, and 2.8%
respectively. The one cylinder per TEG shows the most promise, not only in the percent
improvement in BSFC but also as a model to be used in conjunction with a single
cylinder test engine.

Phase II expanded upon the modeling conducted in Phase I by recreating the Cummins
ISX engine in greater detail. The model now includes a full intake and exhaust system
utilizing an intercooler and a VGT. Also included was an EGR circuit which had the heat
exchanger section replaced with a TEG. Associated control systems for the EGR fraction
and the VGT rack position were also implemented. This phase examined performance at
all five operating points, and the performance of the model was shown to match up with
Cummins test data to within 10% in all but one area. TEG wall temperatures were
collected and TEG efficiencies were predicted. BSFC improvement was estimated in the

1.0 to 1.4% range for the five operating points.

89

The detailed model of the ISX engine from Phase II was modified for further studies
examining possible locations for TEG placement in Phase H1. The placement concepts
examined included: a dual TEG concept with a secondary TEG after the turbocharger in
the exhaust, a revisit of the Phase I one cylinder per TEG concept, and the modification
of the Phase II and the dual TEG concepts which included a heat exchanger in the EGR
circuit to improve engine performance. The BSFC percent improvement calculation was
reevaluated and modified to incorporate a baseline horsepower number to achieve a more
consistent comparison between different configurations. As part of the improved BSFC
calculation, a baseline engine model was developed without a TEG anywhere in the
exhaust system. The single TEG/EGR cooler and dual TEG concepts each with and
without heat exchangers and the six TEGs on exhaust ports concept then had their
respective BSFC percent improvement recalculated with the updated calculation. It was
shown that on the average of all operating points the six TEGs configuration would
decrease the BSFC by 1.3%, with a maximum improvement of only 2.2%. The single and
dual TEG concepts both experienced an improvement in BSFC once a heat exchanger
was incorporated into the EGR circuit. The dual TEG with heat exchanger was the most
promising, with the average BSFC improvement of 3.5% for the five operating points.
This configuration also exhibited the highest improvement at the B-62 load point of
4.9%. The six TEG concept may prove useful when used with a single cylinder test
engine yet. The concept which will be most likely implemented on a production ISX

engine will be the dual TEG with a heat exchanger.

90

5 Conclusion

As transportation fuel costs continue to increase, efficient powertrain systems are more
important than ever before. Efficiency includes reducing losses of energy wherever
possible. Approximately 60% of fuel energy is lost as heat in a modern internal
combustion engine. The use of thermoelectric generators in the exhaust system will help
reduce these losses and consequently increase efficiency. The dual TEG with a heat
exchanger concept will provide 4.9% better BSFC at the B-62 load point and potentially
save over 4200 gallons of fuel over the emissions useful life of the engine. With fuel at
$4.00/gallon that will amount to a reduction of operating costs of over $17,000 for the
emissions useful life of 435,000 miles. This technology shows a good deal of potential

for improving the fuel consumption of an over-the-road truck engine.

6 Future Work

This project has several options for future work. The WAVE model can be coupled into
an advanced 3D heat transfer model developed by Tom Shih at Iowa State University.
The software used by Tim Hogan’s group to predict TE efficiency could be coupled into
the WAVE code to provide material efficiencies along the entire length of the TEG. The
EGR fraction and fuel rate can be increased to match what Cummins estimates will be
necessary to meet the new 2010 emission regulations. Once a TEG prototype is
developed the, test data can be used to calibrate a TEG performance model which can

then be coupled to WAVE.

91

 

APPENDIX: VARIABLE GEOMETRY TURBINE MAPS

 

Int—‘31

99:30.0
abersNo
Q9510

”craved

 

flowxaod u2g
vexurod 02:2

-—

N
n
a
n
w
h
m
m
9

.13.: 32”. Bus.

 

Orv

Wm

0......

can”. 2339:.
m. N

0N

m._. 0.?

 

 

   

L 0.0

.- B tad

........ - to $8.0

....................... T to rxcd

 

voted

 

 

 

 

 

coca;

 

 

8 an

 

co :4...
ans. occur—totem 6583

[Warmth] needs

 

 

Figure 34: Turbine Performance Map VGT Rack Position 1.0

92

 

23m 2320...".
ogv md 06 md ON mhw 0.6

 

     

 

co :3

90:53.0 H2.2
worgm 025

....... - co :3...

..................... r to Fxod

tinny

 

.o :8...

 

42.8.?

and ‘

an: SCGEOtUl 0581:

 

 

 

 

 

 

.9 :ae

 

93

[sawmill] Deeds
Figure 35: Turbine Performance Map VGT Rack Position 0.75

 

 

Kirov

 

«0595;.

 

0.026000 H2.2
«9:35— Hx5).

I I I I I I I I I
Dmmktflnq-I'INI-

‘—

~.£9= ROE an:

80.0 H2_E
500.0 “x52

11:

o..v

can! 2:32.".

 

 

m6

0.0

0N

0N

mé 0.

   

.To :00

- so :35

- 3:60

 

59:66

 

 

 

 

 

no N

3:6;

 

 

no.2 «Generated 059.;

cc :8;

[90me] needs

 

 

Figure 36: Turbine Performance Map VGT Rack Position 0.5

94

 

”-1.560

90:03.0 H2.5.
«Sauna; H52..

2

.".

O

l I I r
9mmntfln¢nNP

«0:9 F.
wowamNP

fins. .6; 8.:

 

can". 0536i

 

 

ir 0.0

£9.00

..... - 6:35

............... 40:80

 

40:60

 

50:6;

 

 

 

2.0 \.

 

.o :&e
ems. coca—Eaton. 655.5%

[g owwdl] paads

 

 

Figure 37: Turbine Performance Map VGT Rack Position 0.25

95

 

uncanny .

02v

md

0.0

caum 9539:.
Wu

0N

mé 0.?

 

 

90rx~n00 “2.2
105000... 09.2
90:58.0
wowxorud
90:80.0
worxhfid

a
E

_.

N
n
.v
n
0
s
w
m
0

«3.0 022

‘_

 

     

 

 

 

no \

 

 

 

 

nus. nucaEcoton 659.;

vowxmwd

50:63

50:63

so :63

[sex/mm] paeds

 

 

Figure 38: Turbine Performance Map VGT Rack Position 0.0

96

REFERENCES

[1] AS. Kushch, Proc. of the 20th Intl. Conf. on Thermoelectrics, 2001, Beijing, China,
pages 422-430.

[2] K. Ikoma, Proc. of the 13th Intl. Conf. on Thermoelectrics, 1998, Nagoya, Japan 464-
467.

[3] K. Matsubara, Proc. of the 21St Intl. Conf. on Thermoelectrics, 2002, Long Beach,
CA, USA.

[4] J. Vazquez, Proc. 7th European Workshop on Thermoelectrics, 2002, Pamplona,
Spain.

[5] G. L. Bennett, CRC Handbook of Thermoelectrics, Ed. D. M. Rowe, CRC Press LLC,
Boca Raton, FL, 1995.

[6] Michigan State University et al., Thermoelectric Conversion of Waste Heat to
Electricity in an IC Engine Powered Vehicle: Phase 1 Final Report, Department of
Energy, October 2005.

[7] Ricardo WAVETM Basic User Manual. Release 7.2. October 2006

[8] Ricardo WAVETM Engine User Manual. Release 7.2. October 2006

[9] Ricardo WAVETM Conduction User Manual. Release 7.2. October 2006

97

   

"11111111111111illlllljillllilllf